diff --git a/ci-scripts/oai-ci-vm-tool b/ci-scripts/oai-ci-vm-tool index 8d3dbb4ddd..dda71d2606 100755 --- a/ci-scripts/oai-ci-vm-tool +++ b/ci-scripts/oai-ci-vm-tool @@ -252,7 +252,7 @@ case $key in VM_NAME=ci-phy-sim ARCHIVES_LOC=phy_sim LOG_PATTERN=.Rel15.txt - NB_PATTERN_FILES=6 + NB_PATTERN_FILES=7 BUILD_OPTIONS="--phy_simulators" RUN_OPTIONS="./run_exec_autotests.bash -g \"01510*\" -q -np -b" NBARGS=$[$NBARGS+256] @@ -330,7 +330,7 @@ case $key in VM_NAME=ci-phy-sim ARCHIVES_LOC=phy_sim LOG_PATTERN=.Rel15.txt - NB_PATTERN_FILES=6 + NB_PATTERN_FILES=7 BUILD_OPTIONS="--phy_simulators" RUN_OPTIONS="./run_exec_autotests.bash -g \"01510*\" -q -np -b" NBARGS=$[$NBARGS+256] diff --git a/ci-scripts/reportBuildLocally.sh b/ci-scripts/reportBuildLocally.sh index 2e78e6f6cd..d523c25baa 100755 --- a/ci-scripts/reportBuildLocally.sh +++ b/ci-scripts/reportBuildLocally.sh @@ -411,6 +411,10 @@ function report_build { then summary_table_row "PBCH Test - Release 15" ./archives/phy_sim/nr_pbchsim.Rel15.txt "Built target nr_pbchsim" ./phy_sim_row6.html fi + if [ -f ./archives/phy_sim/nr_dlschsim.Rel15.txt ] + then + summary_table_row "DLSCH Test - Release 15" ./archives/phy_sim/nr_dlschsim.Rel15.txt "Built target nr_dlschsim" ./phy_sim_row7.html + fi summary_table_footer diff --git a/cmake_targets/CMakeLists.txt b/cmake_targets/CMakeLists.txt index 2e172b94f2..d631abb32d 100644 --- a/cmake_targets/CMakeLists.txt +++ b/cmake_targets/CMakeLists.txt @@ -290,6 +290,8 @@ set(protobuf_generated_dir ${OPENAIR_BIN_DIR}) add_list2_option(RRC_ASN1_VERSION "Rel15" "ASN.1 version of RRC interface" "Rel8" "Rel10" "Rel14" "CBA") +Message("RRC_ASN1_VERSION is ${RRC_ASN1_VERSION}") + if (${RRC_ASN1_VERSION} STREQUAL "Rel8") make_version(RRC_VERSION 8 6 0) set (RRC_GRAMMAR ${OPENAIR2_DIR}/RRC/LTE/MESSAGES/asn1c/ASN1_files/EUTRA-RRC-Definitions-86.asn) @@ -366,9 +368,9 @@ execute_process(COMMAND mkdir -p ${NR_RRC_FULL_DIR} OUTPUT_QUIET ERROR_QUIET) -if (NOT ${ret} STREQUAL 0) - message(FATAL_ERROR "${ret}: error") -endif (NOT ${ret} STREQUAL 0) +#if (NOT ${ret} STREQUAL 0) +# message(FATAL_ERROR "${ret}: error") +#endif (NOT ${ret} STREQUAL 0) file(GLOB nr_rrc_source ${NR_RRC_FULL_DIR}/*.c) file(GLOB nr_rrc_h ${NR_RRC_FULL_DIR}/*.h) @@ -1184,6 +1186,7 @@ set(PHY_SRC_COMMON ${OPENAIR1_DIR}/PHY/CODING/lte_segmentation.c ${OPENAIR1_DIR}/PHY/CODING/nr_segmentation.c ${OPENAIR1_DIR}/PHY/CODING/nr_compute_tbs.c + ${OPENAIR1_DIR}/PHY/CODING/nr_rate_matching.c ${OPENAIR1_DIR}/PHY/CODING/ccoding_byte.c ${OPENAIR1_DIR}/PHY/CODING/ccoding_byte_lte.c ${OPENAIR1_DIR}/PHY/CODING/3gpplte_sse.c @@ -1304,6 +1307,11 @@ set(PHY_SRC_UE ${OPENAIR1_DIR}/PHY/NR_TRANSPORT/nr_pbch.c ${OPENAIR1_DIR}/PHY/NR_TRANSPORT/nr_dci.c ${OPENAIR1_DIR}/PHY/NR_TRANSPORT/nr_dci_tools.c + ${OPENAIR1_DIR}/PHY/NR_TRANSPORT/nr_dlsch.c + ${OPENAIR1_DIR}/PHY/NR_TRANSPORT/nr_dlsch_tools.c + ${OPENAIR1_DIR}/PHY/NR_TRANSPORT/nr_dlsch_coding.c + ${OPENAIR1_DIR}/PHY/NR_TRANSPORT/nr_tbs_tools.c + ${OPENAIR1_DIR}/PHY/NR_TRANSPORT/nr_sch_dmrs.c ${OPENAIR1_DIR}/PHY/NR_REFSIG/nr_gold.c ${OPENAIR1_DIR}/PHY/TOOLS/file_output.c ${OPENAIR1_DIR}/PHY/TOOLS/cadd_vv.c @@ -1333,6 +1341,10 @@ set(PHY_SRC_UE ${OPENAIR1_DIR}/PHY/NR_UE_TRANSPORT/dmrs_nr.c ${OPENAIR1_DIR}/PHY/NR_UE_TRANSPORT/nr_initial_sync.c ${OPENAIR1_DIR}/PHY/NR_UE_TRANSPORT/nr_pbch.c + ${OPENAIR1_DIR}/PHY/NR_UE_TRANSPORT/nr_dlsch_demodulation.c + ${OPENAIR1_DIR}/PHY/NR_UE_TRANSPORT/nr_dlsch_decoding.c + ${OPENAIR1_DIR}/PHY/NR_UE_TRANSPORT/nr_dlsch_llr_computation.c + ${OPENAIR1_DIR}/PHY/NR_TRANSPORT/nr_tbs_tools.c ${OPENAIR1_DIR}/PHY/NR_UE_TRANSPORT/ ${OPENAIR1_DIR}/PHY/NR_UE_TRANSPORT/nr_prach.c ${OPENAIR1_DIR}/PHY/NR_UE_TRANSPORT/srs_modulation_nr.c @@ -1343,6 +1355,7 @@ set(PHY_SRC_UE ${OPENAIR1_DIR}/PHY/NR_REFSIG/nr_dmrs_rx.c ${OPENAIR1_DIR}/PHY/NR_REFSIG/nr_gold_ue.c ${OPENAIR1_DIR}/PHY/NR_UE_ESTIMATION/nr_dl_channel_estimation.c + ${OPENAIR1_DIR}/PHY/NR_UE_ESTIMATION/nr_adjust_synch_ue.c ${OPENAIR1_DIR}/PHY/LTE_ESTIMATION/lte_ue_measurements.c ${OPENAIR1_DIR}/PHY/TOOLS/file_output.c ${OPENAIR1_DIR}/PHY/TOOLS/cadd_vv.c @@ -1523,14 +1536,6 @@ set(NR_L2_SRC_UE ${NR_UE_RRC_DIR}/L2_interface_ue.c ${NR_UE_RRC_DIR}/main_ue.c ${NR_UE_RRC_DIR}/rrc_UE.c -# ${NR_UE_RRC_DIR}/mac_vars.c -#${RRC_DIR}/rrc_UE.c - ) - -set(L2_NR_SRC_UE - ${NR_UE_RRC_DIR}/L2_interface_ue.c - ${NR_UE_RRC_DIR}/main_ue.c - ${NR_UE_RRC_DIR}/rrc_UE.c ) set (MAC_SRC @@ -1636,26 +1641,27 @@ set ( NR_LTE_UE_REUSE_SRC ${OPENAIR1_DIR}/SCHED/phy_procedures_lte_common.c #${OPENAIR1_DIR}/SCHED_UE/phy_procedures_lte_ue.c ${OPENAIR1_DIR}/PHY/LTE_TRANSPORT/lte_mcs.c - ${OPENAIR1_DIR}/PHY/LTE_TRANSPORT/pmch_common.c + #${OPENAIR1_DIR}/PHY/LTE_TRANSPORT/pmch_common.c - ${OPENAIR1_DIR}/PHY/LTE_REFSIG/lte_dl_cell_spec.c - ${OPENAIR1_DIR}/PHY/CODING/lte_segmentation.c - ${OPENAIR1_DIR}/PHY/LTE_TRANSPORT/prach_common.c + #${OPENAIR1_DIR}/PHY/LTE_REFSIG/lte_dl_cell_spec.c + #${OPENAIR1_DIR}/PHY/CODING/lte_segmentation.c + #${OPENAIR1_DIR}/PHY/LTE_TRANSPORT/prach_common.c # ${OPENAIR1_DIR}/PHY/INIT/lte_init_ue.c ${OPENAIR1_DIR}/PHY/CODING/crc_byte.c ${OPENAIR1_DIR}/PHY/CODING/viterbi_lte.c ${OPENAIR1_DIR}/PHY/CODING/viterbi.c - ${OPENAIR1_DIR}/PHY/LTE_TRANSPORT/phich_common.c + #${OPENAIR1_DIR}/PHY/LTE_TRANSPORT/phich_common.c ${OPENAIR1_DIR}/PHY/LTE_UE_TRANSPORT/dlsch_llr_computation.c + # ${OPENAIR1_DIR}/PHY/LTE_TRANSPORT/dci_tools_common.c - ${OPENAIR1_DIR}/PHY/CODING/lte_rate_matching.c +# ${OPENAIR1_DIR}/PHY/CODING/lte_rate_matching.c ${OPENAIR1_DIR}/PHY/CODING/ccoding_byte_lte.c ${OPENAIR1_DIR}/PHY/CODING/ccoding_byte.c ${OPENAIR1_DIR}/PHY/LTE_REFSIG/lte_gold.c ${OPENAIR1_DIR}/PHY/LTE_ESTIMATION/lte_est_freq_offset.c ${OPENAIR1_DIR}/PHY/LTE_ESTIMATION/lte_dl_channel_estimation.c ${OPENAIR1_DIR}/PHY/LTE_ESTIMATION/lte_ue_measurements.c - ${OPENAIR1_DIR}/PHY/LTE_UE_TRANSPORT/dlsch_demodulation.c + #${OPENAIR1_DIR}/PHY/LTE_UE_TRANSPORT/dlsch_demodulation.c #${OPENAIR1_DIR}/PHY/LTE_UE_TRANSPORT/dlsch_llr_computation_avx2.c ) @@ -2555,6 +2561,9 @@ target_link_libraries(polartest SIMU PHY PHY_NR PHY_COMMON m ${ATLAS_LIBRARIES}) add_executable(ldpctest ${OPENAIR1_DIR}/PHY/CODING/TESTBENCH/ldpctest.c) target_link_libraries(ldpctest SIMU PHY PHY_NR m ${ATLAS_LIBRARIES}) +add_executable(nr_dlschsim ${OPENAIR1_DIR}/SIMULATION/NR_PHY/dlschsim.c ${T_SOURCE}) +target_link_libraries(nr_dlschsim -Wl,--start-group UTIL SIMU PHY PHY_COMMON PHY_NR PHY_NR_UE SCHED_NR_LIB CONFIG_LIB -Wl,--end-group m pthread ${ATLAS_LIBRARIES} ${T_LIB} dl) + add_executable(nr_pbchsim ${OPENAIR1_DIR}/SIMULATION/NR_PHY/pbchsim.c ${T_SOURCE}) target_link_libraries(nr_pbchsim -Wl,--start-group UTIL SIMU PHY_COMMON PHY_NR PHY_NR_UE SCHED_NR_LIB CONFIG_LIB -Wl,--end-group m pthread ${ATLAS_LIBRARIES} ${T_LIB} dl) diff --git a/cmake_targets/build_oai b/cmake_targets/build_oai index 82cdba228f..f4b64026e1 100755 --- a/cmake_targets/build_oai +++ b/cmake_targets/build_oai @@ -692,7 +692,7 @@ function main() { echo_info "Compiling unitary tests simulators" # TODO: fix: dlsim_tm4 pucchsim prachsim pdcchsim pbchsim mbmssim #simlist="dlsim_tm4 dlsim ulsim pucchsim prachsim pdcchsim pbchsim mbmssim" - simlist="dlsim ulsim polartest ldpctest nr_pbchsim" + simlist="dlsim ulsim polartest ldpctest nr_pbchsim nr_dlschsim" for f in $simlist ; do compilations \ phy_simulators $f \ diff --git a/nfapi/open-nFAPI/nfapi/public_inc/nfapi_nr_interface.h b/nfapi/open-nFAPI/nfapi/public_inc/nfapi_nr_interface.h index 5ba07281c2..4273186dcf 100644 --- a/nfapi/open-nFAPI/nfapi/public_inc/nfapi_nr_interface.h +++ b/nfapi/open-nFAPI/nfapi/public_inc/nfapi_nr_interface.h @@ -191,6 +191,27 @@ typedef struct { #define NFAPI_NR_SCH_CONFIG_SS_PBCH_BLOCK_POWER 0x5023 #define NFAPI_NR_SCH_CONFIG_N_SSB_CRB 0x5024 +typedef struct { + nfapi_uint16_tlv_t data_scrambling_id; + nfapi_uint16_tlv_t dmrs_typeA_position; + nfapi_uint16_tlv_t dmrs_additional_position; + nfapi_uint16_tlv_t dmrs_type; + nfapi_uint16_tlv_t dmrs_max_length; + nfapi_uint16_tlv_t mapping_type; + nfapi_uint16_tlv_t resource_allocation; + nfapi_uint16_tlv_t time_allocation_list; + nfapi_uint16_tlv_t mcs_table; + nfapi_uint16_tlv_t aggregation_factor; + nfapi_uint16_tlv_t prb_bundling_type; + nfapi_uint16_tlv_t rbg_size; + nfapi_uint16_tlv_t vrb_to_prb_interleaver; + nfapi_uint16_tlv_t code_block_groug_transmission; + nfapi_uint16_tlv_t x_overhead; + nfapi_uint16_tlv_t max_dci_codewords; + +} nfapi_nr_pdsch_config_t; +#define NFAPI_NR_PDSCH_CONFIG_TAG + typedef struct { nfapi_uint16_tlv_t prach_RootSequenceIndex; ///// L1 parameter 'PRACHRootSequenceIndex' @@ -205,12 +226,6 @@ typedef struct { nfapi_uint16_tlv_t preambleReceivedTargetPower; } nfapi_nr_rach_config_t; -typedef struct { - nfapi_uint16_tlv_t dmrs_TypeA_Position; ///// Position of (first) DL DM-RS - nfapi_uint16_tlv_t TimeDomainResourceAllocation_k0; ///// L1 parameter 'K0' - nfapi_uint16_tlv_t TimeDomainResourceAllocation_mappingType; ///// L1 parameter 'Mapping-type' -} nfapi_nr_pdsch_config_t; - typedef struct { nfapi_uint16_tlv_t groupHoppingEnabledTransformPrecoding; ///// L1 parameter 'Group-hopping-enabled-Transform-precoding' nfapi_uint16_tlv_t msg3_DeltaPreamble; ///// L1 parameter 'Delta-preamble-msg3' @@ -269,8 +284,8 @@ typedef struct { nfapi_nr_subframe_config_t subframe_config; nfapi_nr_rf_config_t rf_config; nfapi_nr_sch_config_t sch_config; - nfapi_nr_rach_config_t rach_config; nfapi_nr_pdsch_config_t pdsch_config; + nfapi_nr_rach_config_t rach_config; nfapi_nr_pusch_config_t pusch_config; nfapi_nr_pucch_config_t pucch_config; nfapi_nr_pdcch_config_t pdcch_config; @@ -321,6 +336,13 @@ typedef enum { NFAPI_NR_SEARCH_SPACE_TYPE_UE_SPECIFIC } nfapi_nr_search_space_type_e; +typedef enum { + NFAPI_NR_COMMON_SEARCH_SPACE_TYPE_0=0, + NFAPI_NR_COMMON_SEARCH_SPACE_TYPE_0A, + NFAPI_NR_COMMON_SEARCH_SPACE_TYPE_1, + NFAPI_NR_COMMON_SEARCH_SPACE_TYPE_2 +} nfapi_nr_common_search_space_type_e; + typedef enum { NFAPI_NR_SSB_AND_CSET_MUX_PATTERN_TYPE1=0, NFAPI_NR_SSB_AND_CSET_MUX_PATTERN_TYPE2, @@ -342,6 +364,44 @@ typedef enum { NFAPI_NR_CSET_ALL_CONTIGUOUS_RBS } nfapi_nr_coreset_precoder_granularity_type_e; +typedef enum { + NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_A=0, + NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_B, + NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_C, + NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_ALLOC_LIST +} nfapi_nr_pdsch_time_domain_alloc_type_e; + +typedef enum { + NFAPI_NR_PDSCH_MAPPING_TYPE_A=0, + NFAPI_NR_PDSCH_MAPPING_TYPE_B +} nfapi_nr_pdsch_mapping_type_e; + +typedef enum { + NFAPI_NR_PDSCH_RBG_CONFIG_TYPE1=0, + NFAPI_NR_PDSCH_RBG_CONFIG_TYPE2 +} nfapi_nr_pdsch_rbg_config_type_e; + +typedef enum { + NFAPI_NR_PRG_GRANULARITY_2=2, + NFAPI_NR_PRG_GRANULARITY_4=4, + NFAPI_NR_PRG_GRANULARITY_WIDEBAND +} nfapi_nr_prg_granularity_e; + +typedef enum { + NFAPI_NR_PRB_BUNDLING_TYPE_STATIC=0, + NFAPI_NR_PRB_BUNDLING_TYPE_DYNAMIC +} nfapi_nr_prb_bundling_type_e; + +typedef enum { + NFAPI_NR_MCS_TABLE_QAM64_LOW_SE=0, + NFAPI_NR_MCS_TABLE_QAM256 +} nfapi_nr_pdsch_mcs_table_e; + +typedef enum { + NFAPI_NR_DMRS_TYPE1=0, + NFAPI_NR_DMRS_TYPE2 +} nfapi_nr_dmrs_type_e; + // P7 Sub Structures typedef struct { @@ -445,13 +505,14 @@ typedef struct{ typedef struct { nfapi_tl_t tl; - uint8_t rnti; + uint16_t rnti; uint8_t rnti_type; uint8_t dci_format; /// Number of CRB in BWP that this DCI configures uint16_t n_RB_BWP; uint8_t config_type; - uint8_t search_space_type; + uint8_t search_space_type; + uint8_t common_search_space_type; uint8_t aggregation_level; uint8_t n_rb; uint8_t n_symb; @@ -484,14 +545,50 @@ typedef struct { } nfapi_nr_dl_config_bch_pdu; typedef struct { - nfapi_tl_t tl; - nfapi_nr_search_space_t sib1searchSpace; - nfapi_nr_coreset_t sib1ControlResourceSets; - nfapi_nr_search_space_t sibssearchSpace; - nfapi_nr_coreset_t sibsControlResourceSets; - nfapi_nr_search_space_t ra_SearchSpace; - nfapi_nr_coreset_t ra_ControlResourceSets; + nfapi_tl_t tl; + uint16_t length; + uint8_t pdu_index; + uint16_t rnti; + uint16_t nb_mod_symbols; + uint8_t time_allocation_type; + uint8_t freq_allocation_type; + uint8_t start_prb; + uint8_t n_prb; + uint8_t start_symbol; + uint8_t nb_symbols; + uint8_t mcs_idx; + uint8_t ndi; + uint8_t nb_codewords; + uint8_t nb_layers; + uint16_t coding_rate; + uint8_t modulation; + uint8_t modulation_order; + uint16_t transport_block_size; + uint8_t nb_re_dmrs; + uint8_t time_alloc_list_flag; + uint8_t time_alloc_list; + uint8_t rbg_list; + uint8_t virtual_resource_block_assignment_flag; + uint32_t resource_block_coding; + uint8_t redundancy_version; + uint8_t transport_blocks; + uint8_t transmission_scheme; + uint8_t number_of_subbands; + uint8_t codebook_index[NFAPI_MAX_NUM_SUBBANDS]; + uint8_t ue_category_capacity; + uint8_t pa; + uint8_t delta_power_offset_index; + uint8_t ngap; + uint8_t transmission_mode; + uint8_t num_bf_prb_per_subband; + uint8_t num_bf_vector; + nfapi_bf_vector_t bf_vector[NFAPI_MAX_BF_VECTORS]; }nfapi_nr_dl_config_dlsch_pdu_rel15_t; +#define NFAPI_NR_DL_CONFIG_REQUEST_DLSCH_PDU_REL15_TAG + +typedef struct { + nfapi_nr_dl_config_dlsch_pdu_rel15_t dlsch_pdu_rel15; +} nfapi_nr_dl_config_dlsch_pdu; typedef struct { nfapi_tl_t tl; @@ -501,15 +598,15 @@ typedef struct { typedef struct { -} nfapi_nr_dl_config_nbch_pdu; +} nfapi_nr_dl_config_nbch_pdu_rel15_t; typedef struct { -} nfapi_nr_dl_config_npdcch_pdu; +} nfapi_nr_dl_config_npdcch_pdu_rel15_t; typedef struct { -} nfapi_nr_dl_config_ndlsch_pdu; +} nfapi_nr_dl_config_ndlsch_pdu_rel15_t; typedef struct { nfapi_nr_dl_config_dci_dl_pdu_rel15_t dci_dl_pdu_rel15; @@ -524,11 +621,11 @@ typedef struct { union { nfapi_nr_dl_config_dci_dl_pdu dci_dl_pdu; nfapi_nr_dl_config_bch_pdu_rel15_t bch_pdu_rel15; - nfapi_nr_dl_config_dlsch_pdu_rel15_t dlsch_pdu_rel15; + nfapi_nr_dl_config_dlsch_pdu dlsch_pdu; nfapi_nr_dl_config_pch_pdu_rel15_t pch_pdu_rel15; - //nfapi_nr_dl_config_nbch_pdu_rel15_t nbch_pdu_rel15; - //nfapi_nr_dl_config_npdcch_pdu_rel15_t npdcch_pdu_rel15; - //nfapi_nr_dl_config_ndlsch_pdu_rel15_t ndlsch_pdu_rel15; + nfapi_nr_dl_config_nbch_pdu_rel15_t nbch_pdu_rel15; + nfapi_nr_dl_config_npdcch_pdu_rel15_t npdcch_pdu_rel15; + nfapi_nr_dl_config_ndlsch_pdu_rel15_t ndlsch_pdu_rel15; }; } nfapi_nr_dl_config_request_pdu_t; diff --git a/oaienv b/oaienv index 4cee684b3d..61b607c605 100644 --- a/oaienv +++ b/oaienv @@ -18,6 +18,6 @@ alias oailte='cd $OPENAIR_TARGETS/RT/USER' alias oais='cd $OPENAIR_TARGETS/SIMU/USER' alias oaiex='cd $OPENAIR_TARGETS/SIMU/EXAMPLES' -#export IIOD_REMOTE=192.168.121.32 -export IIOD_REMOTE=192.168.1.11 +export IIOD_REMOTE=192.168.121.32 +#export IIOD_REMOTE=192.168.1.11 diff --git a/openair1/PHY/CODING/TESTBENCH/ldpctest.c b/openair1/PHY/CODING/TESTBENCH/ldpctest.c index 99906721c3..19708b3dca 100644 --- a/openair1/PHY/CODING/TESTBENCH/ldpctest.c +++ b/openair1/PHY/CODING/TESTBENCH/ldpctest.c @@ -24,6 +24,7 @@ #include #include +#include "assertions.h" #include "SIMULATION/TOOLS/sim.h" #include "PHY/CODING/nrLDPC_encoder/defs.h" #include "PHY/CODING/nrLDPC_decoder/nrLDPC_decoder.h" @@ -38,6 +39,10 @@ # endif #endif +#define NR_LDPC_PROFILER_DETAIL +#define NR_LDPC_ENABLE_PARITY_CHECK + + // 4-bit quantizer char quantize4bit(double D,double x) { @@ -53,13 +58,33 @@ char quantize4bit(double D,double x) return((char)qxd); } +char quantize8bit(double D,double x) +{ + double qxd; + //char maxlev; + qxd = floor(x/D); + + //maxlev = 1<<(B-1); + + //printf("x=%f,qxd=%f,maxlev=%d\n",x,qxd, maxlev); + + if (qxd <= -128) + qxd = -128; + else if (qxd >= 128) + qxd = 127; + + return((char)qxd); +} + char quantize(double D,double x,unsigned char B) { double qxd; - char maxlev; + short maxlev; qxd = floor(x/D); - // printf("x=%f,qxd=%f\n",x,qxd); - maxlev = 1<<(B-1); + + maxlev = 1<<(B-1);//(char)(pow(2,B-1)); + + //printf("x=%f,qxd=%f,maxlev=%d\n",x,qxd, maxlev); if (qxd <= -maxlev) qxd = -maxlev; @@ -69,45 +94,78 @@ char quantize(double D,double x,unsigned char B) return((char)qxd); } +typedef struct { +double n_iter_mean; +double n_iter_std; +int n_iter_max; +} n_iter_stats_t; + #define MAX_BLOCK_LENGTH 8448 +short lift_size[51]= {2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,18,20,22,24,26,28,30,32,36,40,44,48,52,56,60,64,72,80,88,96,104,112,120,128,144,160,176,192,208,224,240,256,288,320,352,384}; + int test_ldpc(short No_iteration, - //int nom_rate, - //int denom_rate, + int nom_rate, + int denom_rate, double SNR, unsigned char qbits, - unsigned short block_length, + short block_length, unsigned int ntrials, + int n_segments, unsigned int *errors, - unsigned int *crc_misses) + unsigned int *errors_bit, + double *errors_bit_uncoded, + unsigned int *crc_misses, + time_stats_t *time_optim, + time_stats_t *time_decoder, + n_iter_stats_t * dec_iter) { //clock initiate - time_stats_t time,time_optim,tinput,tprep,tparity,toutput; + //time_stats_t time,time_optim,tinput,tprep,tparity,toutput, time_decoder; + time_stats_t time, tinput,tprep,tparity,toutput; + double n_iter_mean = 0; + double n_iter_std = 0; + int n_iter_max = 0; + unsigned int segment_bler = 0; + + double sigma; + sigma = 1.0/sqrt(2*SNR); + opp_enabled=1; cpu_freq_GHz = get_cpu_freq_GHz(); //short test_input[block_length]; unsigned char *test_input[MAX_NUM_DLSCH_SEGMENTS]={NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL};; //short *c; //padded codeword - unsigned char *estimated_output; + unsigned char *estimated_output[MAX_NUM_DLSCH_SEGMENTS]; + unsigned char *estimated_output_bit[MAX_NUM_DLSCH_SEGMENTS]; + unsigned char *test_input_bit; unsigned char *channel_input[MAX_NUM_DLSCH_SEGMENTS]; + unsigned char *channel_output_uncoded[MAX_NUM_DLSCH_SEGMENTS]; unsigned char *channel_input_optim[MAX_NUM_DLSCH_SEGMENTS]; double *channel_output; - double *modulated_input; - char *channel_output_fixed; + double *modulated_input[MAX_NUM_DLSCH_SEGMENTS]; + char *channel_output_fixed[MAX_NUM_DLSCH_SEGMENTS]; unsigned int i,j,trial=0; - short BG,Zc,Kb; - //short nrows,ncols; - //int no_punctured_columns,removed_bit; + short BG,Zc,Kb,nrows;//,ncols; + int no_punctured_columns,removed_bit; int i1; + int R_ind = 0; //Table of possible lifting sizes - short lift_size[51]= {2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,18,20,22,24,26,28,30,32,36,40,44,48,52,56,60,64,72,80,88,96,104,112,120,128,144,160,176,192,208,224,240,256,288,320,352,384}; - int n_segments=8; - int N; //length encoded block +//short lift_size[51]= {2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,18,20,22,24,26,28,30,32,36,40,44,48,52,56,60,64,72,80,88,96,104,112,120,128,144,160,176,192,208,224,240,256,288,320,352,384}; + //int n_segments=1; + int code_rate_vec[8] = {15, 13, 25, 12, 23, 34, 56, 89}; + //double code_rate_actual_vec[8] = {0.2, 0.33333, 0.4, 0.5, 0.66667, 0.73333, 0.81481, 0.88}; t_nrLDPC_dec_params decParams; - int n_iter; + + t_nrLDPC_time_stats decoder_profiler; + t_nrLDPC_time_stats* p_decoder_profiler =&decoder_profiler ; + + int32_t n_iter = 0; *errors=0; + *errors_bit=0; + *errors_bit_uncoded=0; *crc_misses=0; // generate input block @@ -115,37 +173,454 @@ int test_ldpc(short No_iteration, test_input[j]=(unsigned char *)malloc16(sizeof(unsigned char) * block_length/8); channel_input[j] = (unsigned char *)malloc16(sizeof(unsigned char) * 68*384); channel_input_optim[j] = (unsigned char *)malloc16(sizeof(unsigned char) * 68*384); + channel_output_uncoded[j] = (unsigned char *)malloc16(sizeof(unsigned char) * 68*384); + estimated_output[j] = (unsigned char*) malloc16(sizeof(unsigned char) * block_length); + estimated_output_bit[j] = (unsigned char*) malloc16(sizeof(unsigned char) * block_length); + modulated_input[j] = (double *)malloc16(sizeof(double) * 68*384); + channel_output_fixed[j] = (char *)malloc16(sizeof( char) * 68*384); } - modulated_input = (double *)malloc(sizeof(double) * 68*384); - channel_output = (double *)malloc(sizeof(double) * 68*384); - channel_output_fixed = (char *)malloc16(sizeof(char) * 68*384); - estimated_output = (unsigned char*) malloc16(sizeof(unsigned char) * block_length); - - memset(channel_output_fixed,0,sizeof(char) * 68*384); + //modulated_input = (double *)malloc(sizeof(double) * 68*384); + //channel_output = (double *)malloc(sizeof(double) * 68*384); + //channel_output_fixed = (char *)malloc16(sizeof(char) * 68*384); + //modulated_input = (double *)calloc(68*384, sizeof(double)); + channel_output = (double *)calloc(68*384, sizeof(double)); + //channel_output_fixed = (double *)calloc(68*384, sizeof(double)); + //channel_output_fixed = (unsigned char*)calloc(68*384, sizeof(unsigned char*)); + //estimated_output = (unsigned char*) malloc16(sizeof(unsigned char) * block_length);///8); + //estimated_output_bit = (unsigned char*) malloc16(sizeof(unsigned char) * block_length); + test_input_bit = (unsigned char*) malloc16(sizeof(unsigned char) * block_length); reset_meas(&time); - reset_meas(&time_optim); + reset_meas(time_optim); + reset_meas(time_decoder); reset_meas(&tinput); reset_meas(&tprep); reset_meas(&tparity); reset_meas(&toutput); + //Reset Decoder profiles + reset_meas(&decoder_profiler.llr2llrProcBuf); + reset_meas(&decoder_profiler.llr2CnProcBuf); + reset_meas(&decoder_profiler.cnProc); + reset_meas(&decoder_profiler.cnProcPc); + reset_meas(&decoder_profiler.bnProc); + reset_meas(&decoder_profiler.bnProcPc); + reset_meas(&decoder_profiler.cn2bnProcBuf); + reset_meas(&decoder_profiler.bn2cnProcBuf); + reset_meas(&decoder_profiler.llrRes2llrOut); + reset_meas(&decoder_profiler.llr2bit); + //reset_meas(&decoder_profiler.total); for (j=0;j3840) && (block_length <= MAX_BLOCK_LENGTH)) + if (block_length>3840) + { + BG=1; + Kb = 22; + nrows=46; //parity check bits + //ncols=22; //info bits + } + else if (block_length<=3840) + { + BG=2; + nrows=42; //parity check bits + //ncols=10; // info bits + + if (block_length>640) + Kb = 10; + else if (block_length>560) + Kb = 9; + else if (block_length>192) + Kb = 8; + else + Kb = 6; + } + + if (nom_rate == 1) + if (denom_rate == 5) + if (BG == 2) + R_ind = 0; + else + printf("Not supported"); + else if (denom_rate == 3) + R_ind = 1; + else if (denom_rate == 2) + //R_ind = 3; + printf("Not supported"); + else + printf("Not supported"); + + else if (nom_rate == 2) + if (denom_rate == 5) + //R_ind = 2; + printf("Not supported"); + else if (denom_rate == 3) + R_ind = 4; + else + printf("Not supported"); + + else if ((nom_rate == 22) && (denom_rate == 30)) + //R_ind = 5; + printf("Not supported"); + else if ((nom_rate == 22) && (denom_rate == 27)) + //R_ind = 6; + printf("Not supported"); + else if ((nom_rate == 22) && (denom_rate == 25)) + if (BG == 1) + R_ind = 7; + else + printf("Not supported"); + else + printf("Not supported"); + + //find minimum value in all sets of lifting size + Zc=0; + + for (i1=0; i1 < 51; i1++) + { + if (lift_size[i1] >= (double) block_length/Kb) + { + Zc = lift_size[i1]; + //printf("%d\n",Zc); + break; + } + } + + printf("ldpc_test: codeword_length %d, n_segments %d, block_length %d, BG %d, Zc %d, Kb %d\n",n_segments *block_length, n_segments, block_length,BG, Zc, Kb); + no_punctured_columns=(int)((nrows-2)*Zc+block_length-block_length*(1/((float)nom_rate/(float)denom_rate)))/Zc; + // printf("puncture:%d\n",no_punctured_columns); + removed_bit=(nrows-no_punctured_columns-2) * Zc+block_length-(int)(block_length/((float)nom_rate/(float)denom_rate)); + if (ntrials==0) + ldpc_encoder_orig(test_input[0],channel_input[0], block_length, BG, 1); + + for (trial=0; trial < ntrials; trial++) + { + segment_bler = 0; + //// encoder + start_meas(&time); + for(j=0;j>3; i++) + { + //printf("block_length>>3: %d \n",block_length>>3); + /// printf("i: %d \n",i); + ///printf("estimated_output[%d]: %d \n",i,estimated_output[i]); + ///printf("test_input[0][%d]: %d \n",i,test_input[0][i]); + if (estimated_output[j][i] != test_input[j][i]) + { + //////printf("error pos %d (%d, %d)\n\n",i,estimated_output[i],test_input[0][i]); + segment_bler = segment_bler + 1; + break; + } + } + + for (i=0; i>(i&7); + test_input_bit[i] = (test_input[j][i/8]&(1<<(i&7)))>>(i&7); // Further correct for multiple segments + if (estimated_output_bit[j][i] != test_input_bit[i]) + { + *errors_bit = (*errors_bit) + 1; + } + } + + //if (*errors == 1000) + //break; + + n_iter_mean = n_iter_mean + n_iter; + n_iter_std = n_iter_std + pow(n_iter-1,2); + + if ( n_iter > n_iter_max ) + n_iter_max = n_iter; + + } // end segments + + if (segment_bler != 0) + *errors = (*errors) + 1; + + } + /*else if (trial==0) + printf("decoder is not supported\n");*/ + } + + + dec_iter->n_iter_mean = n_iter_mean/(double)ntrials/(double)n_segments - 1; + dec_iter->n_iter_std = sqrt(n_iter_std/(double)ntrials/(double)n_segments - pow(n_iter_mean/(double)ntrials/(double)n_segments - 1,2)); + dec_iter->n_iter_max = n_iter_max -1; + + *errors_bit_uncoded = *errors_bit_uncoded / (double)((Kb+nrows-no_punctured_columns-2) * Zc-removed_bit); + + for(j=0;j 1280, 2401 -> 2560 + + short No_iteration=5; + int n_segments=1; + //double rate=0.333; + int nom_rate=1; + int denom_rate=3; + double SNR0=-2.0,SNR,SNR_lin; + unsigned char qbits=8; + unsigned int decoded_errors[10000]; // initiate the size of matrix equivalent to size of SNR + int c,i=0, i1 = 0; + + int n_trials = 1; + double SNR_step = 0.1; + + randominit(0); + int test_uncoded= 0; + + time_stats_t time_optim[10], time_decoder[10]; + n_iter_stats_t dec_iter[3]; + + short BG,Zc,Kb; + + while ((c = getopt (argc, argv, "q:r:s:S:l:n:d:i:t:u:h")) != -1) + switch (c) + { + case 'q': + qbits = atoi(optarg); + break; + + case 'r': + nom_rate = atoi(optarg); + break; + + case 'd': + denom_rate = atoi(optarg); + break; + + case 'l': + block_length = atoi(optarg); + break; + + case 'n': + n_trials = atoi(optarg); + break; + + case 's': + SNR0 = atof(optarg); + break; + + case 'S': + n_segments = atof(optarg); + break; + + case 't': + SNR_step = atof(optarg); + break; + + case 'i': + No_iteration = atoi(optarg); + break; + + case 'u': + test_uncoded = atoi(optarg); + break; + + case 'h': + default: + printf("CURRENTLY SUPPORTED CODE RATES: \n"); + printf("BG1 (blocklength > 3840): 1/3, 2/3, 22/25 (8/9) \n"); + printf("BG2 (blocklength <= 3840): 1/5, 1/3, 2/3 \n\n"); + printf("-h This message\n"); + printf("-q Quantization bits, Default: 8\n"); + printf("-r Nominator rate, (1, 2, 22), Default: 1\n"); + printf("-d Denominator rate, (3, 5, 25), Default: 1\n"); + printf("-l Block length (l > 3840 -> BG1, rest BG2 ), Default: 8448\n"); + printf("-n Number of simulation trials, Default: 1\n"); + //printf("-M MCS2 for TB 2\n"); + printf("-s SNR per information bit (EbNo) in dB, Default: -2\n"); + printf("-S Number of segments (Maximum: 8), Default: 1\n"); + printf("-t SNR simulation step, Default: 0.1\n"); + printf("-i Max decoder iterations, Default: 5\n"); + printf("-u Set SNR per coded bit, Default: 0\n"); + exit(1); + break; + } + cpu_freq_GHz = get_cpu_freq_GHz(); + //printf("the decoder supports BG2, Kb=10, Z=128 & 256\n"); + //printf(" range of blocklength: 1201 -> 1280, 2401 -> 2560\n"); + printf("block length %d: \n", block_length); + printf("n_trials %d: \n", n_trials); + printf("SNR0 %f: \n", SNR0); + + + + + //for (block_length=8;block_length<=MAX_BLOCK_LENGTH;block_length+=8) + + + //determine number of bits in codeword + if (block_length>3840) { BG=1; Kb = 22; //nrows=46; //parity check bits //ncols=22; //info bits } - else if (block_length<=3840) + else if (block_length<=3840) { BG=2; //nrows=42; //parity check bits @@ -159,14 +634,11 @@ int test_ldpc(short No_iteration, Kb = 8; else Kb = 6; - } - else { - printf("block_length %d not supported\n",block_length); - exit(-1); - } - + } + //find minimum value in all sets of lifting size Zc=0; + for (i1=0; i1 < 51; i1++) { if (lift_size[i1] >= (double) block_length/Kb) @@ -177,193 +649,87 @@ int test_ldpc(short No_iteration, } } - if (BG==1) - N=66*Zc; - else if (BG==2) - N=50*Zc; + char fname[200]; + sprintf(fname,"ldpctest_BG_%d_Zc_%d_rate_%d-%d_block_length_%d_maxit_%d.txt",BG,Zc,nom_rate,denom_rate,block_length, No_iteration); + FILE *fd=fopen(fname,"w"); + AssertFatal(fd!=NULL,"cannot open %s\n",fname); - printf("ldpc_test: block_length %d, BG %d, Zc %d, Kb %d, N %d\n",block_length,BG, Zc, Kb, N); - //no_punctured_columns=(int)((nrows-2)*Zc+block_length-block_length*(1/((float)nom_rate/(float)denom_rate)))/Zc; - // printf("puncture:%d\n",no_punctured_columns); - //removed_bit=(nrows-no_punctured_columns-2) * Zc+block_length-(int)(block_length/((float)nom_rate/(float)denom_rate)); - if (ntrials==0) - ldpc_encoder_orig(test_input[0],channel_input[0], block_length, BG, 1); - - for (trial=0; trial < ntrials; trial++) + fprintf(fd,"SNR BLER BER UNCODED_BER ENCODER_MEAN ENCODER_STD ENCODER_MAX DECODER_TIME_MEAN DECODER_TIME_STD DECODER_TIME_MAX DECODER_ITER_MEAN DECODER_ITER_STD DECODER_ITER_MAX\n"); + + for (SNR=SNR0;SNR1.0) - channel_output_fixed[i]=127; - else if (channel_output[i]<-1.0) - channel_output_fixed[i]=-128; - else - channel_output_fixed[i] = (char) ((channel_output[i]*128)<0?(channel_output[i]*128-0.5):(channel_output[i]*128+0.5)); //fixed point 9-7 - //printf("llr[%d]=%d (%f)\n",i,channel_output_fixed[i],channel_output[i]); - } - - /* - the LDPC decoder supports several rates and the input has to be adapted accordingly (by padding with zeros to the next lower supported rate). Moreover, the first 2*Z LLRs are zero (since they are punctured and not transmitted) then you have the LLRs corresponding to the systematic bits. After that come the filler bits, the encoder uses 0 for the filler bits hence for the decoder you need to put 127 (max LLR for bit 0) at their place. Then come the LLRs corresponding to the punctured bits followed by zeros to pad for the supported decoder rate. - */ - for (i=0;i<2*Zc;i++) - channel_output_fixed[i] = 0; - - decParams.BG=BG; - decParams.Z=Zc; - decParams.R=13; - decParams.numMaxIter=6; - decParams.outMode = nrLDPC_outMode_BIT; - - // decode the sequence - // decoder supports BG2, Z=128 & 256 - //estimated_output=ldpc_decoder(channel_output_fixed, block_length, No_iteration, (double)((float)nom_rate/(float)denom_rate)); - - n_iter = nrLDPC_decoder(&decParams, (int8_t*) channel_output_fixed, (int8_t*) estimated_output, NULL); - printf("nrLDPC_decoder n_iter=%d\n",n_iter); - - //for (i=(Kb+nrows) * Zc-5;i<(Kb+nrows) * Zc;i++) - // printf("esimated_output[%d]=%d\n",i,esimated_output[i]); - - //count errors - for (i=0; i>3; i++) - { - if (estimated_output[i] != test_input[0][i]) - { - printf("error pos %d (%d, %d)\n",i,estimated_output[i],test_input[0][i]); - *errors = (*errors) + 1; - break; - } - } - - } - - for(j=0;j 1280, 2401 -> 2560 - short No_iteration=25; - //double rate=0.333; - //int nom_rate=1; - //int denom_rate=3; - double SNR0=-2.0,SNR,SNR_lin; - unsigned char qbits=4; - unsigned int decoded_errors[100]; // initiate the size of matrix equivalent to size of SNR - int c,i=0; - - int n_trials = 1; - - randominit(0); - - while ((c = getopt (argc, argv, "q:s:l:n:h")) != -1) - switch (c) - { - case 'q': - qbits = atoi(optarg); - break; - - case 'l': - block_length = atoi(optarg); - break; - - case 'n': - n_trials = atoi(optarg); - break; - - case 's': - SNR0 = atoi(optarg); - break; - - default: - case 'h': - printf("usage ldpctest -h -l blocklength -n n_trials -s SNR\n"); - exit(0); - } - - //printf("the decoder supports BG2, Kb=10, Z=128 & 256\n"); - //printf(" range of blocklength: 1201 -> 1280, 2401 -> 2560\n"); - //printf("rate: %d/%d\n",nom_rate,denom_rate); - printf("block length %d: \n", block_length); - printf("n_trials %d: \n", n_trials); - printf("SNR0 %f: \n", SNR0); - - //for (block_length=8;block_length<=MAX_BLOCK_LENGTH;block_length+=8) - for (SNR=SNR0;SNRn_iter_mean); + printf("SNR %f, Std iterations: %f\n",SNR, dec_iter->n_iter_std); + printf("SNR %f, Max iterations: %d\n",SNR, dec_iter->n_iter_max); + printf("\n"); + printf("Encoding time mean: %15.3f us\n",(double)time_optim->diff/time_optim->trials/1000.0/cpu_freq_GHz); + printf("Encoding time std: %15.3f us\n",sqrt((double)time_optim->diff_square/time_optim->trials/pow(1000,2)/pow(cpu_freq_GHz,2)-pow((double)time_optim->diff/time_optim->trials/1000.0/cpu_freq_GHz,2))); + printf("Encoding time max: %15.3f us\n",(double)time_optim->max/1000.0/cpu_freq_GHz); + printf("\n"); + printf("Decoding time mean: %15.3f us\n",(double)time_decoder->diff/time_decoder->trials/1000.0/cpu_freq_GHz); + printf("Decoding time std: %15.3f us\n",sqrt((double)time_decoder->diff_square/time_decoder->trials/pow(1000,2)/pow(cpu_freq_GHz,2)-pow((double)time_decoder->diff/time_decoder->trials/1000.0/cpu_freq_GHz,2))); + printf("Decoding time max: %15.3f us\n",(double)time_decoder->max/1000.0/cpu_freq_GHz); + + fprintf(fd,"%f %f %f %f %f %f %f %f %f %f %f %f %d \n", + SNR, + (double)decoded_errors[i]/(double)n_trials , + (double)errors_bit/(double)n_trials/(double)block_length/(double)n_segments , + errors_bit_uncoded/(double)n_trials/(double)n_segments , + (double)time_optim->diff/time_optim->trials/1000.0/cpu_freq_GHz, + sqrt((double)time_optim->diff_square/time_optim->trials/pow(1000,2)/pow(cpu_freq_GHz,2)-pow((double)time_optim->diff/time_optim->trials/1000.0/cpu_freq_GHz,2)), + (double)time_optim->max/1000.0/cpu_freq_GHz, + (double)time_decoder->diff/time_decoder->trials/1000.0/cpu_freq_GHz, + sqrt((double)time_decoder->diff_square/time_decoder->trials/pow(1000,2)/pow(cpu_freq_GHz,2)-pow((double)time_decoder->diff/time_decoder->trials/1000.0/cpu_freq_GHz,2)), + (double)time_decoder->max/1000.0/cpu_freq_GHz, + dec_iter->n_iter_mean, + dec_iter->n_iter_std, + dec_iter->n_iter_max + ); + + if (decoded_errors[i] == 0) + break; i=i+1; + + } + fclose(fd); return(0); } + + + + + + + diff --git a/openair1/PHY/CODING/TESTBENCH/polartest.c b/openair1/PHY/CODING/TESTBENCH/polartest.c index 225bd54a0e..6d50070b86 100644 --- a/openair1/PHY/CODING/TESTBENCH/polartest.c +++ b/openair1/PHY/CODING/TESTBENCH/polartest.c @@ -7,6 +7,8 @@ #include #include "PHY/CODING/nrPolar_tools/nr_polar_defs.h" +#include "PHY/CODING/nrPolar_tools/nr_polar_pbch_defs.h" +#include "PHY/CODING/nrPolar_tools/nr_polar_uci_defs.h" #include "PHY/CODING/coding_defs.h" #include "SIMULATION/TOOLS/sim.h" diff --git a/openair1/PHY/CODING/coding_defs.h b/openair1/PHY/CODING/coding_defs.h index 8d2ae12a7a..f6e520b226 100644 --- a/openair1/PHY/CODING/coding_defs.h +++ b/openair1/PHY/CODING/coding_defs.h @@ -38,6 +38,11 @@ #define MAX_TURBO_ITERATIONS_MBSFN 8 #define MAX_TURBO_ITERATIONS max_turbo_iterations +#define MAX_LDPC_ITERATIONS 4 +#define MAX_LDPC_ITERATIONS_MBSFN 4 + +#define LTE_NULL 2 +#define NR_NULL 2 typedef struct { unsigned short nb_bits; unsigned short f1; @@ -443,16 +448,11 @@ unsigned int crcbit (unsigned char * inputptr, int octetlen, unsigned int poly); int16_t reverseBits(int32_t ,int32_t); void phy_viterbi_dot11(int8_t *,uint8_t *,uint16_t); -decoder_if_t phy_threegpplte_turbo_decoder; -decoder_if_t phy_threegpplte_turbo_decoder8; -decoder_if_t phy_threegpplte_turbo_decoder16; - int32_t nr_segmentation(unsigned char *input_buffer, unsigned char **output_buffers, unsigned int B, unsigned int *C, - unsigned int *Kplus, - unsigned int *Kminus, + unsigned int *K, unsigned int *Zout, unsigned int *F); @@ -463,4 +463,40 @@ uint32_t nr_compute_tbs(uint8_t mcs, uint16_t length_dmrs, uint8_t Nl); +void nr_interleaving_ldpc(uint32_t E, uint8_t Qm, uint8_t *e,uint8_t *f); + +void nr_deinterleaving_ldpc(uint32_t E, uint8_t Qm, int16_t *e,int16_t *f); + +uint32_t nr_rate_matching_ldpc(uint8_t Ilbrm, + uint32_t Tbslbrm, + uint8_t BG, + uint16_t Z, + uint32_t G, + uint8_t *w, + uint8_t *e, + uint8_t C, + uint8_t rvidx, + uint8_t Qm, + uint8_t Nl, + uint8_t r); + +int nr_rate_matching_ldpc_rx(uint8_t Ilbrm, + uint32_t Tbslbrm, + uint8_t BG, + uint16_t Z, + uint32_t G, + int16_t *w, + int16_t *soft_input, + uint8_t C, + uint8_t rvidx, + uint8_t clear, + uint8_t Qm, + uint8_t Nl, + uint8_t r, + uint32_t *E_out); + +decoder_if_t phy_threegpplte_turbo_decoder; +decoder_if_t phy_threegpplte_turbo_decoder8; +decoder_if_t phy_threegpplte_turbo_decoder16; + #endif diff --git a/openair1/PHY/CODING/crc_byte.c b/openair1/PHY/CODING/crc_byte.c index c2b016920a..d728d6b9ad 100644 --- a/openair1/PHY/CODING/crc_byte.c +++ b/openair1/PHY/CODING/crc_byte.c @@ -35,12 +35,14 @@ /*ref 36-212 v8.6.0 , pp 8-9 */ /* the highest degree is set by default */ + unsigned int poly24a = 0x864cfb00; // 1000 0110 0100 1100 1111 1011 // D^24 + D^23 + D^18 + D^17 + D^14 + D^11 + D^10 + D^7 + D^6 + D^5 + D^4 + D^3 + D + 1 unsigned int poly24b = 0x80006300; // 1000 0000 0000 0000 0110 0011 // D^24 + D^23 + D^6 + D^5 + D + 1 unsigned int poly24c = 0xb2b11700; // 1011 0010 1011 0001 0001 0111 // D^24+D^23+D^21+D^20+D^17+D^15+D^13+D^12+D^8+D^4+D^2+D+1 + unsigned int poly16 = 0x10210000; // 0001 0000 0010 0001 D^16 + D^12 + D^5 + 1 unsigned int poly12 = 0x80F00000; // 1000 0000 1111 D^12 + D^11 + D^3 + D^2 + D + 1 unsigned int poly8 = 0x9B000000; // 1001 1011 D^8 + D^7 + D^4 + D^3 + D + 1 @@ -253,6 +255,68 @@ unsigned int crcPayload(unsigned char * inptr, int bitlen, uint32_t* crc256Table return crc; } +int check_crc(uint8_t* decoded_bytes, uint32_t n, uint32_t F, uint8_t crc_type) +{ + uint32_t crc=0,oldcrc=0; + uint8_t crc_len,temp; + + switch (crc_type) { + case CRC24_A: + case CRC24_B: + crc_len=3; + break; + + case CRC16: + crc_len=2; + break; + + case CRC8: + crc_len=1; + break; + + default: + crc_len=3; + } + + for (int i=0; i>3)-crc_len+i])<<((crc_len-1-i)<<3); + + switch (crc_type) { + + case CRC24_A: + oldcrc&=0x00ffffff; + crc = crc24a(decoded_bytes, + n-24)>>8; + + break; + + case CRC24_B: + oldcrc&=0x00ffffff; + crc = crc24b(decoded_bytes, + n-24)>>8; + + break; + + case CRC16: + oldcrc&=0x0000ffff; + crc = crc16(decoded_bytes, + n-16)>>16; + + break; + + case CRC8: + oldcrc&=0x000000ff; + crc = crc8(decoded_bytes, + n-8)>>24; + break; + } + + if (crc == oldcrc) + return(1); + else + return(0); + +} #ifdef DEBUG_CRC diff --git a/openair1/PHY/CODING/nrLDPC_decoder/nrLDPC_decoder.c b/openair1/PHY/CODING/nrLDPC_decoder/nrLDPC_decoder.c index 53966befff..fbdb9a3fa4 100644 --- a/openair1/PHY/CODING/nrLDPC_decoder/nrLDPC_decoder.c +++ b/openair1/PHY/CODING/nrLDPC_decoder/nrLDPC_decoder.c @@ -38,6 +38,9 @@ #include "nrLDPC_cnProc.h" #include "nrLDPC_bnProc.h" +#define NR_LDPC_ENABLE_PARITY_CHECK +//#define NR_LDPC_PROFILER_DETAIL + #ifdef NR_LDPC_DEBUG_MODE #include "nrLDPC_tools/nrLDPC_debug.h" #endif diff --git a/openair1/PHY/CODING/nrLDPC_encoder/defs.h b/openair1/PHY/CODING/nrLDPC_encoder/defs.h index 629d50c79c..4230056f31 100644 --- a/openair1/PHY/CODING/nrLDPC_encoder/defs.h +++ b/openair1/PHY/CODING/nrLDPC_encoder/defs.h @@ -38,6 +38,7 @@ int encode_parity_check_part_orig(unsigned char *c,unsigned char *d, short BG,sh void encode_parity_check_part_optim(uint8_t *c,uint8_t *d, short BG,short Zc,short Kb); int ldpc_encoder_optim(unsigned char *test_input,unsigned char *channel_input,short block_length,short BG,time_stats_t *tinput,time_stats_t *tprep,time_stats_t *tparity,time_stats_t *toutput); int ldpc_encoder_optim_8seg(unsigned char **test_input,unsigned char **channel_input,short block_length,short BG,int n_segments,time_stats_t *tinput,time_stats_t *tprep,time_stats_t *tparity,time_stats_t *toutput); +int ldpc_encoder_optim_8seg_multi(unsigned char **test_input,unsigned char **channel_input,short block_length, short BG, int n_segments,unsigned int macro_num, time_stats_t *tinput,time_stats_t *tprep,time_stats_t *tparity,time_stats_t *toutput); /*ldpc_generate_coefficient.c*/ int ldpc_encoder_orig(unsigned char *test_input,unsigned char *channel_input,short block_length,short BG,unsigned char gen_code); diff --git a/openair1/PHY/CODING/nrLDPC_encoder/ldpc_encoder.c b/openair1/PHY/CODING/nrLDPC_encoder/ldpc_encoder.c index b97b780333..a225730845 100644 --- a/openair1/PHY/CODING/nrLDPC_encoder/ldpc_encoder.c +++ b/openair1/PHY/CODING/nrLDPC_encoder/ldpc_encoder.c @@ -68,7 +68,7 @@ int encode_parity_check_part_orig(unsigned char *c,unsigned char *d, short BG,sh printf("problem with BG\n"); return(-1); } - + no_punctured_columns=(int)((nrows-2)*Zc+block_length-block_length*3)/Zc; diff --git a/openair1/PHY/CODING/nrLDPC_encoder/ldpc_encoder2.c b/openair1/PHY/CODING/nrLDPC_encoder/ldpc_encoder2.c index 8f19de8d11..be1a87e6bf 100644 --- a/openair1/PHY/CODING/nrLDPC_encoder/ldpc_encoder2.c +++ b/openair1/PHY/CODING/nrLDPC_encoder/ldpc_encoder2.c @@ -201,13 +201,7 @@ void encode_parity_check_part_optim(uint8_t *c,uint8_t *d, short BG,short Zc,sho int ldpc_encoder_optim(unsigned char *test_input,unsigned char *channel_input,short block_length,short BG,time_stats_t *tinput,time_stats_t *tprep,time_stats_t *tparity,time_stats_t *toutput) { - short Zc; - - //initialize for BG == 1 - short Kb = 22; - short nrows = 46;//parity check bits - short ncols = 22;//info bits - + short Zc,Kb,nrows,ncols; int i,i1; int no_punctured_columns,removed_bit; @@ -217,20 +211,30 @@ int ldpc_encoder_optim(unsigned char *test_input,unsigned char *channel_input,sh int simd_size; //determine number of bits in codeword - if (BG==2) - { - nrows=42; //parity check bits - ncols=10; // info bits - - if (block_length>640) - Kb = 10; - else if (block_length>560) - Kb = 9; - else if (block_length>192) - Kb = 8; - else - Kb = 6; - } + //if (block_length>3840) + if (BG==1) + { + //BG=1; + Kb = 22; + nrows=46; //parity check bits + ncols=22; //info bits + } + //else if (block_length<=3840) + else if (BG==2) + { + //BG=2; + nrows=42; //parity check bits + ncols=10; // info bits + + if (block_length>640) + Kb = 10; + else if (block_length>560) + Kb = 9; + else if (block_length>192) + Kb = 8; + else + Kb = 6; + } //find minimum value in all sets of lifting size Zc=0; @@ -316,13 +320,7 @@ int ldpc_encoder_optim(unsigned char *test_input,unsigned char *channel_input,sh int ldpc_encoder_optim_8seg(unsigned char **test_input,unsigned char **channel_input,short block_length,short BG,int n_segments,time_stats_t *tinput,time_stats_t *tprep,time_stats_t *tparity,time_stats_t *toutput) { - short Zc; - - //initialize for BG == 1 - short Kb = 22; - short nrows = 46;//parity check bits - short ncols = 22;//info bits - + short Zc,Kb,nrows,ncols; int i,i1,j; int no_punctured_columns,removed_bit; //Table of possible lifting sizes @@ -349,8 +347,18 @@ int ldpc_encoder_optim_8seg(unsigned char **test_input,unsigned char **channel_i AssertFatal(n_segments>0&&n_segments<=8,"0 < n_segments %d <= 8\n",n_segments); //determine number of bits in codeword - if (BG==2) + //if (block_length>3840) + if (BG==1) { + //BG=1; + Kb = 22; + nrows=46; //parity check bits + ncols=22; //info bits + } + //else if (block_length<=3840) + else if (BG==2) + { + //BG=2; nrows=42; //parity check bits ncols=10; // info bits @@ -362,7 +370,7 @@ int ldpc_encoder_optim_8seg(unsigned char **test_input,unsigned char **channel_i Kb = 8; else Kb = 6; - } + } //find minimum value in all sets of lifting size Zc=0; @@ -481,12 +489,12 @@ int ldpc_encoder_optim_8seg(unsigned char **test_input,unsigned char **channel_i // if (((block_length-(2*Zc))&31)>0) l1++; for (i=0;i0) l2++; for (i1=0;i1>j)&1; for (i=0;i<((nrows-no_punctured_columns) * Zc-removed_bit);i++) - for (j=0; j>j)&1; } @@ -508,3 +516,217 @@ int ldpc_encoder_optim_8seg(unsigned char **test_input,unsigned char **channel_i stop_meas(toutput); return 0; } + +int ldpc_encoder_optim_8seg_multi(unsigned char **test_input,unsigned char **channel_input,short block_length, short BG, int n_segments,unsigned int macro_num, time_stats_t *tinput,time_stats_t *tprep,time_stats_t *tparity,time_stats_t *toutput) +{ + + short Zc,Kb,nrows,ncols; + int i,i1,j; + int no_punctured_columns,removed_bit; + //Table of possible lifting sizes + short lift_size[51]= {2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,18,20,22,24,26,28,30,32,36,40,44,48,52,56,60,64,72,80,88,96,104,112,120,128,144,160,176,192,208,224,240,256,288,320,352,384}; + char temp; + int simd_size; + int macro_segment, macro_segment_end; + + + macro_segment = 8*macro_num; + macro_segment_end = (n_segments > 8*(macro_num+1)) ? 8*(macro_num+1) : n_segments; + ///printf("macro_segment: %d\n", macro_segment); + ///printf("macro_segment_end: %d\n", macro_segment_end ); + +#ifdef __AVX2__ + __m256i shufmask = _mm256_set_epi64x(0x0303030303030303, 0x0202020202020202,0x0101010101010101, 0x0000000000000000); + __m256i andmask = _mm256_set1_epi64x(0x8040201008040201); // every 8 bits -> 8 bytes, pattern repeats. + __m256i zero256 = _mm256_setzero_si256(); + __m256i masks[8]; + register __m256i c256; + masks[0] = _mm256_set1_epi8(0x1); + masks[1] = _mm256_set1_epi8(0x2); + masks[2] = _mm256_set1_epi8(0x4); + masks[3] = _mm256_set1_epi8(0x8); + masks[4] = _mm256_set1_epi8(0x10); + masks[5] = _mm256_set1_epi8(0x20); + masks[6] = _mm256_set1_epi8(0x40); + masks[7] = _mm256_set1_epi8(0x80); +#endif + + ///AssertFatal(n_segments>0&&n_segments<=8,"0 < n_segments %d <= 8\n",n_segments); + + //determine number of bits in codeword + if (BG==1) + { + BG=1; + Kb = 22; + nrows=46; //parity check bits + ncols=22; //info bits + } + else if (BG==2) + { + BG=2; + nrows=42; //parity check bits + ncols=10; // info bits + + if (block_length>640) + Kb = 10; + else if (block_length>560) + Kb = 9; + else if (block_length>192) + Kb = 8; + else + Kb = 6; + } + + //find minimum value in all sets of lifting size + Zc=0; + for (i1=0; i1 < 51; i1++) + { + if (lift_size[i1] >= (double) block_length/Kb) + { + Zc = lift_size[i1]; + //printf("%d\n",Zc); + break; + } + } + +#ifdef DEBUG_LDPC + LOG_D(PHY,"ldpc_encoder_optim_8seg: BG %d, Zc %d, Kb %d, block_length %d, segments %d\n",BG,Zc,Kb,block_length,n_segments); + LOG_D(PHY,"ldpc_encoder_optim_8seg: PDU (seg 0) %x %x %x %x\n",test_input[0][0],test_input[0][1],test_input[0][2],test_input[0][3]); +#endif + + AssertFatal(Zc>0,"no valid Zc found for block length %d\n",block_length); + + if ((Zc&31) > 0) simd_size = 16; + else simd_size = 32; + + unsigned char c[22*Zc] __attribute__((aligned(32))); //padded input, unpacked, max size + unsigned char d[46*Zc] __attribute__((aligned(32))); //coded parity part output, unpacked, max size + + unsigned char c_extension[2*22*Zc*simd_size] __attribute__((aligned(32))); //double size matrix of c + + // calculate number of punctured bits + no_punctured_columns=(int)((nrows-2)*Zc+block_length-block_length*3)/Zc; + removed_bit=(nrows-no_punctured_columns-2) * Zc+block_length-(int)(block_length*3); + // printf("%d\n",no_punctured_columns); + // printf("%d\n",removed_bit); + // unpack input + memset(c,0,sizeof(unsigned char) * ncols * Zc); + memset(d,0,sizeof(unsigned char) * nrows * Zc); + + start_meas(tinput); +#if 0 + for (i=0; i>(i&7); + //printf("c(%d,%d)=%d\n",j,i,temp); + c[i] |= (temp << (j-macro_segment)); + } + } +#else +#ifdef __AVX2__ + for (i=0; i>5; i++) { + c256 = _mm256_and_si256(_mm256_cmpeq_epi8(_mm256_andnot_si256(_mm256_shuffle_epi8(_mm256_set1_epi32(((uint32_t*)test_input[macro_segment])[i]), shufmask),andmask),zero256),masks[0]); + //for (j=1; j>5)<<5;i>(i&7); + //printf("c(%d,%d)=%d\n",j,i,temp); + c[i] |= (temp << (j-macro_segment)); + } + } +#else + AssertFatal(1==0,"Need AVX2 for this\n"); +#endif +#endif + + stop_meas(tinput); + + if ((BG==1 && Zc>176) || (BG==2 && Zc>64)) { + // extend matrix + start_meas(tprep); + for (i1=0; i1 < ncols; i1++) + { + memcpy(&c_extension[2*i1*Zc], &c[i1*Zc], Zc*sizeof(unsigned char)); + memcpy(&c_extension[(2*i1+1)*Zc], &c[i1*Zc], Zc*sizeof(unsigned char)); + } + for (i1=1;i1>5; + uint32_t l2 = ((nrows-no_punctured_columns) * Zc-removed_bit)>>5; + __m256i *c256p = (__m256i *)&c[2*Zc]; + __m256i *d256p = (__m256i *)&d[0]; + // if (((block_length-(2*Zc))&31)>0) l1++; + + for (i=0;i0) l2++; + + for (i1=0;i1>(j-macro_segment))&1; + for (i=0;i<((nrows-no_punctured_columns) * Zc-removed_bit);i++) + //for (j=0; j>(j-macro_segment))&1; + } + +#else + AssertFatal(1==0,"Need AVX2 for now\n"); +#endif + + stop_meas(toutput); + return 0; +} + diff --git a/openair1/PHY/CODING/nrLDPC_encoder/ldpc_generate_coefficient.c b/openair1/PHY/CODING/nrLDPC_encoder/ldpc_generate_coefficient.c index 24098cdf38..d5be3e685f 100644 --- a/openair1/PHY/CODING/nrLDPC_encoder/ldpc_generate_coefficient.c +++ b/openair1/PHY/CODING/nrLDPC_encoder/ldpc_generate_coefficient.c @@ -361,7 +361,7 @@ short *choose_generator_matrix(short BG,short Zc) return Gen_shift_values; } -int ldpc_encoder_orig(unsigned char *test_input,unsigned char *channel_input,short block_length,short BG,unsigned char gen_code) +int ldpc_encoder_orig(unsigned char *test_input,unsigned char *channel_input,short block_length, short BG,unsigned char gen_code) { unsigned char c[22*384]; //padded input, unpacked, max size unsigned char d[68*384]; //coded output, unpacked, max size @@ -384,20 +384,30 @@ int ldpc_encoder_orig(unsigned char *test_input,unsigned char *channel_input,sho int indlist2[1000]; //determine number of bits in codeword - if (BG==2) - { - nrows=42; //parity check bits - ncols=10; // info bits + //if (block_length>3840) + if (BG==1) + { + //BG=1; + Kb = 22; + nrows=46; //parity check bits + ncols=22; //info bits + } + //else if (block_length<=3840) + else if (BG==2) + { + //BG=2; + nrows=42; //parity check bits + ncols=10; // info bits - if (block_length>640) - Kb = 10; - else if (block_length>560) - Kb = 9; - else if (block_length>192) - Kb = 8; - else - Kb = 6; - } + if (block_length>640) + Kb = 10; + else if (block_length>560) + Kb = 9; + else if (block_length>192) + Kb = 8; + else + Kb = 6; + } //find minimum value in all sets of lifting size Zc=0; @@ -415,8 +425,6 @@ int ldpc_encoder_orig(unsigned char *test_input,unsigned char *channel_input,sho return(-1); } - //int K = ncols*Zc; //unused variable - Gen_shift_values=choose_generator_matrix(BG,Zc); if (Gen_shift_values==NULL) { printf("ldpc_encoder_orig: could not find generator matrix\n"); @@ -438,15 +446,9 @@ int ldpc_encoder_orig(unsigned char *test_input,unsigned char *channel_input,sho } no_punctured_columns=(int)((nrows-2)*Zc+block_length-block_length*3)/Zc; - //nrows - no_punctured_columns = 2 +2*block_length/Zc removed_bit=(nrows-no_punctured_columns-2) * Zc+block_length-(block_length*3); - // ((nrows-no_punctured_columns) * Zc-removed_bit) = - // 2Zc + 2*block_length //printf("%d\n",no_punctured_columns); //printf("%d\n",removed_bit); - //printf("%d\n",nrows-no_punctured_columns); - //printf("%d\n",((nrows-no_punctured_columns) * Zc-removed_bit)); - // unpack input memset(c,0,sizeof(unsigned char) * ncols * Zc); memset(d,0,sizeof(unsigned char) * nrows * Zc); @@ -614,8 +616,8 @@ int ldpc_encoder_orig(unsigned char *test_input,unsigned char *channel_input,sho } // information part and puncture columns - memcpy(&channel_input[0], &c[2*Zc], (block_length-2*Zc)*sizeof(unsigned char)); //systematic bits - memcpy(&channel_input[block_length-2*Zc], &d[0], ((nrows-no_punctured_columns) * Zc-removed_bit)*sizeof(unsigned char)); //systematic bits 2Zc + 2*block_length + memcpy(&channel_input[0], &c[2*Zc], (block_length-2*Zc)*sizeof(unsigned char)); + memcpy(&channel_input[block_length-2*Zc], &d[0], ((nrows-no_punctured_columns) * Zc-removed_bit)*sizeof(unsigned char)); //memcpy(channel_input,c,Kb*Zc*sizeof(unsigned char)); return 0; } diff --git a/openair1/PHY/CODING/nrPolar_tools/nr_polar_uci_defs.h b/openair1/PHY/CODING/nrPolar_tools/nr_polar_uci_defs.h index 724da73fa0..42051eb1b5 100644 --- a/openair1/PHY/CODING/nrPolar_tools/nr_polar_uci_defs.h +++ b/openair1/PHY/CODING/nrPolar_tools/nr_polar_uci_defs.h @@ -35,5 +35,27 @@ #define NR_POLAR_UCI_MESSAGE_TYPE -1 //int8_t #define NR_POLAR_UCI_CRC_ERROR_CORRECTION_BITS 3 +#define NR_POLAR_PUCCH_PAYLOAD_BITS 32 +#define NR_POLAR_PUCCH_E 32 + +//Ref. 38-212, Section 6.3.1.2.1 +#define NR_POLAR_PUCCH_CRC_PARITY_BITS_SHORT 6 +#define NR_POLAR_PUCCH_CRC_PARITY_BITS_LONG 11 + +#define NR_POLAR_PUCCH_I_SEG_LONG 1 +#define NR_POLAR_PUCCH_I_SEG_SHORT 0 + +//Ref. 38-212, Section 6.3.1.3.1 +#define NR_POLAR_PUCCH_N_MAX 10 +#define NR_POLAR_PUCCH_I_IL 0 +#define NR_POLAR_PUCCH_N_PC_SHORT 3 +#define NR_POLAR_PUCCH_N_PC_LONG 0 +#define NR_POLAR_PUCCH_N_PC_WM_LONG 0 +#define NR_POLAR_PUCCH_N_PC_WM_SHORT 1 + +//Ref. 38-212, Section 6.3.1.4.1 +#define NR_POLAR_PUCCH_I_BIL 1 + + #endif diff --git a/openair1/PHY/CODING/nr_compute_tbs.c b/openair1/PHY/CODING/nr_compute_tbs.c index 3bfe56e80f..e1c15d4a86 100644 --- a/openair1/PHY/CODING/nr_compute_tbs.c +++ b/openair1/PHY/CODING/nr_compute_tbs.c @@ -28,35 +28,8 @@ #define INDEX_MAX_TBS_TABLE (93) //Table 5.1.3.1-1 -uint16_t Mcsindextable1[29][3] = {{2,120,0.2344}, - {2,157,0.3066}, - {2,193,0.3770}, - {2,251,0.4902}, - {2,308,0.6016}, - {2,379,0.7402}, - {2,449,0.8770}, - {2,526,1.0273}, - {2,602,1.1758}, - {2,679,1.3262}, - {4,340,1.3281}, - {4,378,1.4766}, - {4,434,1.6953}, - {4,490,1.9141}, - {4,553,2.1602}, - {4,616,2.4063}, - {4,658,2.5703}, - {6,438,2.5664}, - {6,466,2.7305}, - {6,517,3.0293}, - {6,567,3.3223}, - {6,616,3.6094}, - {6,666,3.9023}, - {6,719,4.2129}, - {6,772,4.5234}, - {6,822,4.8164}, - {6,873,5.1152}, - {6,910,5.3320}, - {6,948,5.5547}}; +uint16_t Mcsindextable1[29][2] = {{2,120},{2,157},{2,193},{2,251},{2,308},{2,379},{2,449},{2,526},{2,602},{2,679},{4,340},{4,378},{4,434},{4,490},{4,553},{4,616}, + {4,658},{6,438},{6,466},{6,517},{6,567},{6,616},{6,666},{6,719},{6,772},{6,822},{6,873}, {6,910}, {6,948}}; //Table 5.1.2.2-2 uint16_t Tbstable_nr[INDEX_MAX_TBS_TABLE] = {24,32,40,48,56,64,72,80,88,96,104,112,120,128,136,144,152,160,168,176,184,192,208,224,240,256,272,288,304,320,336,352,368,384,408,432,456,480,504,528,552,576,608,640,672,704,736,768,808,848,888,928,984,1032,1064,1128,1160,1192,1224,1256,1288,1320,1352,1416,1480,1544,1608,1672,1736,1800,1864,1928,2024,2088,2152,2216,2280,2408,2472,2536,2600,2664,2728,2792,2856,2976,3104,3240,3368,3496,3624,3752,3824}; @@ -67,9 +40,9 @@ uint32_t nr_compute_tbs(uint8_t mcs, uint16_t length_dmrs, uint8_t Nl) { - uint16_t nbp_re, nb_re, nb_dmrs_prb, nb_rb_oh, Ninfo,Np_info,n,Qm,R,C; - - uint32_t nr_tbs = 0;//Initialization to remove [-Wmaybe-uninitialized] + uint16_t nbp_re, nb_re, nb_dmrs_prb, nb_rb_oh,Qm,R; + uint32_t nr_tbs=0; + double Ninfo,Np_info,n,C; nb_rb_oh = 0; //set to 0 if not configured by higher layer Qm = Mcsindextable1[mcs][0]; @@ -81,17 +54,17 @@ uint32_t nr_compute_tbs(uint8_t mcs, nb_re = min(156, nbp_re) * nb_rb; // Intermediate number of information bits - Ninfo = (nb_re * R * Qm * Nl)/1024; + Ninfo = (double)((nb_re * R * Qm * Nl)/1024); - //printf("Ninfo %d nbp_re %d nb_re %d Qm %d, R %d\n", Ninfo, nbp_re, nb_re, Qm, R); + //printf("Ninfo %lf nbp_re %d nb_re %d mcs %d Qm %d, R %d\n", Ninfo, nbp_re, nb_re,mcs, Qm, R); if (Ninfo <=3824) { n = max(3, floor(log2(Ninfo)) - 6); - Np_info = max(24, pow(n,2) * floor(Ninfo/pow(n,2))); + Np_info = max(24, pow(2,n) * floor(Ninfo/pow(2,n))); for (int i=0; i= Np_info){ + if ((double)Tbstable_nr[i] >= Np_info){ nr_tbs = Tbstable_nr[i]; break; } @@ -99,19 +72,21 @@ uint32_t nr_compute_tbs(uint8_t mcs, } else { n = floor(log2(Ninfo-24)) - 5; - Np_info = max(3840, pow(n,2) * round((Ninfo - 24)/pow(n,2))); + Np_info = max(3840, pow(2,n) * round((Ninfo - 24)/pow(2,n))); - if (R <= 1024/4) { + if (R <= 256) { //1/4 C = ceil( (Np_info + 24)/3816 ); - nr_tbs = 8 * C * ceil( (Np_info + 24)/(8*C) ) - 24; + nr_tbs = (uint32_t)(8 * C * ceil( (Np_info + 24)/(8*C) ) - 24); } else { if (Np_info > 8424){ C = ceil( (Np_info + 24)/8424 ); - nr_tbs = 8 * C * ceil( (Np_info + 24)/(8*C) ) - 24; + nr_tbs = (uint32_t)(8 * C * ceil( (Np_info + 24)/(8*C) ) - 24); + } + else { + nr_tbs = (uint32_t)(8 * ceil( (Np_info + 24)/8 ) - 24); + //printf("n %lf Np_info %f pow %f ceil %f \n",n, Np_info,pow(2,6),ceil( (Np_info + 24)/8 )); } - else - nr_tbs = 8 * ceil( (Np_info + 24)/8 ) - 24; } diff --git a/openair1/PHY/CODING/nr_polar_init.c b/openair1/PHY/CODING/nr_polar_init.c index e808175930..76dd5c4ad0 100644 --- a/openair1/PHY/CODING/nr_polar_init.c +++ b/openair1/PHY/CODING/nr_polar_init.c @@ -51,7 +51,7 @@ void nr_polar_init(t_nrPolar_paramsPtr *polarParams, //Parse the list. If the node is already created, return without initialization. while (currentPtr != NULL) { - // printf("currentPtr->idx %d, (%d,%d,%d)\n",currentPtr->idx,messageType,messageLength,aggregation_prime); + printf("currentPtr->idx %d, (%d,%d)\n",currentPtr->idx,currentPtr->payloadBits,currentPtr->encoderLength); if (currentPtr->idx == (messageType * messageLength * aggregation_prime)) return; else currentPtr = currentPtr->nextPtr; } @@ -64,6 +64,7 @@ void nr_polar_init(t_nrPolar_paramsPtr *polarParams, newPolarInitNode->idx = (messageType * messageLength * aggregation_prime); newPolarInitNode->nextPtr = NULL; + printf("newPolarInitNode->idx %d, (%d,%d,%d:%d)\n",newPolarInitNode->idx,messageType,messageLength,aggregation_prime,aggregation_level); if (messageType == 0) { //PBCH newPolarInitNode->n_max = NR_POLAR_PBCH_N_MAX; diff --git a/openair1/PHY/CODING/nr_rate_matching.c b/openair1/PHY/CODING/nr_rate_matching.c new file mode 100644 index 0000000000..245f3b9b75 --- /dev/null +++ b/openair1/PHY/CODING/nr_rate_matching.c @@ -0,0 +1,241 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.1 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +/* file: nr_rate_matching.c + purpose: Procedures for rate matching/interleaving for NR LDPC + author: hongzhi.wang@tcl.com +*/ + +#include "PHY/defs_gNB.h" +#include "PHY/defs_nr_UE.h" + +//#define RM_DEBUG 1 + +uint8_t index_k0[2][4] = {{0,17,33,56},{0,13,25,43}}; + + +void nr_interleaving_ldpc(uint32_t E, uint8_t Qm, uint8_t *e,uint8_t *f) +{ + uint32_t EQm; + + EQm = E/Qm; + memset(f,0,E*sizeof(uint8_t)); + + for (int j = 0; j< EQm; j++){ + for (int i = 0; i< Qm; i++){ + f[(i+j*Qm)] = e[(i*EQm + j)]; + } + } + +} + + +void nr_deinterleaving_ldpc(uint32_t E, uint8_t Qm, int16_t *e,int16_t *f) +{ + + uint32_t EQm; + + EQm = E/Qm; + + for (int j = 0; j< EQm; j++){ + for (int i = 0; i< Qm; i++){ + e[(i*EQm + j)] = f[(i+j*Qm)]; + } + } +} + + +uint32_t nr_rate_matching_ldpc(uint8_t Ilbrm, + uint32_t Tbslbrm, + uint8_t BG, + uint16_t Z, + uint32_t G, + uint8_t *w, + uint8_t *e, + uint8_t C, + uint8_t rvidx, + uint8_t Qm, + uint8_t Nl, + uint8_t r) +{ + uint8_t Cprime; + uint32_t Ncb,E,ind,k,Nref,N; + uint8_t *e2; + + AssertFatal(Nl>0,"Nl is 0\n"); + AssertFatal(Qm>0,"Qm is 0\n"); + + //Bit selection + N = (BG==1)?(66*Z):(50*Z); + + if (Ilbrm == 0) + Ncb = N; + else { + Nref = 3*Tbslbrm/(2*C); //R_LBRM = 2/3 + Ncb = min(N, Nref); + } + +#ifdef RM_DEBUG + printf("nr_rate_matching: Ncb %d, rvidx %d, G %d, Qm %d, Nl%d, r %d\n",Ncb,rvidx, G, Qm,Nl,r); +#endif + + Cprime = C; //assume CBGTI not present + + if (r <= Cprime - ((G/(Nl*Qm))%Cprime) - 1) + E = Nl*Qm*(G/(Nl*Qm*Cprime)); + else + E = Nl*Qm*((G/(Nl*Qm*Cprime))+1); + + ind = (index_k0[BG-1][rvidx]*Ncb/N)*Z; + +#ifdef RM_DEBUG + printf("nr_rate_matching: E %d, k0 %d Cprime %d modcprime %d\n",E,ind, Cprime,((G/(Nl*Qm))%Cprime)); +#endif + + e2 = e; + + k=0; + + for (; (ind0,"Nl is 0\n"); + AssertFatal(Qm>0,"Qm is 0\n"); + + //Bit selection + N = (BG==1)?(66*Z):(50*Z); + + if (Ilbrm == 0) + Ncb = N; + else { + Nref = (3*Tbslbrm/(2*C)); //R_LBRM = 2/3 + Ncb = min(N, Nref); + } + + Cprime = C; //assume CBGTI not present + + if (r <= Cprime - ((G/(Nl*Qm))%Cprime) - 1) + E = Nl*Qm*(G/(Nl*Qm*Cprime)); + else + E = Nl*Qm*((G/(Nl*Qm*Cprime))+1); + + ind = (index_k0[BG-1][rvidx]*Ncb/N)*Z; + +#ifdef RM_DEBUG + printf("nr_rate_matching_ldpc_rx: Clear %d, E %d, Ncb %d,rvidx %d, G %d, Qm %d, Nl%d, r %d\n",clear,E,Ncb,rvidx, G, Qm,Nl,r); +#endif + + if (clear==1) + memset(w,0,Ncb*sizeof(int16_t)); + + soft_input2 = soft_input; + k=0; + + for (; (indMAX_NUM_DLSCH_SEGMENTS) { - LOG_E(PHY,"lte_segmentation.c: too many segments %d, B %d, L %d, Bprime %d\n",*C,B,L,Bprime); + LOG_E(PHY,"nr_segmentation.c: too many segments %d, B %d, L %d, Bprime %d\n",*C,B,L,Bprime); return(-1); } // Find K+ Bprime_by_C = Bprime/(*C); - /*if (Bprime <=192) { + if (Bprime <=192) { Kb = 6; } else if (Bprime <=560) { Kb = 8; @@ -74,9 +72,9 @@ int32_t nr_segmentation(unsigned char *input_buffer, Kb = 9; } else if (Bprime <=3840) { Kb = 10;; - } else {*/ + } else { Kb = 22; - //} + } if ((Bprime_by_C%Kb) > 0) @@ -89,97 +87,85 @@ else #endif if (Z <= 2) { - *Kplus = 2; - *Kminus = 0; + *K = 2; } else if (Z<=16) { // increase by 1 byte til here - *Kplus = Z; - *Kminus = Z-1; + *K = Z; } else if (Z <=32) { // increase by 2 bytes til here - *Kplus = (Z>>1)<<1; + *K = (Z>>1)<<1; - if (*Kplus < Z) - *Kplus = *Kplus + 2; + if (*K < Z) + *K = *K + 2; - *Kminus = (*Kplus - 2); } else if (Z <= 64) { // increase by 4 bytes til here - *Kplus = (Z>>2)<<2; + *K = (Z>>2)<<2; - if (*Kplus < Z) - *Kplus = *Kplus + 4; + if (*K < Z) + *K = *K + 4; - *Kminus = (*Kplus - 4); } else if (Z <=128 ) { // increase by 8 bytes til here - *Kplus = (Z>>3)<<3; + *K = (Z>>3)<<3; - if (*Kplus < Z) - *Kplus = *Kplus + 8; + if (*K < Z) + *K = *K + 8; #ifdef DEBUG_SEGMENTATION - printf("Z_by_C %d , Kplus2 %d\n",Z,*Kplus); + printf("Z_by_C %d , K2 %d\n",Z,*K); #endif - *Kminus = (*Kplus - 8); } else if (Z <= 256) { // increase by 4 bytes til here - *Kplus = (Z>>4)<<4; + *K = (Z>>4)<<4; - if (*Kplus < Z) - *Kplus = *Kplus + 16; + if (*K < Z) + *K = *K + 16; - *Kminus = (*Kplus - 16); } else if (Z <= 384) { // increase by 4 bytes til here - *Kplus = (Z>>5)<<5; + *K = (Z>>5)<<5; - if (*Kplus < Z) - *Kplus = *Kplus + 32; + if (*K < Z) + *K = *K + 32; - *Kminus = (*Kplus - 32); } else { //msg("nr_segmentation.c: Illegal codeword size !!!\n"); return(-1); } - *Zout = *Kplus; - *Kplus = *Kplus*Kb; - *Kminus = *Kminus*Kb; + *Zout = *K; + *K = *K*Kb; - - - *F = ((*C)*(*Kplus) - (Bprime)); + *F = ((*K) - Bprime_by_C); #ifdef DEBUG_SEGMENTATION - printf("final nr seg output Z %d Kplus %d F %d \n", *Zout, *Kplus, *F); - printf("C %d, Kplus %d, Kminus %d, Bprime_bytes %d, Bprime %d, F %d\n",*C,*Kplus,*Kminus,Bprime>>3,Bprime,*F); + printf("final nr seg output Z %d K %d F %d \n", *Zout, *K, *F); + printf("C %d, K %d, Bprime_bytes %d, Bprime %d, F %d\n",*C,*K,Bprime>>3,Bprime,*F); #endif if ((input_buffer) && (output_buffers)) { - for (k=0; k<*F>>3; k++) { - output_buffers[0][k] = 0; - } - s=0; for (r=0; r<*C; r++) { - //if (r<(B%(*C))) - Kprime = *Kplus; - //else - // Kprime = *Kminus; + Kprime = Bprime_by_C; while (k<((Kprime - L)>>3)) { output_buffers[r][k] = input_buffer[s]; - // printf("encoding segment %d : byte %d (%d) => %d\n",r,k,Kr>>3,input_buffer[s]); + //printf("encoding segment %d : byte %d (%d) => %d\n",r,k,(Kprime-L)>>3,input_buffer[s]); k++; s++; } - /* - if (*C > 1) { // add CRC - crc = crc24b(output_buffers[r],Kprime-24)>>8; - output_buffers[r][(Kprime-24)>>3] = ((uint8_t*)&crc)[2]; - output_buffers[r][1+((Kprime-24)>>3)] = ((uint8_t*)&crc)[1]; - output_buffers[r][2+((Kprime-24)>>3)] = ((uint8_t*)&crc)[0]; + if (*F>0) + for (k=Kprime>>3; k<(*K)>>3; k++) { + output_buffers[r][k] = NR_NULL; + //printf("r %d filler bits [%d] = %d Kprime %d \n", r,k, output_buffers[r][k], Kprime); } - */ + + if (*C > 1) { // add CRC + crc = crc24b(output_buffers[r],Kprime-L)>>8; + output_buffers[r][(Kprime-L)>>3] = ((uint8_t*)&crc)[2]; + output_buffers[r][1+((Kprime-L)>>3)] = ((uint8_t*)&crc)[1]; + output_buffers[r][2+((Kprime-L)>>3)] = ((uint8_t*)&crc)[0]; + } + k=0; } } @@ -193,12 +179,12 @@ else main() { - unsigned int Kplus,Kminus,C,F,Bbytes; + unsigned int K,C,F,Bbytes, Zout; for (Bbytes=5; Bbytes<8; Bbytes++) { - nr_segmentation(0,0,Bbytes<<3,&C,&Kplus,&Kminus, &F); - printf("Bbytes %d : C %d, Kplus %d, Kminus %d, F %d\n", - Bbytes, C, Kplus, Kminus, F); + nr_segmentation(0,0,Bbytes<<3,&C,&K,&Zout, &F); + printf("Bbytes %d : C %d, K %d, F %d\n", + Bbytes, C, K, F); } } #endif diff --git a/openair1/PHY/INIT/nr_init.c b/openair1/PHY/INIT/nr_init.c index b69844cd52..a4ac55904e 100644 --- a/openair1/PHY/INIT/nr_init.c +++ b/openair1/PHY/INIT/nr_init.c @@ -24,6 +24,7 @@ #include "PHY/NR_REFSIG/nr_refsig.h" #include "PHY/INIT/phy_init.h" #include "PHY/CODING/nrPolar_tools/nr_polar_pbch_defs.h" +#include "PHY/NR_TRANSPORT/nr_transport.h" #include "RadioResourceConfigCommonSIB.h" #include "RadioResourceConfigDedicated.h" #include "TDD-Config.h" @@ -146,6 +147,27 @@ int phy_init_nr_gNB(PHY_VARS_gNB *gNB, nr_init_pdcch_dmrs(gNB, cfg->sch_config.physical_cell_id.value); nr_init_pbch_interleaver(gNB->nr_pbch_interleaver); + + //PDSCH DMRS init + gNB->nr_gold_pdsch_dmrs = (uint32_t ****)malloc16(fp->slots_per_frame*sizeof(uint32_t***)); + uint32_t ****pdsch_dmrs = gNB->nr_gold_pdsch_dmrs; + for (int slot=0; slotslots_per_frame; slot++) { + pdsch_dmrs[slot] = (uint32_t ***)malloc16(fp->symbols_per_slot*sizeof(uint32_t**)); + AssertFatal(pdsch_dmrs[slot]!=NULL, "NR init: pdsch_dmrs for slot %d - malloc failed\n", slot); + for (int symb=0; symbsymbols_per_slot; symb++){ + pdsch_dmrs[slot][symb] = (uint32_t **)malloc16(NR_MAX_NB_CODEWORDS*sizeof(uint32_t*)); + AssertFatal(pdsch_dmrs[slot][symb]!=NULL, "NR init: pdsch_dmrs for slot %d symbol %d - malloc failed\n", slot, symb); + for (int q=0; qsch_config.physical_cell_id.value); + /// Transport init necessary for NR synchro + init_nr_transport(gNB); + /* lte_gold(fp,gNB->lte_gold_table,fp->Nid_cell); generate_pcfich_reg_mapping(fp); @@ -451,3 +473,69 @@ void nr_phy_config_request(NR_PHY_Config_t *phy_config) } + +void init_nr_transport(PHY_VARS_gNB *gNB) { + + int i; + int j; + NR_DL_FRAME_PARMS *fp = &gNB->frame_parms; + nfapi_nr_config_request_t *cfg = &gNB->gNB_config; + + LOG_I(PHY, "Initialise nr transport\n"); + + for (i=0; idlsch[i][j] = new_gNB_dlsch(1,16,NSOFT,0,fp,cfg); + if (!gNB->dlsch[i][j]) { + LOG_E(PHY,"Can't get gNB dlsch structures for UE %d \n", i); + exit(-1); + }/* else { + gNB->dlsch[i][j]->rnti=0; + LOG_D(PHY,"dlsch[%d][%d] => %p rnti:%d\n",i,j,gNB->dlsch[i][j], gNB->dlsch[i][j]->rnti); + }*/ + } + + //LOG_I(PHY,"Allocating Transport Channel Buffer for ULSCH, UE %d\n",i); + //gNB->ulsch[1+i] = new_gNB_ulsch(MAX_TURBO_ITERATIONS,fp->N_RB_UL, 0); + + /*if (!gNB->ulsch[1+i]) { + LOG_E(PHY,"Can't get gNB ulsch structures\n"); + exit(-1); + }*/ + + // this is the transmission mode for the signalling channels + // this will be overwritten with the real transmission mode by the RRC once the UE is connected + //gNB->transmission_mode[i] = fp->nb_antenna_ports_gNB==1 ? 1 : 2; + } + // ULSCH for RA + //gNB->ulsch[0] = new_gNB_ulsch(MAX_TURBO_ITERATIONS, fp->N_RB_UL, 0); + + /*if (!gNB->ulsch[0]) { + LOG_E(PHY,"Can't get gNB ulsch structures\n"); + exit(-1); + }*/ + gNB->dlsch_SI = new_gNB_dlsch(1,8,NSOFT, 0, fp, cfg); + LOG_D(PHY,"gNB %d.%d : SI %p\n",gNB->Mod_id,gNB->CC_id,gNB->dlsch_SI); + gNB->dlsch_ra = new_gNB_dlsch(1,8,NSOFT, 0, fp, cfg); + LOG_D(PHY,"gNB %d.%d : RA %p\n",gNB->Mod_id,gNB->CC_id,gNB->dlsch_ra); + gNB->dlsch_MCH = new_gNB_dlsch(1,8,NSOFT, 0, fp, cfg); + LOG_D(PHY,"gNB %d.%d : MCH %p\n",gNB->Mod_id,gNB->CC_id,gNB->dlsch_MCH); + + + gNB->rx_total_gain_dB=130; + + for(i=0; imu_mimo_mode[i].dl_pow_off = 2; + + gNB->check_for_total_transmissions = 0; + + gNB->check_for_MUMIMO_transmissions = 0; + + gNB->FULL_MUMIMO_transmissions = 0; + + gNB->check_for_SUMIMO_transmissions = 0; + + //fp->pucch_config_common.deltaPUCCH_Shift = 1; + +} diff --git a/openair1/PHY/INIT/nr_init_ue.c b/openair1/PHY/INIT/nr_init_ue.c index a5f0558e8e..17f0d8a5d2 100644 --- a/openair1/PHY/INIT/nr_init_ue.c +++ b/openair1/PHY/INIT/nr_init_ue.c @@ -32,7 +32,7 @@ #include #include "PHY/NR_UE_TRANSPORT/nr_transport_ue.h" #include "PHY/NR_UE_TRANSPORT/nr_transport_proto_ue.h" -#include "PHY/LTE_REFSIG/lte_refsig.h" +//#include "PHY/LTE_REFSIG/lte_refsig.h" #include "PHY/CODING/nrPolar_tools/nr_polar_pbch_defs.h" #include "PHY/INIT/phy_init.h" #include "PHY/NR_REFSIG/pss_nr.h" @@ -816,10 +816,10 @@ int init_nr_ue_signal(PHY_VARS_NR_UE *ue, // 100 PRBs * 12 REs/PRB * 4 PDCCH SYMBOLS * 2 LLRs/RE for (th_id=0; th_idllr = (uint16_t*)malloc16_clear( 2*4*100*12*sizeof(uint16_t) ); - (*pdcch_vars_th)[th_id][eNB_id]->llr16 = (uint16_t*)malloc16_clear( 2*4*100*12*sizeof(uint16_t) ); - (*pdcch_vars_th)[th_id][eNB_id]->wbar = (uint16_t*)malloc16_clear( 2*4*100*12*sizeof(uint16_t) ); - (*pdcch_vars_th)[th_id][eNB_id]->e_rx = (int8_t*)malloc16_clear( 4*2*100*12 ); + (*pdcch_vars_th)[th_id][eNB_id]->llr = (int16_t*)malloc16_clear( 2*4*100*12*sizeof(uint16_t) ); + (*pdcch_vars_th)[th_id][eNB_id]->llr16 = (int16_t*)malloc16_clear( 2*4*100*12*sizeof(uint16_t) ); + (*pdcch_vars_th)[th_id][eNB_id]->wbar = (int16_t*)malloc16_clear( 2*4*100*12*sizeof(uint16_t) ); + (*pdcch_vars_th)[th_id][eNB_id]->e_rx = (int16_t*)malloc16_clear( 4*2*100*12 ); (*pdcch_vars_th)[th_id][eNB_id]->rxdataF_comp = (int32_t**)malloc16_clear( 8*sizeof(int32_t*) ); (*pdcch_vars_th)[th_id][eNB_id]->dl_ch_rho_ext = (int32_t**)malloc16_clear( 8*sizeof(int32_t*) ); @@ -912,7 +912,6 @@ int init_nr_ue_signal(PHY_VARS_NR_UE *ue, } -/* void init_nr_ue_transport(PHY_VARS_NR_UE *ue,int abstraction_flag) { int i,j,k; @@ -920,25 +919,25 @@ void init_nr_ue_transport(PHY_VARS_NR_UE *ue,int abstraction_flag) { for (i=0; idlsch[k][i][j] = new_ue_dlsch(1,NUMBER_OF_HARQ_PID_MAX,NSOFT,MAX_TURBO_ITERATIONS,ue->frame_parms.N_RB_DL, abstraction_flag))!=NULL,"Can't get ue dlsch structures\n"); + AssertFatal((ue->dlsch[k][i][j] = new_nr_ue_dlsch(1,NUMBER_OF_HARQ_PID_MAX,NSOFT,MAX_LDPC_ITERATIONS,ue->frame_parms.N_RB_DL, abstraction_flag))!=NULL,"Can't get ue dlsch structures\n"); LOG_D(PHY,"dlsch[%d][%d][%d] => %p\n",k,i,j,ue->dlsch[i][j]); } } - AssertFatal((ue->ulsch[i] = new_ue_ulsch(ue->frame_parms.N_RB_UL, abstraction_flag))!=NULL,"Can't get ue ulsch structures\n"); + //AssertFatal((ue->ulsch[i] = new_ue_ulsch(ue->frame_parms.N_RB_UL, abstraction_flag))!=NULL,"Can't get ue ulsch structures\n"); - ue->dlsch_SI[i] = new_ue_dlsch(1,1,NSOFT,MAX_TURBO_ITERATIONS,ue->frame_parms.N_RB_DL, abstraction_flag); - ue->dlsch_ra[i] = new_ue_dlsch(1,1,NSOFT,MAX_TURBO_ITERATIONS,ue->frame_parms.N_RB_DL, abstraction_flag); + ue->dlsch_SI[i] = new_nr_ue_dlsch(1,1,NSOFT,MAX_LDPC_ITERATIONS,ue->frame_parms.N_RB_DL, abstraction_flag); + ue->dlsch_ra[i] = new_nr_ue_dlsch(1,1,NSOFT,MAX_LDPC_ITERATIONS,ue->frame_parms.N_RB_DL, abstraction_flag); ue->transmission_mode[i] = ue->frame_parms.nb_antenna_ports_eNB==1 ? 1 : 2; } //ue->frame_parms.pucch_config_common.deltaPUCCH_Shift = 1; - ue->dlsch_MCH[0] = new_ue_dlsch(1,NUMBER_OF_HARQ_PID_MAX,NSOFT,MAX_TURBO_ITERATIONS_MBSFN,ue->frame_parms.N_RB_DL,0); + ue->dlsch_MCH[0] = new_nr_ue_dlsch(1,NUMBER_OF_HARQ_PID_MAX,NSOFT,MAX_LDPC_ITERATIONS_MBSFN,ue->frame_parms.N_RB_DL,0); -}*/ +} void phy_init_nr_top(PHY_VARS_NR_UE *ue) { diff --git a/openair1/PHY/INIT/phy_init.h b/openair1/PHY/INIT/phy_init.h index 8e579c69be..fdb7fab1b7 100644 --- a/openair1/PHY/INIT/phy_init.h +++ b/openair1/PHY/INIT/phy_init.h @@ -385,6 +385,7 @@ void nr_phy_config_request(NR_PHY_Config_t *gNB); void nr_phy_config_request_sim(PHY_VARS_gNB *gNB,int N_RB_DL,int N_RB_UL,int mu); void phy_free_nr_gNB(PHY_VARS_gNB *gNB); int l1_north_init_gNB(void); +void init_nr_transport(PHY_VARS_gNB *gNB); void init_dfts(void); diff --git a/openair1/PHY/LTE_TRANSPORT/transport_vars.h b/openair1/PHY/LTE_TRANSPORT/transport_vars.h index 02dfbc6a5e..6ac121dc13 100644 --- a/openair1/PHY/LTE_TRANSPORT/transport_vars.h +++ b/openair1/PHY/LTE_TRANSPORT/transport_vars.h @@ -58,9 +58,6 @@ char lte_cqi_snr_dB[15] = { -2, unsigned char ue_power_offsets[25] = {14,11,9,8,7,6,6,5,4,4,4,3,3,3,2,2,2,1,1,1,1,1,0,0,0}; -short conjugate[8]__attribute__((aligned(16))) = {-1,1,-1,1,-1,1,-1,1}; -short conjugate2[8]__attribute__((aligned(16))) = {1,-1,1,-1,1,-1,1,-1}; - int qam64_table[8],qam16_table[4],qpsk_table[2]; unsigned char cs_ri_normal[4] = {1,4,7,10}; diff --git a/openair1/PHY/MODULATION/slot_fep_nr.c b/openair1/PHY/MODULATION/slot_fep_nr.c index 330817bfc9..8ef29d016d 100644 --- a/openair1/PHY/MODULATION/slot_fep_nr.c +++ b/openair1/PHY/MODULATION/slot_fep_nr.c @@ -49,7 +49,7 @@ int nr_slot_fep(PHY_VARS_NR_UE *ue, unsigned char symbol = l;//+((7-frame_parms->Ncp)*(Ns&1)); ///symbol within sub-frame unsigned int nb_prefix_samples = (no_prefix ? 0 : frame_parms->nb_prefix_samples); unsigned int nb_prefix_samples0 = (no_prefix ? 0 : frame_parms->nb_prefix_samples0); - unsigned int subframe_offset;//,subframe_offset_F; + //unsigned int subframe_offset;//,subframe_offset_F; unsigned int slot_offset; //int i; unsigned int frame_length_samples = frame_parms->samples_per_subframe * 10; @@ -57,13 +57,10 @@ int nr_slot_fep(PHY_VARS_NR_UE *ue, NR_UE_PDCCH *pdcch_vars = ue->pdcch_vars[ue->current_thread_id[Ns>>1]][0]; uint16_t coreset_start_subcarrier = frame_parms->first_carrier_offset+((int)floor(frame_parms->ssb_start_subcarrier/NR_NB_SC_PER_RB)+pdcch_vars->coreset[0].rb_offset)*NR_NB_SC_PER_RB; uint16_t nb_rb_coreset = 24; - uint16_t bwp_start_subcarrier = frame_parms->first_carrier_offset; - uint16_t nb_rb_pdsch = 100; - - /*LTE_UE_DLSCH_t **dlsch_ue = phy_vars_ue->dlsch_ue[eNB_id]; - unsigned char harq_pid = dlsch_ue[0]->current_harq_pid; - LTE_DL_UE_HARQ_t *dlsch0_harq = dlsch_ue[0]->harq_processes[harq_pid]; - int uespec_pilot[9][1200];*/ + uint16_t bwp_start_subcarrier = frame_parms->first_carrier_offset+516; + uint16_t nb_rb_pdsch = 50; + uint8_t p=0; + uint8_t l0 = 2; void (*dft)(int16_t *,int16_t *, int); int tmp_dft_in[8192] __attribute__ ((aligned (32))); // This is for misalignment issues for 6 and 15 PRBs @@ -107,11 +104,9 @@ int nr_slot_fep(PHY_VARS_NR_UE *ue, } if (no_prefix) { - subframe_offset = frame_parms->ofdm_symbol_size * frame_parms->symbols_per_tti * (Ns>>1); - slot_offset = frame_parms->ofdm_symbol_size * (frame_parms->symbols_per_tti>>1) * (Ns%2); + slot_offset = frame_parms->ofdm_symbol_size * (frame_parms->symbols_per_slot*frame_parms->slots_per_subframe) * (Ns>>1); } else { - subframe_offset = frame_parms->samples_per_tti * (Ns>>1); - slot_offset = (frame_parms->samples_per_tti>>1) * (Ns%2); + slot_offset = (frame_parms->samples_per_subframe) * (Ns>>1); } /*if (l<0 || l>=7-frame_parms->Ncp) { @@ -129,7 +124,7 @@ int nr_slot_fep(PHY_VARS_NR_UE *ue, for (aa=0; aanb_antennas_rx; aa++) { memset(&common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[Ns>>1]].rxdataF[aa][frame_parms->ofdm_symbol_size*symbol],0,frame_parms->ofdm_symbol_size*sizeof(int)); - rx_offset = sample_offset + slot_offset + nb_prefix_samples0 + subframe_offset - SOFFSET; + rx_offset = sample_offset + slot_offset + nb_prefix_samples0 - SOFFSET; // Align with 256 bit // rx_offset = rx_offset&0xfffffff8; @@ -167,7 +162,6 @@ int nr_slot_fep(PHY_VARS_NR_UE *ue, rx_offset += (frame_parms->ofdm_symbol_size+nb_prefix_samples)*l;// + // (frame_parms->ofdm_symbol_size+nb_prefix_samples)*(l-1); - if (rx_offset > (frame_length_samples - frame_parms->ofdm_symbol_size)) memcpy((void *)&common_vars->rxdata[aa][frame_length_samples], (void *)&common_vars->rxdata[aa][0], @@ -196,7 +190,7 @@ int nr_slot_fep(PHY_VARS_NR_UE *ue, #ifdef DEBUG_FEP // if (ue->frame <100) - printf("slot_fep: frame %d: symbol %d rx_offset %d\n", ue->proc.proc_rxtx[(Ns>>1)&1].frame_rx, symbol,rx_offset); + printf("slot_fep: frame %d: symbol %d rx_offset %d\n", ue->proc.proc_rxtx[(Ns)&1].frame_rx, symbol,rx_offset); #endif } @@ -272,12 +266,18 @@ int nr_slot_fep(PHY_VARS_NR_UE *ue, #if UE_TIMING_TRACE start_meas(&ue->dlsch_channel_estimation_stats); #endif + + ue->frame_parms.nushift = (p>>1)&1;; + + if (symbol ==l0) nr_pdsch_channel_estimation(ue,eNB_id,0, Ns, + p, l, symbol, bwp_start_subcarrier, nb_rb_pdsch); + #if UE_TIMING_TRACE stop_meas(&ue->dlsch_channel_estimation_stats); #endif diff --git a/openair1/PHY/NR_REFSIG/nr_dmrs_rx.c b/openair1/PHY/NR_REFSIG/nr_dmrs_rx.c index 6e14272550..2e5ad47d91 100644 --- a/openair1/PHY/NR_REFSIG/nr_dmrs_rx.c +++ b/openair1/PHY/NR_REFSIG/nr_dmrs_rx.c @@ -30,25 +30,12 @@ * \warning */ -//#define DEBUG_DL_DMRS -//#define DEBUG_PDCCH //#define NR_PBCH_DMRS_LENGTH_DWORD 5 //#define NR_PBCH_DMRS_LENGTH 144 -#ifdef USER_MODE -#include -#include -#endif - #include "refsig_defs_ue.h" #include "PHY/defs_nr_UE.h" -//#include "nr_mod_table.h" -#include "common/utils/LOG/log.h" -#ifdef LOG_I -#undef LOG_I -#define LOG_I(A,B...) printf(B) -#endif /*Table 7.4.1.1.2-1/2 from 38.211 */ int wf1[8][2] = {{1,1},{1,-1},{1,1},{1,-1},{1,1},{1,-1},{1,1},{1,1}}; int wt1[8][2] = {{1,1},{1,1},{1,1},{1,1},{1,-1},{1,-1},{1,-1},{1,-1}}; @@ -56,102 +43,28 @@ int wf2[12][2] = {{1,1},{1,-1},{1,1},{1,-1},{1,1},{1,-1},{1,1},{1,1},{1,1},{1,-1 int wt2[12][2] = {{1,1},{1,1},{1,1},{1,1},{1,1},{1,1},{1,-1},{1,-1},{1,-1},{1,-1},{1,-1},{1,-1}}; -short nr_rx_mod_table[NR_MOD_TABLE_SIZE_SHORT] = {0,0,23170,-23170,-23170,23170,23170,-23170,23170,23170,-23170,-23170,-23170,23170}; +short nr_rx_mod_table[14] = {0,0,23170,-23170,-23170,23170,23170,-23170,23170,23170,-23170,-23170,-23170,23170}; +short nr_rx_nmod_table[14] = {0,0,-23170,23170,23170,-23170,-23170,23170,-23170,-23170,23170,23170,23170,-23170}; -/*int nr_pdcch_dmrs_rx(PHY_VARS_NR_UE *ue, - uint8_t eNB_offset, - unsigned int Ns, - unsigned int nr_gold_pdcch[7][20][3][10], - int32_t *output, - unsigned short p, - int length_dmrs, - unsigned short nb_rb_coreset) - { - int32_t qpsk[4],n; - int w,ind,l,ind_dword,ind_qpsk_symb,kp,k; - short pamp; - pamp = ONE_OVER_SQRT2_Q15; - // This includes complex conjugate for channel estimation - ((short *)&qpsk[0])[0] = pamp; - ((short *)&qpsk[0])[1] = -pamp; - ((short *)&qpsk[1])[0] = -pamp; - ((short *)&qpsk[1])[1] = -pamp; - ((short *)&qpsk[2])[0] = pamp; - ((short *)&qpsk[2])[1] = pamp; - ((short *)&qpsk[3])[0] = -pamp; - ((short *)&qpsk[3])[1] = pamp; - - if (p==2000) { - for (n=0; n>4; - ind_qpsk_symb = ind&0xf; - - output[k] = qpsk[(nr_gold_pdcch[eNB_offset][Ns][l][ind_dword]>>(2*ind_qpsk_symb))&3]; - - #ifdef DEBUG_DL_DMRS - LOG_I(PHY,"Ns %d, p %d, ind_dword %d, ind_qpsk_symbol %d\n", - Ns,p,idx_dword,idx_qpsk_symb); - LOG_I(PHY,"index = %d\n",(nr_gold_pdsch[0][Ns][lprime][ind_dword]>>(2*ind_qpsk_symb))&3); - #endif - - k++; - } - } - } - } else { - LOG_E(PHY,"Illegal PDCCH DMRS port %d\n",p); - } - - return(0); - }*/ int nr_pdsch_dmrs_rx(PHY_VARS_NR_UE *ue, - uint8_t eNB_offset, - unsigned int Ns, - unsigned int nr_gold_pdsch[2][20][2][21], - int32_t *output, - unsigned short p, - int length_dmrs, - unsigned short nb_rb_pdsch) + unsigned int Ns, + unsigned int *nr_gold_pdsch, + int32_t *output, + unsigned short p, + unsigned char lp, + unsigned short nb_pdsch_rb) { - int32_t qpsk[4],nqpsk[4],*qpsk_p, n; - //int w,mprime,ind,l,ind_dword,ind_qpsk_symb,kp,lp, config_type, k; - int w,ind,ind_dword,ind_qpsk_symb,kp,lp, config_type, k=0; - short pamp; + int8_t w,config_type; + short *mod_table; + unsigned char idx=0; typedef int array_of_w[2]; array_of_w *wf; array_of_w *wt; - config_type = 1; - printf("dmrs config type %d port %d\n", config_type, p); - - // Compute the correct pilot amplitude, sqrt_rho_b = Q3.13 - pamp = ONE_OVER_SQRT2_Q15; - - // This includes complex conjugate for channel estimation - ((short *)&qpsk[0])[0] = pamp; - ((short *)&qpsk[0])[1] = -pamp; - ((short *)&qpsk[1])[0] = -pamp; - ((short *)&qpsk[1])[1] = -pamp; - ((short *)&qpsk[2])[0] = pamp; - ((short *)&qpsk[2])[1] = pamp; - ((short *)&qpsk[3])[0] = -pamp; - ((short *)&qpsk[3])[1] = pamp; - - ((short *)&nqpsk[0])[0] = -pamp; - ((short *)&nqpsk[0])[1] = pamp; - ((short *)&nqpsk[1])[0] = pamp; - ((short *)&nqpsk[1])[1] = pamp; - ((short *)&nqpsk[2])[0] = -pamp; - ((short *)&nqpsk[2])[1] = -pamp; - ((short *)&nqpsk[3])[0] = pamp; - ((short *)&nqpsk[3])[1] = -pamp; + config_type = 0; //to be updated by higher layer wf = (config_type==0) ? wf1 : wf2; wt = (config_type==0) ? wt1 : wt2; @@ -160,31 +73,26 @@ int nr_pdsch_dmrs_rx(PHY_VARS_NR_UE *ue, LOG_E(PHY,"Bad PDSCH DMRS config type %d\n", config_type); if ((p>=1000) && (p<((config_type==0) ? 1008 : 1012))) { + if (ue->frame_parms.Ncp == NORMAL) { - // r_n from 38.211 7.4.1.1 - for (n=0; n>4; - ind_qpsk_symb = ind&0xf; + w = (wf[p-1000][i&1])*(wt[p-1000][lp]); + mod_table = (w==1) ? nr_rx_mod_table : nr_rx_nmod_table; - output[k] = qpsk_p[(nr_gold_pdsch[0][Ns][lp][ind_dword]>>(2*ind_qpsk_symb))&3]; - - -#ifdef DEBUG_DL_DMRS - LOG_I(PHY,"Ns %d, p %d, ind_dword %d, ind_qpsk_symbol %d\n", - Ns,p,ind_dword,ind_qpsk_symb); - LOG_I(PHY,"index = %d\n",(nr_gold_pdsch[0][Ns][lp][ind_dword]>>(2*ind_qpsk_symb))&3); + idx = ((((nr_gold_pdsch[(i<<1)>>5])>>((i<<1)&0x1f))&1)<<1) ^ (((nr_gold_pdsch[((i<<1)+1)>>5])>>(((i<<1)+1)&0x1f))&1); + ((int16_t*)output)[i<<1] = mod_table[(NR_MOD_TABLE_QPSK_OFFSET + idx)<<1]; + ((int16_t*)output)[(i<<1)+1] = mod_table[((NR_MOD_TABLE_QPSK_OFFSET + idx)<<1) + 1]; +#ifdef DEBUG_PDSCH + printf("nr_pdsch_dmrs_rx dmrs config type %d port %d nb_pdsch_rb %d\n", config_type, p, nb_pdsch_rb); + printf("wf[%d] = %d wt[%d]= %d\n", i&1, wf[p-1000][i&1], lp, wt[p-1000][lp]); + printf("i %d idx %d pdsch gold %u b0-b1 %d-%d mod_dmrs %d %d\n", i, idx, nr_gold_pdsch[(i<<1)>>5], (((nr_gold_pdsch[(i<<1)>>5])>>((i<<1)&0x1f))&1), + (((nr_gold_pdsch[((i<<1)+1)>>5])>>(((i<<1)+1)&0x1f))&1), ((int16_t*)output)[i<<1], ((int16_t*)output)[(i<<1)+1]); #endif - - k++; - } + } + } else { + LOG_E(PHY,"extended cp not supported for PDSCH DMRS yet\n"); } - } } else { LOG_E(PHY,"Illegal p %d PDSCH DMRS port\n",p); } @@ -256,3 +164,5 @@ int nr_pbch_dmrs_rx(int symbol,unsigned int *nr_gold_pbch,int32_t *output ) return(0); } + + diff --git a/openair1/PHY/NR_REFSIG/nr_gen_mod_table.m b/openair1/PHY/NR_REFSIG/nr_gen_mod_table.m index a081ad1baa..13a8d4ad09 100644 --- a/openair1/PHY/NR_REFSIG/nr_gen_mod_table.m +++ b/openair1/PHY/NR_REFSIG/nr_gen_mod_table.m @@ -3,12 +3,38 @@ % 0 .. "0" % 1,2 .. BPSK(0),BPSK(1) +K = 1/sqrt(2); + % Amplitude for BPSK (\f$ 2^15 \times 1/\sqrt{2}\f$) BPSK = 23170; % Amplitude for QPSK (\f$ 2^15 \times 1/\sqrt{2}\f$) QPSK = 23170; + +% First Amplitude for QAM16 (\f$ 2^15 \times 2/\sqrt{10}\f$) +QAM16_n1 = 20724; +% Second Amplitude for QAM16 (\f$ 2^15 \times 1/\sqrt{10}\f$) +QAM16_n2 = 10362; + +% First Amplitude for QAM64 (\f$ 2^15 \times 4/\sqrt{42}\f$) +QAM64_n1 = 20225; +% Second Amplitude for QAM64 (\f$ 2^15 \times 2/\sqrt{42}\f$) +QAM64_n2 = 10112; +% Third Amplitude for QAM64 (\f$ 2^15 \times 1/\sqrt{42}\f$) +QAM64_n3 = 5056; + + +% First Amplitude for QAM256 (\f$ 2^15 \times 8/\sqrt{170}\f$) +QAM256_n1 = 20105; +% Second Amplitude for QAM256 (\f$ 2^15 \times 4/\sqrt{170}\f$) +QAM256_n2 = 10052; +% Third Amplitude for QAM256 (\f$ 2^15 \times 2/\sqrt{170}\f$) +QAM256_n3 = 5026; +% Third Amplitude for QAM256 (\f$ 2^15 \times 1/\sqrt{170}\f$) +QAM256_n4 = 2513; + + % BPSK for b = 0:1 bpsk_table(b+1) = (1 - 2*b)*BPSK + 1j*(1-2*b)*BPSK; @@ -17,12 +43,91 @@ end %% QPSK for r=0:1 %0 -- 1 LS for j=0:1 %0 -- 1 MS + %% Formula is dispalayed + qpsk_table(2*r+j+1) = ((1-r*2)*QPSK + 1j*(1-2*j)*QPSK); end end -table = [ 0; bpsk_table(:); qpsk_table(:) ]; +%% QAM16 + for a=-1:2:1 + for b=-1:2:1 + index = (1+a) + (1+b)/2; + qam16_table(index+1) = -a*(QAM16_n1 + (b*QAM16_n2)); + end + end +for b0=0:1 +for b1=0:1 +for b2=0:1 +for b3=0:1 +qam16_table2(b0*8+b1*4+b2*2+b3*1+1) = qam16_table(b0*2+b2*1+1) + 1j*qam16_table(b1*2+b3*1+1); +end +end +end +end + + +%% QAM64 + for a=-1:2:1 + for b=-1:2:1 + for c=-1:2:1 + index = (1+a)*2 + (1+b) + (1+c)/2; + qam64_table(index+1) = -a*(QAM64_n1 + b*(QAM64_n2 + (c*QAM64_n3))); + end + end + end + +for b0=0:1 +for b1=0:1 +for b2=0:1 +for b3=0:1 +for b4=0:1 +for b5=0:1 +qam64_table2(b0*32+b1*16+b2*8+b3*4+b4*2+b5*1+1) = qam64_table(b0*4+b2*2+b4*1+1) + 1j*qam64_table(b1*4+b3*2+b5*1+1); +end +end +end +end +end +end + + +%%256QAM +%% QAM256 ******************************************************************************************* + for a=-1:2:1 + for b=-1:2:1 + for c=-1:2:1 + for d = -1:2:1 + + index = (1+a)*4 + (1+b)*2 + (1+c)+(1+d)/2; + qam256_table(index+1) = -a*(QAM256_n1 + b*(QAM256_n2 + (c*(QAM256_n3+d*QAM256_n4)))); + end + end + end +end +for b0=0:1 +for b1=0:1 +for b2=0:1 +for b3=0:1 +for b4=0:1 +for b5=0:1 +for b6=0:1 +for b7=0:1 + +%%qam64_table2(b0*32+b1*16+b2*8+b3*4+b4*2+b5*1+1) = qam64_table(b0*4+b2*2+b4*1+1) + 1j*qam64_table(b1*4+b3*2+b5*1+1); +qam256_table2(b0*128+b1*64+b2*32+b3*16+b4*8+b5*4+b6*2+b7*1+1) = qam256_table(b0*8+b2*4+b4*2+b6*1+1) + 1j*qam256_table(b1*8+b3*4+b5*2+b7*1+1); +end +end +end +end +end +end +end +end + +table = round(K * [ 0; bpsk_table(:); qpsk_table(:);qam16_table2(:); qam64_table2(:);qam256_table2(:) ]); +%scatter (real(qam256_table2), imag(qam256_table2), 'x'); save mod_table.mat table table2 = zeros(1,length(table)*2); @@ -33,7 +138,10 @@ fd = fopen("nr_mod_table.h","w"); fprintf(fd,"#define NR_MOD_TABLE_SIZE_SHORT %d\n", length(table)*2); fprintf(fd,"#define NR_MOD_TABLE_BPSK_OFFSET %d\n", 1); fprintf(fd,"#define NR_MOD_TABLE_QPSK_OFFSET %d\n", 3); -fprintf(fd,"short nr_mod_table[MOD_TABLE_SIZE_SHORT] = {"); +fprintf(fd,"#define NR_MOD_TABLE_QAM16_OFFSET %d\n", 7); +fprintf(fd,"#define NR_MOD_TABLE_QAM64_OFFSET %d\n", 23); +fprintf(fd,"#define NR_MOD_TABLE_QAM256_OFFSET %d\n", 87); +fprintf(fd,"short nr_mod_table[NR_MOD_TABLE_SIZE_SHORT] = {"); fprintf(fd,"%d,",table2(1:end-1)); fprintf(fd,"%d};\n",table2(end)); fclose(fd); diff --git a/openair1/PHY/NR_REFSIG/nr_gold.c b/openair1/PHY/NR_REFSIG/nr_gold.c index c4e59f907a..968f7fa7a4 100644 --- a/openair1/PHY/NR_REFSIG/nr_gold.c +++ b/openair1/PHY/NR_REFSIG/nr_gold.c @@ -65,7 +65,7 @@ void nr_init_pdcch_dmrs(PHY_VARS_gNB* gNB, uint32_t Nid) for (uint8_t symb=0; symbsymbols_per_slot; symb++) { reset = 1; - x2 = ((1<<17) * (14*slot+symb+1) * ((Nid<<1)+1) + (Nid<<1))&(((uint32_t)1<<31)-1); + x2 = ((1<<17) * (14*slot+symb+1) * ((Nid<<1)+1) + (Nid<<1)); for (uint32_t n=0; nframe_parms; + uint32_t ****pdsch_dmrs = gNB->nr_gold_pdsch_dmrs; + + uint16_t N_n_scid[NR_MAX_NB_CODEWORDS]={Nid, Nid}; // Not correct, appropriate scrambling IDs have to be updated to support DCI 1_1 + uint8_t n_scid=0; // again works only for 1_0 + for (uint8_t slot=0; slotslots_per_frame; slot++) { + for (uint8_t symb=0; symbsymbols_per_slot; symb++) { + for (uint8_t q=0; qsymbols_per_slot*slot+symb+1) * ((N_n_scid[n_scid]<<1)+1) +((N_n_scid[n_scid]<<1)+n_scid)); + for (uint32_t n=0; nframe_parms.Nid_cell; @@ -123,6 +123,8 @@ void nr_gold_pdsch(PHY_VARS_NR_UE* ue,unsigned short lbar,unsigned short *n_idDM nid = n_idDMRS[nscid]; else nid = ue->frame_parms.Nid_cell; + + //printf("gold pdsch nid %d lbar %d\n",nid,lbar); for (ns=0; ns<20; ns++) { @@ -130,6 +132,8 @@ void nr_gold_pdsch(PHY_VARS_NR_UE* ue,unsigned short lbar,unsigned short *n_idDM x2tmp0 = ((14*ns+(lbar+l)+1)*((nid<<1)+1))<<17; x2 = (x2tmp0+(nid<<1)+nscid)%(1<<31); //cinit + + //printf("ns %d gold pdsch x2 %d\n",ns,x2); x1 = 1+ (1<<31); x2=x2 ^ ((x2 ^ (x2>>1) ^ (x2>>2) ^ (x2>>3))<<31); @@ -143,12 +147,13 @@ void nr_gold_pdsch(PHY_VARS_NR_UE* ue,unsigned short lbar,unsigned short *n_idDM //printf("x1 : %x, x2 : %x\n",x1,x2); } - for (n=0; n<14; n++) { + for (n=0; n<52; n++) { x1 = (x1>>1) ^ (x1>>4); x1 = x1 ^ (x1<<31) ^ (x1<<28); x2 = (x2>>1) ^ (x2>>2) ^ (x2>>3) ^ (x2>>4); x2 = x2 ^ (x2<<31) ^ (x2<<30) ^ (x2<<29) ^ (x2<<28); ue->nr_gold_pdsch[nscid][ns][l][n] = x1^x2; + // if ((ns==2)&&(l==0)) //printf("n=%d : c %x\n",n,x1^x2); } diff --git a/openair1/PHY/NR_REFSIG/nr_mod_table.h b/openair1/PHY/NR_REFSIG/nr_mod_table.h index 3c0c808d1d..a998429699 100644 --- a/openair1/PHY/NR_REFSIG/nr_mod_table.h +++ b/openair1/PHY/NR_REFSIG/nr_mod_table.h @@ -1,4 +1,7 @@ -#define NR_MOD_TABLE_SIZE_SHORT 14 +#define NR_MOD_TABLE_SIZE_SHORT 686 #define NR_MOD_TABLE_BPSK_OFFSET 1 #define NR_MOD_TABLE_QPSK_OFFSET 3 -short nr_mod_table[NR_MOD_TABLE_SIZE_SHORT] = {0,0,23170,23170,-23170,-23170,23170,23170,23170,-23170,-23170,23170,-23170,-23170}; +#define NR_MOD_TABLE_QAM16_OFFSET 7 +#define NR_MOD_TABLE_QAM64_OFFSET 23 +#define NR_MOD_TABLE_QAM256_OFFSET 87 +short nr_mod_table[NR_MOD_TABLE_SIZE_SHORT] = {0,0,16384,16384,-16384,-16384,16384,16384,16384,-16384,-16384,16384,-16384,-16384,7327,7327,7327,21981,21981,7327,21981,21981,7327,-7327,7327,-21981,21981,-7327,21981,-21981,-7327,7327,-7327,21981,-21981,7327,-21981,21981,-7327,-7327,-7327,-21981,-21981,-7327,-21981,-21981,10726,10726,10726,3576,3576,10726,3576,3576,10726,17876,10726,25027,3576,17876,3576,25027,17876,10726,17876,3576,25027,10726,25027,3576,17876,17876,17876,25027,25027,17876,25027,25027,10726,-10726,10726,-3576,3576,-10726,3576,-3576,10726,-17876,10726,-25027,3576,-17876,3576,-25027,17876,-10726,17876,-3576,25027,-10726,25027,-3576,17876,-17876,17876,-25027,25027,-17876,25027,-25027,-10726,10726,-10726,3576,-3576,10726,-3576,3576,-10726,17876,-10726,25027,-3576,17876,-3576,25027,-17876,10726,-17876,3576,-25027,10726,-25027,3576,-17876,17876,-17876,25027,-25027,17876,-25027,25027,-10726,-10726,-10726,-3576,-3576,-10726,-3576,-3576,-10726,-17876,-10726,-25027,-3576,-17876,-3576,-25027,-17876,-10726,-17876,-3576,-25027,-10726,-25027,-3576,-17876,-17876,-17876,-25027,-25027,-17876,-25027,-25027,8886,8886,8886,12439,12439,8886,12439,12439,8886,5332,8886,1778,12439,5332,12439,1778,5332,8886,5332,12439,1778,8886,1778,12439,5332,5332,5332,1778,1778,5332,1778,1778,8886,19547,8886,15993,12439,19547,12439,15993,8886,23101,8886,26655,12439,23101,12439,26655,5332,19547,5332,15993,1778,19547,1778,15993,5332,23101,5332,26655,1778,23101,1778,26655,19547,8886,19547,12439,15993,8886,15993,12439,19547,5332,19547,1778,15993,5332,15993,1778,23101,8886,23101,12439,26655,8886,26655,12439,23101,5332,23101,1778,26655,5332,26655,1778,19547,19547,19547,15993,15993,19547,15993,15993,19547,23101,19547,26655,15993,23101,15993,26655,23101,19547,23101,15993,26655,19547,26655,15993,23101,23101,23101,26655,26655,23101,26655,26655,8886,-8886,8886,-12439,12439,-8886,12439,-12439,8886,-5332,8886,-1778,12439,-5332,12439,-1778,5332,-8886,5332,-12439,1778,-8886,1778,-12439,5332,-5332,5332,-1778,1778,-5332,1778,-1778,8886,-19547,8886,-15993,12439,-19547,12439,-15993,8886,-23101,8886,-26655,12439,-23101,12439,-26655,5332,-19547,5332,-15993,1778,-19547,1778,-15993,5332,-23101,5332,-26655,1778,-23101,1778,-26655,19547,-8886,19547,-12439,15993,-8886,15993,-12439,19547,-5332,19547,-1778,15993,-5332,15993,-1778,23101,-8886,23101,-12439,26655,-8886,26655,-12439,23101,-5332,23101,-1778,26655,-5332,26655,-1778,19547,-19547,19547,-15993,15993,-19547,15993,-15993,19547,-23101,19547,-26655,15993,-23101,15993,-26655,23101,-19547,23101,-15993,26655,-19547,26655,-15993,23101,-23101,23101,-26655,26655,-23101,26655,-26655,-8886,8886,-8886,12439,-12439,8886,-12439,12439,-8886,5332,-8886,1778,-12439,5332,-12439,1778,-5332,8886,-5332,12439,-1778,8886,-1778,12439,-5332,5332,-5332,1778,-1778,5332,-1778,1778,-8886,19547,-8886,15993,-12439,19547,-12439,15993,-8886,23101,-8886,26655,-12439,23101,-12439,26655,-5332,19547,-5332,15993,-1778,19547,-1778,15993,-5332,23101,-5332,26655,-1778,23101,-1778,26655,-19547,8886,-19547,12439,-15993,8886,-15993,12439,-19547,5332,-19547,1778,-15993,5332,-15993,1778,-23101,8886,-23101,12439,-26655,8886,-26655,12439,-23101,5332,-23101,1778,-26655,5332,-26655,1778,-19547,19547,-19547,15993,-15993,19547,-15993,15993,-19547,23101,-19547,26655,-15993,23101,-15993,26655,-23101,19547,-23101,15993,-26655,19547,-26655,15993,-23101,23101,-23101,26655,-26655,23101,-26655,26655,-8886,-8886,-8886,-12439,-12439,-8886,-12439,-12439,-8886,-5332,-8886,-1778,-12439,-5332,-12439,-1778,-5332,-8886,-5332,-12439,-1778,-8886,-1778,-12439,-5332,-5332,-5332,-1778,-1778,-5332,-1778,-1778,-8886,-19547,-8886,-15993,-12439,-19547,-12439,-15993,-8886,-23101,-8886,-26655,-12439,-23101,-12439,-26655,-5332,-19547,-5332,-15993,-1778,-19547,-1778,-15993,-5332,-23101,-5332,-26655,-1778,-23101,-1778,-26655,-19547,-8886,-19547,-12439,-15993,-8886,-15993,-12439,-19547,-5332,-19547,-1778,-15993,-5332,-15993,-1778,-23101,-8886,-23101,-12439,-26655,-8886,-26655,-12439,-23101,-5332,-23101,-1778,-26655,-5332,-26655,-1778,-19547,-19547,-19547,-15993,-15993,-19547,-15993,-15993,-19547,-23101,-19547,-26655,-15993,-23101,-15993,-26655,-23101,-19547,-23101,-15993,-26655,-19547,-26655,-15993,-23101,-23101,-23101,-26655,-26655,-23101,-26655,-26655}; diff --git a/openair1/PHY/NR_REFSIG/nr_refsig.h b/openair1/PHY/NR_REFSIG/nr_refsig.h index f6a8b5683b..cb165eed5e 100644 --- a/openair1/PHY/NR_REFSIG/nr_refsig.h +++ b/openair1/PHY/NR_REFSIG/nr_refsig.h @@ -37,6 +37,6 @@ void nr_init_pbch_dmrs(PHY_VARS_gNB* gNB); @param Nid is used for the initialization of x2, Physical cell Id by default or upper layer configured pdcch_scrambling_ID */ void nr_init_pdcch_dmrs(PHY_VARS_gNB* gNB, uint32_t Nid); - +void nr_init_pdsch_dmrs(PHY_VARS_gNB* gNB, uint32_t Nid); #endif diff --git a/openair1/PHY/NR_REFSIG/refsig_defs_ue.h b/openair1/PHY/NR_REFSIG/refsig_defs_ue.h index a4cb2be7b8..a47e65d0b0 100644 --- a/openair1/PHY/NR_REFSIG/refsig_defs_ue.h +++ b/openair1/PHY/NR_REFSIG/refsig_defs_ue.h @@ -43,13 +43,12 @@ int nr_pdcch_dmrs_rx(PHY_VARS_NR_UE *ue, unsigned short nb_rb_corset); int nr_pdsch_dmrs_rx(PHY_VARS_NR_UE *ue, - uint8_t eNB_offset, unsigned int Ns, - unsigned int nr_gold_pdsch[2][20][2][21], + unsigned int *nr_gold_pdsch, int32_t *output, unsigned short p, - int length_dmrs, - unsigned short nb_rb_pdsch); + unsigned char lp, + unsigned short nb_pdsch_rb); void nr_gold_pbch(PHY_VARS_NR_UE* ue); diff --git a/openair1/PHY/NR_TRANSPORT/nr_dci.c b/openair1/PHY/NR_TRANSPORT/nr_dci.c index 52d5f05a1a..7fa649bff8 100644 --- a/openair1/PHY/NR_TRANSPORT/nr_dci.c +++ b/openair1/PHY/NR_TRANSPORT/nr_dci.c @@ -20,7 +20,7 @@ */ /*! \file PHY/NR_TRANSPORT/nr_dci.c -* \brief Implements DCI encoding/decoding and PDCCH TX/RX procedures (38.212/38.213/38.214). V15.2.0 2018-06. +* \brief Implements DCI encoding and PDCCH TX procedures (38.212/38.213/38.214). V15.2.0 2018-06. * \author Guy De Souza * \date 2018 * \version 0.1 @@ -336,9 +336,5 @@ printf("scrambled output: [0]->0x%08x \t [1]->0x%08x \t [2]->0x%08x \t [3]->0x%0 } } -#ifdef DEBUG_DCI - write_output("txdataF_dci.m", "txdataF_dci", txdataF[0], frame_parms.samples_per_frame_wCP>>1, 1, 1); -#endif - return 0; } diff --git a/openair1/PHY/NR_TRANSPORT/nr_dci.h b/openair1/PHY/NR_TRANSPORT/nr_dci.h index 517f5672a1..4bf250dbff 100644 --- a/openair1/PHY/NR_TRANSPORT/nr_dci.h +++ b/openair1/PHY/NR_TRANSPORT/nr_dci.h @@ -51,7 +51,8 @@ void nr_fill_dci_and_dlsch(PHY_VARS_gNB *gNB, int subframe, gNB_L1_rxtx_proc_t *proc, NR_gNB_DCI_ALLOC_t *dci_alloc, - nfapi_nr_dl_config_request_pdu_t *pdu); + nfapi_nr_dl_config_dci_dl_pdu *pdu, + nfapi_nr_dl_config_dlsch_pdu *dlsch_pdu); void nr_fill_cce_list(NR_gNB_DCI_ALLOC_t* dci_alloc, uint16_t n_shift, uint8_t m); diff --git a/openair1/PHY/NR_TRANSPORT/nr_dci_tools.c b/openair1/PHY/NR_TRANSPORT/nr_dci_tools.c index 59c061553c..1d29b057fc 100644 --- a/openair1/PHY/NR_TRANSPORT/nr_dci_tools.c +++ b/openair1/PHY/NR_TRANSPORT/nr_dci_tools.c @@ -31,8 +31,12 @@ */ #include "nr_dci.h" + //#define DEBUG_FILL_DCI +#include "nr_dlsch.h" + + void nr_fill_cce_list(NR_gNB_DCI_ALLOC_t* dci_alloc, uint16_t n_shift, uint8_t m) { nr_cce_t* cce; @@ -61,7 +65,7 @@ void nr_fill_cce_list(NR_gNB_DCI_ALLOC_t* dci_alloc, uint16_t n_shift, uint8_t m AssertFatal(cond==0, "CCE to REG interleaving: Invalid configuration leading to non integer C\n"); C = N_reg/(bsize*R); - tmp = L * (( Y + (uint16_t)(floor((m*N_cce)/(L*M_s_max))) + n_CI ) % ((uint16_t)floor(N_cce/L))); + tmp = L * (( Y + (m*N_cce)/(L*M_s_max) + n_CI ) % (N_cce/L)); LOG_I(PHY, "CCE list generation for candidate %d: bundle size %d ilv size %d tmp %d\n", m, bsize, R, tmp); for (uint8_t cce_idx=0; cce_idxdci_pdu; memset((void*)dci_pdu,0,2*sizeof(uint64_t)); - nfapi_nr_dl_config_dci_dl_pdu_rel15_t *pdu_rel15 = &pdu->dci_dl_pdu.dci_dl_pdu_rel15; - nfapi_nr_dl_config_pdcch_parameters_rel15_t *params_rel15 = &pdu->dci_dl_pdu.pdcch_params_rel15; + nfapi_nr_dl_config_dci_dl_pdu_rel15_t *pdu_rel15 = &pdcch_pdu->dci_dl_pdu_rel15; + nfapi_nr_dl_config_pdcch_parameters_rel15_t *params_rel15 = &pdcch_pdu->pdcch_params_rel15; nfapi_nr_config_request_t *cfg = &gNB->gNB_config; - + NR_gNB_DLSCH_t *dlsch = gNB->dlsch[0][0]; + NR_DL_gNB_HARQ_t **harq = dlsch->harq_processes; uint16_t N_RB = params_rel15->n_RB_BWP; uint8_t fsize=0, pos=0, cand_idx=0; + dci_alloc->L = 8; memcpy((void*)&dci_alloc->pdcch_params, (void*)params_rel15, sizeof(nfapi_nr_dl_config_pdcch_parameters_rel15_t)); dci_alloc->size = nr_get_dci_size(dci_alloc->pdcch_params.dci_format, @@ -142,6 +150,7 @@ void nr_fill_dci_and_dlsch(PHY_VARS_gNB *gNB, /// Payload generation switch(params_rel15->dci_format) { + case NFAPI_NR_DL_DCI_FORMAT_1_0: switch(params_rel15->rnti_type) { case NFAPI_NR_RNTI_RA: @@ -182,70 +191,76 @@ void nr_fill_dci_and_dlsch(PHY_VARS_gNB *gNB, case NFAPI_NR_RNTI_C: // indicating a DL DCI format 1bit - *dci_pdu |= (pdu_rel15->format_indicator&1)<<(dci_alloc->size-pos++); + pos++; + *dci_pdu |= (pdu_rel15->format_indicator&1)<<(dci_alloc->size-pos); // Freq domain assignment (275rb >> fsize = 16) fsize = (int)ceil( log2( (N_RB*(N_RB+1))>>1 ) ); - for (int i=0; ifrequency_domain_assignment>>(fsize-i-1))&1)<<(dci_alloc->size-pos++); + pos+=fsize; + *dci_pdu |= ((pdu_rel15->frequency_domain_assignment&((1<size-pos)); if (((pdu_rel15->frequency_domain_assignment+1)&1) ==0) //fsize are all 1 38.212 p86 { // ra_preamble_index 6 bits - for (int i=0; i<6; i++) - *dci_pdu |= ((pdu_rel15->ra_preamble_index>>(5-i))&1)<<(dci_alloc->size-pos++); + pos+=6; + *dci_pdu |= ((pdu_rel15->ra_preamble_index&0x3f)<<(dci_alloc->size-pos)); // UL/SUL indicator 1 bit - *dci_pdu |= (pdu_rel15->ul_sul_indicator&1)<<(dci_alloc->size-pos++); + pos++; + *dci_pdu |= (pdu_rel15->ul_sul_indicator&1)<<(dci_alloc->size-pos); // SS/PBCH index 6 bits - for (int i=0; i<6; i++) - *dci_pdu |= ((pdu_rel15->ss_pbch_index>>(5-i))&1)<<(dci_alloc->size-pos++); + pos+=6; + *dci_pdu |= ((pdu_rel15->ss_pbch_index&0x3f)<<(dci_alloc->size-pos)); // prach_mask_index 4 bits - for (int i=0; i<4; i++) - *dci_pdu |= ((pdu_rel15->prach_mask_index>>(3-i))&1)<<(dci_alloc->size-pos++); + pos+=4; + *dci_pdu |= ((pdu_rel15->prach_mask_index&0xf)<<(dci_alloc->size-pos)); } //end if else { // Time domain assignment 4bit - for (int i=0; i<4; i++) - *dci_pdu |= ((pdu_rel15->time_domain_assignment>>(3-i))&1)<<(dci_alloc->size-pos++); + + pos+=4; + *dci_pdu |= ((pdu_rel15->time_domain_assignment&0xf) << (dci_alloc->size-pos)); + // VRB to PRB mapping 1bit - *dci_pdu |= (pdu_rel15->vrb_to_prb_mapping&1)<<(dci_alloc->size-pos++); + pos++; + *dci_pdu |= (pdu_rel15->vrb_to_prb_mapping&1)<<(dci_alloc->size-pos); // MCS 5bit //bit over 32, so dci_pdu ++ - for (int i=0; i<5; i++) - *dci_pdu |= ((pdu_rel15->mcs>>(4-i))&1)<<(dci_alloc->size-pos++); + pos+=5; + *dci_pdu |= (pdu_rel15->mcs&0x1f)<<(dci_alloc->size-pos); // New data indicator 1bit - *dci_pdu |= (pdu_rel15->ndi&1)<<(dci_alloc->size-pos++); + pos++; + *dci_pdu |= (pdu_rel15->ndi&1)<<(dci_alloc->size-pos); // Redundancy version 2bit - for (int i=0; i<2; i++) - *dci_pdu |= ((pdu_rel15->rv>>(1-i))&1)<<(dci_alloc->size-pos++); + pos+=2; + *dci_pdu |= (pdu_rel15->rv&0x3)<<(dci_alloc->size-pos); // HARQ process number 4bit - for (int i=0; i<4; i++) - *dci_pdu |= ((pdu_rel15->harq_pid>>(3-i))&1)<<(dci_alloc->size-pos++); + pos+=4; + *dci_pdu |= ((pdu_rel15->harq_pid&0xf)<<(dci_alloc->size-pos)); // Downlink assignment index 2bit - for (int i=0; i<2; i++) - *dci_pdu |= ((pdu_rel15->dai>>(1-i))&1)<<(dci_alloc->size-pos++); + pos+=2; + *dci_pdu |= ((pdu_rel15->dai&3)<<(dci_alloc->size-pos)); // TPC command for scheduled PUCCH 2bit - for (int i=0; i<2; i++) - *dci_pdu |= ((pdu_rel15->tpc>>(1-i))&1)<<(dci_alloc->size-pos++); + pos+=2; + *dci_pdu |= ((pdu_rel15->tpc&3)<<(dci_alloc->size-pos)); // PUCCH resource indicator 3bit - for (int i=0; i<3; i++) - *dci_pdu |= ((pdu_rel15->pucch_resource_indicator>>(2-i))&1)<<(dci_alloc->size-pos++); + pos+=3; + *dci_pdu |= ((pdu_rel15->pucch_resource_indicator&0x7)<<(dci_alloc->size-pos)); // PDSCH-to-HARQ_feedback timing indicator 3bit - for (int i=0; i<3; i++) - *dci_pdu |= ((pdu_rel15->pdsch_to_harq_feedback_timing_indicator>>(2-i))&1)<<(dci_alloc->size-pos++); + pos+=3; + *dci_pdu |= ((pdu_rel15->pdsch_to_harq_feedback_timing_indicator&0x7)<<(dci_alloc->size-pos)); } //end else break; @@ -275,6 +290,7 @@ void nr_fill_dci_and_dlsch(PHY_VARS_gNB *gNB, for (int i=0; i<2; i++) *dci_pdu |= ((pdu_rel15->tb_scaling>>(1-i))&1)<<(dci_alloc->size-pos++); + break; case NFAPI_NR_RNTI_SI: @@ -328,9 +344,10 @@ void nr_fill_dci_and_dlsch(PHY_VARS_gNB *gNB, for (int i=0; i<2; i++) *dci_pdu |= ((pdu_rel15->tpc>>(1-i))&1)<<(dci_alloc->size-pos++); - // PUCCH resource indicator – 3 bits - for (int i=0; i<3; i++) - *dci_pdu |= ((pdu_rel15->pucch_resource_indicator>>(2-i))&1)<<(dci_alloc->size-pos++); + + // LOG_I(PHY, "DCI PDU: [0]->0x%08llx \t [1]->0x%08llx \t [2]->0x%08llx \t [3]->0x%08llx\n", + // dci_pdu[0], dci_pdu[1], dci_pdu[2], dci_pdu[3]); + // PDSCH-to-HARQ_feedback timing indicator – 3 bits for (int i=0; i<3; i++) @@ -425,4 +442,7 @@ void nr_fill_dci_and_dlsch(PHY_VARS_gNB *gNB, LOG_I(PHY, "DCI PDU: [0]->0x%16llxx \t [1]->0x%16llx \n",(long long unsigned int)dci_pdu[0], (long long unsigned int)dci_pdu[1]); LOG_I(PHY, "DCI type %d payload (size %d) generated on candidate %d\n", dci_alloc->pdcch_params.dci_format, dci_alloc->size, cand_idx); + /// DLSCH struct + memcpy((void*)&harq[dci_alloc->harq_pid]->dlsch_pdu, (void*)dlsch_pdu, sizeof(nfapi_nr_dl_config_dlsch_pdu)); + } diff --git a/openair1/PHY/NR_TRANSPORT/nr_dlsch.c b/openair1/PHY/NR_TRANSPORT/nr_dlsch.c new file mode 100644 index 0000000000..b65b48c39e --- /dev/null +++ b/openair1/PHY/NR_TRANSPORT/nr_dlsch.c @@ -0,0 +1,374 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.1 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +/*! \file PHY/NR_TRANSPORT/dlsch_decoding.c +* \brief Top-level routines for transmission of the PDSCH 38211 v 15.2.0 +* \author Guy De Souza +* \date 2018 +* \version 0.1 +* \company Eurecom +* \email: desouza@eurecom.fr +* \note +* \warning +*/ + +#include "nr_dlsch.h" +#include "nr_dci.h" +#include "nr_sch_dmrs.h" + +//#define DEBUG_DLSCH +//#define DEBUG_DLSCH_MAPPING + +uint8_t mod_order[5] = {1, 2, 4, 6, 8}; +uint16_t mod_offset[5] = {1,3,7,23,87}; + +void nr_pdsch_codeword_scrambling(uint8_t *in, + uint16_t size, + uint8_t q, + uint32_t Nid, + uint32_t n_RNTI, + uint32_t* out) { + + uint8_t reset, b_idx; + uint32_t x1, x2, s=0; + + reset = 1; + x2 = (n_RNTI<<15) + (q<<14) + Nid; + + for (int i=0; i>b_idx)&1))<>b_idx)&1)<<(order-j-1); + } + + out[i<<1] = nr_mod_table[(offset+idx)<<1]; + out[(i<<1)+1] = nr_mod_table[((offset+idx)<<1)+1]; + } +} + +void nr_pdsch_codeword_modulation(uint32_t *in, + uint8_t Qm, + uint32_t length, + int16_t *out) { + + uint16_t offset = (Qm==2)? NR_MOD_TABLE_QPSK_OFFSET : (Qm==4)? NR_MOD_TABLE_QAM16_OFFSET : \ + (Qm==6)? NR_MOD_TABLE_QAM64_OFFSET: (Qm==8)? NR_MOD_TABLE_QAM256_OFFSET : 0; + AssertFatal(offset, "Invalid modulation order %d\n", Qm); + + for (int i=0; i>b_idx)&1)<<(Qm-j-1); + } + + out[i<<1] = nr_mod_table[(offset+idx)<<1]; + out[(i<<1)+1] = nr_mod_table[((offset+idx)<<1)+1]; + } +} + +void nr_pdsch_layer_mapping(int16_t **mod_symbs, + uint8_t n_layers, + uint16_t n_symbs, + int16_t **tx_layers) { + + switch (n_layers) { + + case 1: + memcpy((void*)tx_layers[0], (void*)mod_symbs[0], (n_symbs<<1)*sizeof(int16_t)); + break; + + case 2: + case 3: + case 4: + for (int i=0; i>1; i++) + for (int l=0; l<2; l++) { + tx_layers[l][i<<1] = mod_symbs[0][((i<<1)+l)<<1]; + tx_layers[l][(i<<1)+1] = mod_symbs[0][(((i<<1)+l)<<1)+1]; + } + for (int i=0; i>2; i++) + for (int l=0; l<3; l++) { + tx_layers[l][i<<1] = mod_symbs[q][((i<<2)+l)<<1]; + tx_layers[l][(i<<1)+1] = mod_symbs[q][(((i<<2)+l)<<1)+1]; + } + break; + + default: + AssertFatal(0, "Invalid number of layers %d\n", n_layers); + } +} + +static inline uint16_t get_pdsch_dmrs_idx(uint8_t n, uint8_t k_prime, uint8_t delta, uint8_t dmrs_type) { + uint16_t dmrs_idx = (dmrs_type)? (6*n+k_prime+delta):((n<<2)+(k_prime<<1)+delta); + return dmrs_idx; +} + +uint8_t nr_generate_pdsch(NR_gNB_DLSCH_t dlsch, + NR_gNB_DCI_ALLOC_t dci_alloc, + uint32_t ***pdsch_dmrs, + int32_t** txdataF, + int16_t amp, + uint8_t slot, + NR_DL_FRAME_PARMS frame_parms, + nfapi_nr_config_request_t config) { + + NR_DL_gNB_HARQ_t *harq = dlsch.harq_processes[dci_alloc.harq_pid]; + nfapi_nr_dl_config_dlsch_pdu_rel15_t *rel15 = &harq->dlsch_pdu.dlsch_pdu_rel15; + nfapi_nr_dl_config_pdcch_parameters_rel15_t pdcch_params = dci_alloc.pdcch_params; + uint32_t scrambled_output[NR_MAX_NB_CODEWORDS][NR_MAX_PDSCH_ENCODED_LENGTH>>5]; + int16_t **mod_symbs = (int16_t**)dlsch.mod_symbs; + int16_t **tx_layers = (int16_t**)dlsch.txdataF; + int8_t Wf[2], Wt[2], l0, l_prime[2], delta; + uint16_t nb_symbols = rel15->nb_mod_symbols; + uint8_t Qm = rel15->modulation_order; + uint16_t encoded_length = nb_symbols*Qm; + + /// CRC, coding, interleaving and rate matching + nr_dlsch_encoding(harq->pdu, slot, &dlsch, &frame_parms); +#ifdef DEBUG_DLSCH +printf("PDSCH encoding:\nPayload:\n"); +for (int i=0; i>7; i++) { + for (int j=0; j<16; j++) + printf("0x%02x\t", harq->pdu[(i<<4)+j]); + printf("\n"); +} +printf("\nEncoded payload:\n"); +for (int i=0; i>3; i++) { + for (int j=0; j<8; j++) + printf("%d", harq->f[(i<<3)+j]); + printf("\t"); +} +printf("\n"); +#endif + + /// scrambling + for (int q=0; qnb_codewords; q++) + memset((void*)scrambled_output[q], 0, (encoded_length>>5)*sizeof(uint32_t)); + uint16_t n_RNTI = (pdcch_params.search_space_type == NFAPI_NR_SEARCH_SPACE_TYPE_UE_SPECIFIC)? \ + ((pdcch_params.scrambling_id)?pdcch_params.rnti:0) : 0; + uint16_t Nid = (pdcch_params.search_space_type == NFAPI_NR_SEARCH_SPACE_TYPE_UE_SPECIFIC)? \ + pdcch_params.scrambling_id : config.sch_config.physical_cell_id.value; + for (int q=0; qnb_codewords; q++) + nr_pdsch_codeword_scrambling(harq->f, + encoded_length, + q, + Nid, + n_RNTI, + scrambled_output[q]); +#ifdef DEBUG_DLSCH +printf("PDSCH scrambling:\n"); +for (int i=0; i>8; i++) { + for (int j=0; j<8; j++) + printf("0x%08x\t", scrambled_output[0][(i<<3)+j]); + printf("\n"); +} +#endif + + /// Modulation + for (int q=0; qnb_codewords; q++) + nr_pdsch_codeword_modulation(scrambled_output[q], + Qm, + encoded_length, + mod_symbs[q]); +#ifdef DEBUG_DLSCH +printf("PDSCH Modulation: Qm %d(%d)\n", Qm, nb_symbols); +for (int i=0; i>3; i++) { + for (int j=0; j<8; j++) { + printf("%d %d\t", mod_symbs[0][((i<<3)+j)<<1], mod_symbs[0][(((i<<3)+j)<<1)+1]); + } + printf("\n"); +} +#endif + + + /// Layer mapping + nr_pdsch_layer_mapping(mod_symbs, + rel15->nb_layers, + nb_symbols, + tx_layers); +#ifdef DEBUG_DLSCH +printf("Layer mapping (%d layers):\n", rel15->nb_layers); +for (int l=0; lnb_layers; l++) + for (int i=0; i<(nb_symbols/rel15->nb_layers)>>3; i++) { + for (int j=0; j<8; j++) { + printf("%d %d\t", tx_layers[l][((i<<3)+j)<<1], tx_layers[l][(((i<<3)+j)<<1)+1]); + } + printf("\n"); + } +#endif + + /// Antenna port mapping + //to be moved to init phase potentially, for now tx_layers 1-8 are mapped on antenna ports 1000-1007 + + /// DMRS QPSK modulation + uint16_t n_dmrs = (rel15->n_prb*rel15->nb_re_dmrs)<<1; + int16_t mod_dmrs[n_dmrs<<1]; + uint8_t dmrs_type = config.pdsch_config.dmrs_type.value; + l0 = get_l0(dmrs_type, 2);//config.pdsch_config.dmrs_typeA_position.value); + nr_modulation(pdsch_dmrs[l0][0], n_dmrs, MOD_QPSK, mod_dmrs); // currently only codeword 0 is modulated +#ifdef DEBUG_DLSCH +printf("DMRS modulation (single symbol %d, %d symbols, type %d):\n", l0, n_dmrs>>1, dmrs_type); +for (int i=0; i>4; i++) { + for (int j=0; j<8; j++) { + printf("%d %d\t", mod_dmrs[((i<<3)+j)<<1], mod_dmrs[(((i<<3)+j)<<1)+1]); + } + printf("\n"); +} +#endif + + /// Resource mapping + + + // Non interleaved VRB to PRB mapping + uint16_t start_sc = frame_parms.first_carrier_offset + frame_parms.ssb_start_subcarrier; + if (start_sc >= frame_parms.ofdm_symbol_size) + start_sc -= frame_parms.ofdm_symbol_size; + /*rel15->start_prb*NR_NB_SC_PER_RB + + ((pdcch_params.search_space_type == NFAPI_NR_SEARCH_SPACE_TYPE_COMMON) && (pdcch_params.dci_format == NFAPI_NR_DL_DCI_FORMAT_1_0))?\ + ((frame_parms.ssb_start_subcarrier/NR_NB_SC_PER_RB + pdcch_params.rb_offset)*NR_NB_SC_PER_RB) : 0;*/ + +#ifdef DEBUG_DLSCH_MAPPING +printf("PDSCH resource mapping started (start SC %d\tstart symbol %d\tN_PRB %d\tnb_symbols %d)\n", +start_sc, rel15->start_symbol, rel15->n_prb, rel15->nb_symbols); +#endif + + for (int ap=0; apnb_layers; ap++) { + + // DMRS params for this ap + get_Wt(Wt, ap, dmrs_type); + get_Wf(Wf, ap, dmrs_type); + delta = get_delta(ap, dmrs_type); + l_prime[0] = 0; // single symbol ap 0 + uint8_t dmrs_symbol = l0+l_prime[0]; +#ifdef DEBUG_DLSCH_MAPPING +printf("DMRS params for ap %d: Wt %d %d \t Wf %d %d \t delta %d \t l_prime %d \t l0 %d\tDMRS symbol %d\n", +ap, Wt[0], Wt[1], Wf[0], Wf[1], delta, l_prime[0], l0, dmrs_symbol); +#endif + uint8_t k_prime=0; + uint16_t m=0, n=0, dmrs_idx=0, k=0; + + for (int l=rel15->start_symbol; lstart_symbol+rel15->nb_symbols; l++) { + k = start_sc; + for (int i=0; in_prb*NR_NB_SC_PER_RB; i++) { +// Note the dmrs index modulo 2048 only works for that symbol size -- to be fixed soon +// Also the different amplitudes are for debug purposes and temporary + if ((l == dmrs_symbol) && (k == ((start_sc+get_pdsch_dmrs_idx(n, k_prime, delta, dmrs_type))&((1<<11)-1)))) { + ((int16_t*)txdataF[ap])[(l*frame_parms.ofdm_symbol_size + k)<<1] = (Wt[l_prime[0]]*Wf[k_prime]*(amp>>1)*mod_dmrs[dmrs_idx<<1]) >> 15; + ((int16_t*)txdataF[ap])[((l*frame_parms.ofdm_symbol_size + k)<<1) + 1] = (Wt[l_prime[0]]*Wf[k_prime]*(amp>>1)*mod_dmrs[(dmrs_idx<<1) + 1]) >> 15; +#ifdef DEBUG_DLSCH_MAPPING +printf("dmrs_idx %d\t l %d \t k %d \t k_prime %d \t n %d \t txdataF: %d %d\n", +dmrs_idx, l, k, k_prime, n, ((int16_t*)txdataF[ap])[(l*frame_parms.ofdm_symbol_size + k)<<1], +((int16_t*)txdataF[ap])[((l*frame_parms.ofdm_symbol_size + k)<<1) + 1]); +#endif + dmrs_idx++; + k_prime++; + k_prime&=1; + n+=(k_prime)?0:1; + } + + else { + + ((int16_t*)txdataF[ap])[(l*frame_parms.ofdm_symbol_size + k)<<1] = ((amp<<1) * tx_layers[ap][m<<1]) >> 15; + ((int16_t*)txdataF[ap])[((l*frame_parms.ofdm_symbol_size + k)<<1) + 1] = ((amp<<1) * tx_layers[ap][(m<<1) + 1]) >> 15; +#ifdef DEBUG_DLSCH_MAPPING +printf("m %d\t l %d \t k %d \t txdataF: %d %d\n", +m, l, k, ((int16_t*)txdataF[ap])[(l*frame_parms.ofdm_symbol_size + k)<<1], +((int16_t*)txdataF[ap])[((l*frame_parms.ofdm_symbol_size + k)<<1) + 1]); +#endif + m++; + } + if (++k >= frame_parms.ofdm_symbol_size) + k -= frame_parms.ofdm_symbol_size; + } + } + } + return 0; +} diff --git a/openair1/PHY/NR_TRANSPORT/nr_dlsch.h b/openair1/PHY/NR_TRANSPORT/nr_dlsch.h new file mode 100644 index 0000000000..745b92f5fe --- /dev/null +++ b/openair1/PHY/NR_TRANSPORT/nr_dlsch.h @@ -0,0 +1,99 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.1 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +/*! \file PHY/LTE_TRANSPORT/defs.h +* \brief data structures for PDSCH/DLSCH/PUSCH/ULSCH physical and transport channel descriptors (TX/RX) +* \author R. Knopp +* \date 2011 +* \version 0.1 +* \company Eurecom +* \email: raymond.knopp@eurecom.fr, florian.kaltenberger@eurecom.fr, oscar.tonelli@yahoo.it +* \note +* \warning +*/ + +#include "PHY/defs_gNB.h" + +extern short nr_mod_table[NR_MOD_TABLE_SIZE_SHORT]; + +void nr_get_time_domain_allocation_type(nfapi_nr_config_request_t config, + nfapi_nr_dl_config_dci_dl_pdu dci_pdu, + nfapi_nr_dl_config_dlsch_pdu *dlsch_pdu); + +void nr_check_time_alloc(uint8_t S, uint8_t L, nfapi_nr_config_request_t config); + +uint16_t get_RIV(uint16_t rb_start, uint16_t L, uint16_t N_RB); + +uint16_t get_SLIV(uint8_t S, uint8_t L); + +uint8_t nr_get_S(uint8_t row_idx, uint8_t CP, uint8_t time_alloc_type, uint8_t dmrs_typeA_position); + +void nr_get_rbg_parms(NR_BWP_PARMS* bwp, uint8_t config_type); + +void nr_get_rbg_list(uint32_t bitmap, uint8_t n_rbg, uint8_t* rbg_list); + +void nr_get_PRG_parms(NR_BWP_PARMS* bwp, NR_gNB_DCI_ALLOC_t dci_alloc, uint8_t prb_bundling_type); + +uint8_t nr_get_Qm(uint8_t Imcs, uint8_t table_idx); + +uint32_t nr_get_code_rate(uint8_t Imcs, uint8_t table_idx); + +void nr_get_tbs(nfapi_nr_dl_config_dlsch_pdu *dlsch_pdu, + nfapi_nr_dl_config_dci_dl_pdu dci_pdu, + nfapi_nr_config_request_t config); + +void nr_pdsch_codeword_scrambling(uint8_t *in, + uint16_t size, + uint8_t q, + uint32_t Nid, + uint32_t n_RNTI, + uint32_t* out); + +void nr_pdsch_codeword_modulation(uint32_t *in, + uint8_t Qm, + uint32_t length, + int16_t *out); + +void nr_pdsch_layer_mapping(int16_t **mod_symbs, + uint8_t n_layers, + uint16_t n_symbs, + int16_t **tx_layers); + +uint8_t nr_generate_pdsch(NR_gNB_DLSCH_t dlsch, + NR_gNB_DCI_ALLOC_t dci_alloc, + uint32_t ***pdsch_dmrs, + int32_t** txdataF, + int16_t amp, + uint8_t subframe, + NR_DL_FRAME_PARMS frame_parms, + nfapi_nr_config_request_t config); + +/** \brief Computes available bits G. + @param nb_rb, nb_symb_sch, nb_re_dmrs, length_dmrs */ +uint32_t nr_get_G(uint16_t nb_rb, uint16_t nb_symb_sch,uint8_t nb_re_dmrs,uint16_t length_dmrs,uint8_t Qm, uint8_t Nl); + +int nr_dlsch_encoding(unsigned char *a, + uint8_t subframe, + NR_gNB_DLSCH_t *dlsch, + NR_DL_FRAME_PARMS* frame_parms); + +void nr_emulate_dlsch_payload(uint8_t* payload, uint16_t size); + diff --git a/openair1/PHY/NR_TRANSPORT/nr_dlsch_coding.c b/openair1/PHY/NR_TRANSPORT/nr_dlsch_coding.c new file mode 100644 index 0000000000..4618f6022f --- /dev/null +++ b/openair1/PHY/NR_TRANSPORT/nr_dlsch_coding.c @@ -0,0 +1,451 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.0 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +/*! \file PHY/LTE_TRANSPORT/dlsch_coding.c +* \brief Top-level routines for implementing LDPC-coded (DLSCH) transport channels from 38-212, 15.2 +* \author H.Wang +* \date 2018 +* \version 0.1 +* \company Eurecom +* \email: +* \note +* \warning +*/ + +#include "PHY/defs_gNB.h" +#include "PHY/phy_extern.h" +#include "PHY/CODING/coding_extern.h" +#include "PHY/CODING/coding_defs.h" +#include "PHY/CODING/lte_interleaver_inline.h" +#include "PHY/CODING/nrLDPC_encoder/defs.h" +#include "PHY/NR_TRANSPORT/nr_transport.h" +#include "PHY/NR_TRANSPORT/nr_transport_common_proto.h" +#include "PHY/NR_TRANSPORT/nr_dlsch.h" +#include "SCHED_NR/sched_nr.h" +#include "defs.h" +#include "common/utils/LOG/vcd_signal_dumper.h" +#include "common/utils/LOG/log.h" +#include + +//#define DEBUG_DLSCH_CODING +//#define DEBUG_DLSCH_FREE 1 + +void free_gNB_dlsch(NR_gNB_DLSCH_t *dlsch) +{ + int i; + int r; + + if (dlsch) { +#ifdef DEBUG_DLSCH_FREE + printf("Freeing dlsch %p\n",dlsch); +#endif + + for (i=0; iMdlharq; i++) { +#ifdef DEBUG_DLSCH_FREE + printf("Freeing dlsch process %d\n",i); +#endif + + if (dlsch->harq_processes[i]) { +#ifdef DEBUG_DLSCH_FREE + printf("Freeing dlsch process %d (%p)\n",i,dlsch->harq_processes[i]); +#endif + + if (dlsch->harq_processes[i]->b) { + free16(dlsch->harq_processes[i]->b,MAX_DLSCH_PAYLOAD_BYTES); + dlsch->harq_processes[i]->b = NULL; +#ifdef DEBUG_DLSCH_FREE + printf("Freeing dlsch process %d b (%p)\n",i,dlsch->harq_processes[i]->b); +#endif + } + +#ifdef DEBUG_DLSCH_FREE + printf("Freeing dlsch process %d c (%p)\n",i,dlsch->harq_processes[i]->c); +#endif + + for (r=0; rharq_processes[i]->c[r]); +#endif + + if (dlsch->harq_processes[i]->c[r]) { + free16(dlsch->harq_processes[i]->c[r],1056); + dlsch->harq_processes[i]->c[r] = NULL; + } + if (dlsch->harq_processes[i]->d[r]) { + free16(dlsch->harq_processes[i]->d[r],3*8448); + dlsch->harq_processes[i]->d[r] = NULL; + } + + } + free16(dlsch->harq_processes[i],sizeof(NR_DL_gNB_HARQ_t)); + dlsch->harq_processes[i] = NULL; + } + } + + free16(dlsch,sizeof(NR_gNB_DLSCH_t)); + dlsch = NULL; + } + +} + +NR_gNB_DLSCH_t *new_gNB_dlsch(unsigned char Kmimo, + unsigned char Mdlharq, + uint32_t Nsoft, + uint8_t abstraction_flag, + NR_DL_FRAME_PARMS *frame_parms, + nfapi_nr_config_request_t *config) +{ + + NR_gNB_DLSCH_t *dlsch; + unsigned char exit_flag = 0,i,r,aa,layer; + int re; + unsigned char bw_scaling =1; + uint16_t N_RB = config->rf_config.dl_carrier_bandwidth.value; + + switch (N_RB) { + + case 106: + bw_scaling =2; + break; + + default: + bw_scaling =1; + break; + } + + dlsch = (NR_gNB_DLSCH_t *)malloc16(sizeof(NR_gNB_DLSCH_t)); + + if (dlsch) { + bzero(dlsch,sizeof(NR_gNB_DLSCH_t)); + dlsch->Kmimo = Kmimo; + dlsch->Mdlharq = Mdlharq; + dlsch->Mlimit = 4; + dlsch->Nsoft = Nsoft; + + for (layer=0; layerue_spec_bf_weights[layer] = (int32_t**)malloc16(64*sizeof(int32_t*)); + + for (aa=0; aa<64; aa++) { + dlsch->ue_spec_bf_weights[layer][aa] = (int32_t *)malloc16(OFDM_SYMBOL_SIZE_COMPLEX_SAMPLES*sizeof(int32_t)); + for (re=0;reue_spec_bf_weights[layer][aa][re] = 0x00007fff; + } + } + + dlsch->txdataF[layer] = (int32_t *)malloc16((NR_MAX_PDSCH_ENCODED_LENGTH>>1)*sizeof(int32_t*)); + } + + for (int q=0; qmod_symbs[q] = (int32_t *)malloc16((NR_MAX_PDSCH_ENCODED_LENGTH>>1)*sizeof(int32_t*)); + + dlsch->calib_dl_ch_estimates = (int32_t**)malloc16(64*sizeof(int32_t*)); + for (aa=0; aa<64; aa++) { + dlsch->calib_dl_ch_estimates[aa] = (int32_t *)malloc16(OFDM_SYMBOL_SIZE_COMPLEX_SAMPLES*sizeof(int32_t)); + + } + + for (i=0; i<10; i++) + dlsch->harq_ids[i] = 0; + + + for (i=0; iharq_processes[i] = (NR_DL_gNB_HARQ_t *)malloc16(sizeof(NR_DL_gNB_HARQ_t)); + LOG_T(PHY, "Required mem size %d (bw scaling %d), dlsch->harq_processes[%d] %p\n", + MAX_NR_DLSCH_PAYLOAD_BYTES/bw_scaling,bw_scaling, i,dlsch->harq_processes[i]); + + if (dlsch->harq_processes[i]) { + bzero(dlsch->harq_processes[i],sizeof(NR_DL_gNB_HARQ_t)); + // dlsch->harq_processes[i]->first_tx=1; + dlsch->harq_processes[i]->b = (unsigned char*)malloc16(MAX_NR_DLSCH_PAYLOAD_BYTES/bw_scaling); + dlsch->harq_processes[i]->pdu = (uint8_t*)malloc16(MAX_NR_DLSCH_PAYLOAD_BYTES/bw_scaling); + if (dlsch->harq_processes[i]->pdu) { + bzero(dlsch->harq_processes[i]->pdu,MAX_NR_DLSCH_PAYLOAD_BYTES/bw_scaling); + nr_emulate_dlsch_payload(dlsch->harq_processes[i]->pdu, (MAX_NR_DLSCH_PAYLOAD_BYTES/bw_scaling)>>3); + } else { + printf("Can't allocate PDU\n"); + exit_flag=1; + } + + if (dlsch->harq_processes[i]->b) { + bzero(dlsch->harq_processes[i]->b,MAX_NR_DLSCH_PAYLOAD_BYTES/bw_scaling); + } else { + printf("Can't get b\n"); + exit_flag=1; + } + + if (abstraction_flag==0) { + for (r=0; rharq_processes[i]->c[r] = (uint8_t*)malloc16(8448); + dlsch->harq_processes[i]->d[r] = (uint8_t*)malloc16(68*384); + if (dlsch->harq_processes[i]->c[r]) { + bzero(dlsch->harq_processes[i]->c[r],8448); + } else { + printf("Can't get c\n"); + exit_flag=2; + } + if (dlsch->harq_processes[i]->d[r]) { + bzero(dlsch->harq_processes[i]->d[r],(3*8448)); + } else { + printf("Can't get d\n"); + exit_flag=2; + } + } + } + } else { + printf("Can't get harq_p %d\n",i); + exit_flag=3; + } + } + + if (exit_flag==0) { + for (i=0; iharq_processes[i]->round=0; + } + + return(dlsch); + } + } + + LOG_D(PHY,"new_gNB_dlsch exit flag %d, size of %ld\n", + exit_flag, sizeof(NR_gNB_DLSCH_t)); + free_gNB_dlsch(dlsch); + return(NULL); + + +} + +void clean_gNB_dlsch(NR_gNB_DLSCH_t *dlsch) +{ + + unsigned char Mdlharq; + unsigned char i,j,r; + + if (dlsch) { + Mdlharq = dlsch->Mdlharq; + dlsch->rnti = 0; + dlsch->active = 0; + + for (i=0; i<10; i++) + dlsch->harq_ids[i] = Mdlharq; + + for (i=0; iharq_processes[i]) { + // dlsch->harq_processes[i]->Ndi = 0; + //dlsch->harq_processes[i]->status = 0; + dlsch->harq_processes[i]->round = 0; + + for (j=0; j<96; j++) + for (r=0; rharq_processes[i]->d[r]) + dlsch->harq_processes[i]->d[r][j] = NR_NULL; + + } + } + } +} + +int nr_dlsch_encoding(unsigned char *a, + uint8_t slot, + NR_gNB_DLSCH_t *dlsch, + NR_DL_FRAME_PARMS* frame_parms) +{ + + unsigned int G; + unsigned int crc=1; + + uint8_t harq_pid = dlsch->harq_ids[slot]; + nfapi_nr_dl_config_dlsch_pdu_rel15_t rel15 = dlsch->harq_processes[harq_pid]->dlsch_pdu.dlsch_pdu_rel15; + uint16_t nb_rb = rel15.n_prb; + uint8_t nb_symb_sch = rel15.nb_symbols; + uint32_t A, Z; + uint32_t *pz = &Z; + uint8_t mod_order = rel15.modulation_order; + uint16_t Kr=0,r,r_offset=0;//Kr_bytes + uint8_t *d_tmp[MAX_NUM_DLSCH_SEGMENTS]; + uint8_t kb,BG=1; + uint32_t E; + uint8_t Ilbrm = 0; + uint32_t Tbslbrm = 950984; //max tbs + uint8_t nb_re_dmrs = rel15.nb_re_dmrs; + uint16_t length_dmrs = 1; + + /* + uint8_t *channel_input[MAX_NUM_DLSCH_SEGMENTS]; //unsigned char + for(j=0;jsymbols_per_slot,0,0, rel15.nb_layers); + + // if (dlsch->harq_processes[harq_pid]->Ndi == 1) { // this is a new packet + if (dlsch->harq_processes[harq_pid]->round == 0) { // this is a new packet +#ifdef DEBUG_DLSCH_CODING + printf("encoding thinks this is a new packet \n"); +#endif + /* + int i; + printf("dlsch (tx): \n"); + for (i=0;i<(A>>3);i++) + printf("%02x.",a[i]); + printf("\n"); + */ + // Add 24-bit crc (polynomial A) to payload + crc = crc24a(a,A)>>8; + a[A>>3] = ((uint8_t*)&crc)[2]; + a[1+(A>>3)] = ((uint8_t*)&crc)[1]; + a[2+(A>>3)] = ((uint8_t*)&crc)[0]; + //printf("CRC %x (A %d)\n",crc,A); + //printf("a0 %d a1 %d a2 %d\n", a[A>>3], a[1+(A>>3)], a[2+(A>>3)]); + + dlsch->harq_processes[harq_pid]->B = A+24; + // dlsch->harq_processes[harq_pid]->b = a; + memcpy(dlsch->harq_processes[harq_pid]->b,a,(A/8)+4); + + nr_segmentation(dlsch->harq_processes[harq_pid]->b, + dlsch->harq_processes[harq_pid]->c, + dlsch->harq_processes[harq_pid]->B, + &dlsch->harq_processes[harq_pid]->C, + &dlsch->harq_processes[harq_pid]->K, + pz, + &dlsch->harq_processes[harq_pid]->F); + + kb = dlsch->harq_processes[harq_pid]->K/(*pz); + if ( kb==22){ + BG = 1; + } + else{ + BG = 2; + } + + Kr = dlsch->harq_processes[harq_pid]->K; + //Kr_bytes = Kr>>3; + + //printf("segment Z %d kb %d k %d Kr %d BG %d\n", *pz,kb,dlsch->harq_processes[harq_pid]->K,Kr,BG); + + //start_meas(te_stats); + for (r=0; rharq_processes[harq_pid]->C; r++) { + d_tmp[r] = &dlsch->harq_processes[harq_pid]->d[r][0]; + //channel_input[r] = &dlsch->harq_processes[harq_pid]->d[r][0]; +#ifdef DEBUG_DLSCH_CODING + printf("Encoder: B %d F %d \n",dlsch->harq_processes[harq_pid]->B, dlsch->harq_processes[harq_pid]->F); + printf("start ldpc encoder segment %d/%d\n",r,dlsch->harq_processes[harq_pid]->C); + printf("input %d %d %d %d %d \n", dlsch->harq_processes[harq_pid]->c[r][0], dlsch->harq_processes[harq_pid]->c[r][1], dlsch->harq_processes[harq_pid]->c[r][2],dlsch->harq_processes[harq_pid]->c[r][3], dlsch->harq_processes[harq_pid]->c[r][4]); + for (int cnt =0 ; cnt < 22*(*pz)/8; cnt ++){ + printf("%d ", dlsch->harq_processes[harq_pid]->c[r][cnt]); + } + printf("\n"); + +#endif + //ldpc_encoder_orig((unsigned char*)dlsch->harq_processes[harq_pid]->c[r],dlsch->harq_processes[harq_pid]->d[r],Kr,BG,0); + //ldpc_encoder_optim((unsigned char*)dlsch->harq_processes[harq_pid]->c[r],(unsigned char*)&dlsch->harq_processes[harq_pid]->d[r][0],Kr,BG,NULL,NULL,NULL,NULL); + } + + //for (int i=0;i<68*384;i++) + // printf("channel_input[%d]=%d\n",i,channel_input[i]); + + + + /*printf("output %d %d %d %d %d \n", dlsch->harq_processes[harq_pid]->d[0][0], dlsch->harq_processes[harq_pid]->d[0][1], dlsch->harq_processes[harq_pid]->d[r][2],dlsch->harq_processes[harq_pid]->d[0][3], dlsch->harq_processes[harq_pid]->d[0][4]); + for (int cnt =0 ; cnt < 66*(*pz); cnt ++){ + printf("%d \n", dlsch->harq_processes[harq_pid]->d[0][cnt]); + } + printf("\n");*/ + + ldpc_encoder_optim_8seg(dlsch->harq_processes[harq_pid]->c,d_tmp,Kr,BG,dlsch->harq_processes[harq_pid]->C,NULL,NULL,NULL,NULL); + + //stop_meas(te_stats); + //printf("end ldpc encoder -- output\n"); + +#ifdef DEBUG_DLSCH_CODING + write_output("enc_input0.m","enc_in0",&dlsch->harq_processes[harq_pid]->c[0][0],Kr_bytes,1,4); + write_output("enc_output0.m","enc0",&dlsch->harq_processes[harq_pid]->d[0][0],(3*8*Kr_bytes)+12,1,4); +#endif + + } + + for (r=0; rharq_processes[harq_pid]->C; r++) { +#ifdef DEBUG_DLSCH_CODING + printf("Rate Matching, Code segment %d (coded bits (G) %d,unpunctured/repeated bits per code segment %d,mod_order %d, nb_rb %d)...\n", + r, + G, + Kr*3, + mod_order,nb_rb); +#endif + + //start_meas(rm_stats); +#ifdef DEBUG_DLSCH_CODING + printf("rvidx in encoding = %d\n", dlsch->harq_processes[harq_pid]->rvidx); +#endif + + E = nr_rate_matching_ldpc(Ilbrm, + Tbslbrm, + BG, + *pz, + G, + dlsch->harq_processes[harq_pid]->d[r], + dlsch->harq_processes[harq_pid]->e+r_offset, + dlsch->harq_processes[harq_pid]->C, + rel15.redundancy_version, + mod_order, + rel15.nb_layers, + r); + +#ifdef DEBUG_DLSCH_CODING + for (int i =0; i<16; i++) + printf("output ratematching e[%d]= %d r_offset %d\n", i,dlsch->harq_processes[harq_pid]->e[i], r_offset); +#endif + //stop_meas(rm_stats); + + //start_meas(i_stats); + nr_interleaving_ldpc(E, + mod_order, + dlsch->harq_processes[harq_pid]->e+r_offset, + dlsch->harq_processes[harq_pid]->f+r_offset); + //stop_meas(i_stats); + + r_offset += E; +#ifdef DEBUG_DLSCH_CODING + for (int i =0; i<16; i++) + printf("output interleaving f[%d]= %d r_offset %d\n", i,dlsch->harq_processes[harq_pid]->f[i+r*r_offset], r_offset); +#endif + +#ifdef DEBUG_DLSCH_CODING + + if (r==dlsch->harq_processes[harq_pid]->C-1) + write_output("enc_output.m","enc",dlsch->harq_processes[harq_pid]->f,r_offset,1,4); + +#endif + } + + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_ENB_DLSCH_ENCODING, VCD_FUNCTION_OUT); + + return(0); +} diff --git a/openair1/PHY/NR_TRANSPORT/nr_dlsch_tools.c b/openair1/PHY/NR_TRANSPORT/nr_dlsch_tools.c new file mode 100644 index 0000000000..c1ac34075a --- /dev/null +++ b/openair1/PHY/NR_TRANSPORT/nr_dlsch_tools.c @@ -0,0 +1,250 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.1 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +/*! \file PHY/NR_TRANSPORT/nr_dlsch_tools.c +* \brief Top-level routines for decoding DLSCH transport channel from 38-214, V15.2.0 2018-06 +* \author Guy De Souza +* \date 2018 +* \version 0.1 +* \company Eurecom +* \email: desouza@eurecom.fr +* \note +* \warning +*/ + +#include "nr_dlsch.h" + +extern void set_taus_seed(unsigned int seed_type); + +uint8_t nr_pdsch_default_time_alloc_A_S_nCP[23] = {2,3,2,3,2,3,2,3,2,3,9,10,4,6,5,5,9,12,1,1,2,4,8}; +uint8_t nr_pdsch_default_time_alloc_A_L_nCP[23] = {12,11,10,9,9,8,7,6,5,4,4,4,4,4,7,2,2,2,13,6,4,7,4}; +uint8_t nr_pdsch_default_time_alloc_A_S_eCP[23] = {2,3,2,3,2,3,2,3,2,3,6,8,4,6,5,5,9,10,1,1,2,4,8}; +uint8_t nr_pdsch_default_time_alloc_A_L_eCP[23] = {6,5,10,9,9,8,7,6,5,4,4,2,4,4,6,2,2,2,11,6,4,6,4}; +uint8_t nr_pdsch_default_time_alloc_B_S[16] = {2,4,6,8,10,2,4,2,4,6,8,10,2,2,3,2}; +uint8_t nr_pdsch_default_time_alloc_B_L[16] = {2,2,2,2,2,2,2,4,4,4,4,4,7,12,11,4}; +uint8_t nr_pdsch_default_time_alloc_C_S[15] = {2,4,6,8,10,2,4,6,8,10,2,2,3,0,2}; +uint8_t nr_pdsch_default_time_alloc_C_L[15] = {2,2,2,2,2,4,4,4,4,4,7,12,11,6,6}; + + + /// Time domain allocation routines + +void nr_get_time_domain_allocation_type(nfapi_nr_config_request_t config, + nfapi_nr_dl_config_dci_dl_pdu dci_pdu, + nfapi_nr_dl_config_dlsch_pdu *dlsch_pdu) { + + nfapi_nr_dl_config_pdcch_parameters_rel15_t params_rel15 = dci_pdu.pdcch_params_rel15; + uint8_t *alloc_type = &dlsch_pdu->dlsch_pdu_rel15.time_allocation_type; + uint8_t mux_pattern = params_rel15.mux_pattern; + uint8_t alloc_list_flag = dlsch_pdu->dlsch_pdu_rel15.time_alloc_list_flag; + + switch(params_rel15.rnti_type) { + + case NFAPI_NR_RNTI_SI: + AssertFatal(params_rel15.common_search_space_type == NFAPI_NR_COMMON_SEARCH_SPACE_TYPE_0, + "Invalid common search space type %d for SI RNTI, expected %d\n", + params_rel15.common_search_space_type, NFAPI_NR_COMMON_SEARCH_SPACE_TYPE_0); + + if (mux_pattern == NFAPI_NR_SSB_AND_CSET_MUX_PATTERN_TYPE1) + AssertFatal(config.subframe_config.dl_cyclic_prefix_type.value == NFAPI_CP_NORMAL, + "Invalid configuration CP extended for SI RNTI type 0 search space\n"); + + *alloc_type = (mux_pattern == NFAPI_NR_SSB_AND_CSET_MUX_PATTERN_TYPE1)?NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_A : + (mux_pattern == NFAPI_NR_SSB_AND_CSET_MUX_PATTERN_TYPE2)?NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_B : + NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_C; + break; + + case NFAPI_NR_RNTI_RA: + case NFAPI_NR_RNTI_TC: + /*AssertFatal(dci_alloc.pdcch_params.common_search_space_type == NFAPI_NR_COMMON_SEARCH_SPACE_TYPE_1, + "Invalid common search space type %d for RNTI %d, expected %d\n",dci_alloc.pdcch_params.common_search_space_type, + NFAPI_NR_COMMON_SEARCH_SPACE_TYPE_1, dci_alloc.rnti_type);*/ + *alloc_type = (alloc_list_flag) ? NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_ALLOC_LIST : NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_A; + break; + + case NFAPI_NR_RNTI_P: + break; + + case NFAPI_NR_RNTI_C: + case NFAPI_NR_RNTI_CS: + if (params_rel15.search_space_type == NFAPI_NR_SEARCH_SPACE_TYPE_COMMON) + *alloc_type = (alloc_list_flag)? NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_ALLOC_LIST : NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_A; + else + *alloc_type = NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_ALLOC_LIST; + + break; + + } +} + +uint16_t get_SLIV(uint8_t S, uint8_t L) { + return ( (uint16_t)(((L-1)<=7)? (14*(L-1)+S) : (14*(15-L)+(13-S))) ); +} + +static inline uint8_t get_K0(uint8_t row_idx, uint8_t time_alloc_type) { + return ( (time_alloc_type == NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_A)|| + (time_alloc_type == NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_C)? 0 : + ((row_idx==6)||(row_idx==7)||(row_idx==15))? 1 : 0); +} + +/*ideally combine the calculation of L in the same function once the right struct is defined*/ +uint8_t nr_get_S(uint8_t row_idx, uint8_t CP, uint8_t time_alloc_type, uint8_t dmrs_typeA_position) { + + uint8_t idx; + //uint8_t S; + + switch(time_alloc_type) { + case NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_A: + idx = (row_idx>7)? (row_idx+6) : (((row_idx-1)<<1)-1+((dmrs_typeA_position==2)?0:1)); + return ((CP==NFAPI_CP_NORMAL)?nr_pdsch_default_time_alloc_A_S_nCP[idx] : nr_pdsch_default_time_alloc_A_S_eCP[idx]); + break; + + case NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_B: + idx = (row_idx<14)? (row_idx-1) : (row_idx == 14)? row_idx-1+((dmrs_typeA_position==2)?0:1) : 15; + return (nr_pdsch_default_time_alloc_B_S[idx]); + break; + + case NFAPI_NR_PDSCH_TIME_DOMAIN_ALLOC_TYPE_DEFAULT_C: + AssertFatal((row_idx!=6)&&(row_idx!=7)&&(row_idx<17), "Invalid row index %d in %s %s\n", row_idx, __FUNCTION__, __FILE__); + idx = (row_idx<6)? (row_idx-1) : (row_idx<14)? (row_idx-3) : (row_idx == 14)? row_idx-3+((dmrs_typeA_position==2)?0:1) : (row_idx-2); + break; + + default: + AssertFatal(0, "Invalid Time domain allocation type %d in %s %s\n", time_alloc_type, __FUNCTION__, __FILE__); + } + return 0; // temp warning fix +} + +void nr_check_time_alloc(uint8_t S, uint8_t L, nfapi_nr_config_request_t config) { + + switch (config.subframe_config.dl_cyclic_prefix_type.value) { + case NFAPI_CP_NORMAL: + if (config.pdsch_config.mapping_type.value == NFAPI_NR_PDSCH_MAPPING_TYPE_A) { + AssertFatal(S<4, "Invalid value of S(%d) for mapping type A and normal CP\n", S); + + if (S==3) + AssertFatal(config.pdsch_config.mapping_type.value == 3, "Invalid S %d for dmrs_typeA_position %d\n", + S, config.pdsch_config.dmrs_typeA_position.value); + + AssertFatal((L>2)&&(L<15), "Invalid L %d for mapping type A and normal CP\n", L); + + AssertFatal(((S+L)>2)&&((S+L)<15), "Invalid S+L %d for mapping type A and normal CP\n", S+L); + } + else { + AssertFatal(S<13, "Invalid value of S(%d) for mapping type B and normal CP\n", S); + + AssertFatal((L>1)&&(L<8), "Invalid L %d for mapping type B and normal CP\n", L); + + AssertFatal(((S+L)>1)&&((S+L)<15), "Invalid S+L %d for mapping type B and normal CP\n", S+L); + } + break; + + case NFAPI_CP_EXTENDED: + if (config.pdsch_config.mapping_type.value == NFAPI_NR_PDSCH_MAPPING_TYPE_A) { + AssertFatal(S<4, "Invalid value of S(%d) for mapping type A and extended CP\n", S); + + if (S==3) + AssertFatal(config.pdsch_config.dmrs_typeA_position.value == 3, "Invalid S %d for dmrs_typeA_position %d\n", + S, config.pdsch_config.dmrs_typeA_position.value); + + AssertFatal((L>2)&&(L<13), "Invalid L %d for mapping type A and extended CP\n", L); + + AssertFatal(((S+L)>2)&&((S+L)<13), "Invalid S+L %d for mapping type A and extended CP\n", S+L); + } + else { + AssertFatal(S<11, "Invalid value of S(%d) for mapping type B and extended CP\n", S); + + AssertFatal((L>1)&&(L<7), "Invalid L %d for mapping type B and extended CP\n", L); + + AssertFatal(((S+L)>1)&&((S+L)<13), "Invalid S+L %d for mapping type B and extended CP\n", S+L); + } + break; + } +} + + + /// Frequency domain allocation routines + + // DL alloc type 0 +static inline uint8_t get_RBG_size_P(uint16_t n_RB, uint8_t RBG_config) { + if (RBG_config == NFAPI_NR_PDSCH_RBG_CONFIG_TYPE1) + return ((n_RB<37)?2:(n_RB<73)?4:(n_RB<145)?8:16); + else if (RBG_config == NFAPI_NR_PDSCH_RBG_CONFIG_TYPE2) + return ((n_RB<37)?4:(n_RB<73)?8:(n_RB<145)?16:16); + else + AssertFatal(0, "Invalid RBG config type (%d)\n", RBG_config); +} + +void nr_get_rbg_parms(NR_BWP_PARMS* bwp, uint8_t config_type) { + nr_rbg_parms_t* rbg_parms = &bwp->rbg_parms; + + rbg_parms->P = get_RBG_size_P(bwp->N_RB, config_type); + rbg_parms->start_size = rbg_parms->P - bwp->location%rbg_parms->P; + rbg_parms->end_size = ((bwp->location + bwp->N_RB)%rbg_parms->P)? ((bwp->location + bwp->N_RB)%rbg_parms->P) : rbg_parms->P; + rbg_parms->N_RBG = (uint8_t)ceil( (bwp->N_RB + (bwp->location%rbg_parms->P))/(float)rbg_parms->P); + LOG_I(PHY, "RBG parameters for BWP %d location %d N_RB %d:\n", bwp->bwp_id, bwp->location, bwp->N_RB); + LOG_I(PHY, "P %d\t start size %d\t endsize %d\t N_RBG %d\n", rbg_parms->P, rbg_parms->start_size, rbg_parms->end_size, rbg_parms->N_RBG); +} + +void nr_get_rbg_list(uint32_t bitmap, uint8_t n_rbg, uint8_t* rbg_list) { + uint8_t idx=0; + for (int i=0; i>(n_rbg-i-1))&1) + rbg_list[idx++]=i; +} + + // DL alloc type 1 + +uint16_t get_RIV(uint16_t rb_start, uint16_t L, uint16_t N_RB) { + if ((L-1)<=(N_RB>>1)) + return (N_RB*(L-1)+rb_start); + else + return (N_RB*(N_RB-L+1) + (N_RB-1-rb_start)); +} + + /// PRB bundling routines + +// Precoding granularity +static inline uint8_t nr_get_P_prime(uint8_t rnti_type, uint8_t dci_format, uint8_t prb_bundling_type) { + if (dci_format == NFAPI_NR_DL_DCI_FORMAT_1_0) + return (NFAPI_NR_PRG_GRANULARITY_2); + else // NFAPI_NR_DL_DCI_FORMAT_1_1 + return ((prb_bundling_type)?0:2);// incomplete for 1_1 +} + +void nr_get_PRG_parms(NR_BWP_PARMS* bwp, NR_gNB_DCI_ALLOC_t dci_alloc, uint8_t prb_bundling_type) { + nr_prg_parms_t* prg_parms = &bwp->prg_parms; + + prg_parms->P_prime = nr_get_P_prime(dci_alloc.pdcch_params.rnti_type, dci_alloc.pdcch_params.dci_format, prb_bundling_type); + prg_parms->start_size = prg_parms->P_prime - bwp->location%prg_parms->P_prime; + prg_parms->end_size = (bwp->location + bwp->N_RB)%prg_parms->P_prime; + if (!prg_parms->end_size) + prg_parms->end_size = prg_parms->P_prime; + prg_parms->N_PRG = ceil((float)bwp->N_RB/prg_parms->P_prime); + LOG_I(PHY, "PRG parameters for BWP %d location %d N_RB %d:\n", bwp->bwp_id, bwp->location, bwp->N_RB); + LOG_I(PHY, "P_prime %d\t start size %d\t endsize %d\t N_PRG %d\n", prg_parms->P_prime, prg_parms->start_size, prg_parms->end_size, prg_parms->N_PRG); +} + + /// Payload emulation +void nr_emulate_dlsch_payload(uint8_t* pdu, uint16_t size) { + set_taus_seed(0); + for (int i=0; isamples_per_frame_wCP>>1, 1, 1); -#endif - - return 0; } diff --git a/openair1/PHY/NR_TRANSPORT/nr_sch_dmrs.c b/openair1/PHY/NR_TRANSPORT/nr_sch_dmrs.c new file mode 100644 index 0000000000..4ec1083cfc --- /dev/null +++ b/openair1/PHY/NR_TRANSPORT/nr_sch_dmrs.c @@ -0,0 +1,89 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.1 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +/*! \file PHY/NR_TRANSPORT/nr_sch_dmrs.c +* \brief +* \author +* \date +* \version +* \company Eurecom +* \email: +* \note +* \warning +*/ + +#include "nr_sch_dmrs.h" + + +/*Table 7.4.1.1.2-1 and 7.4.1.1.2-2 38211 Columns: ap - CDM group - Delta - Wf(0) - Wf(1) - Wt(0) - Wt(1)*/ +int8_t pdsch_dmrs_1[8][7] = {{0,0,0,1,1,1,1}, + {1,0,0,1,-1,1,1}, + {2,1,1,1,1,1,1}, + {3,1,1,1,-1,1,1}, + {4,0,0,1,1,1,-1}, + {5,0,0,1,-1,1,-1}, + {6,1,1,1,1,1,-1}, + {7,1,1,1,-1,1,-1}}; + +int8_t pdsch_dmrs_2[12][7] = {{0,0,0,1,1,1,1}, + {1,0,0,1,-1,1,1}, + {2,1,2,1,1,1,1}, + {3,1,2,1,-1,1,1}, + {4,2,4,1,1,1,1}, + {5,2,4,1,-1,1,1}, + {6,0,0,1,1,1,-1}, + {7,0,0,1,-1,1,-1}, + {8,1,2,1,1,1,-1}, + {9,1,2,1,-1,1,-1}, + {10,2,4,1,1,1,-1}, + {11,2,4,1,-1,1,-1}}; + +void get_l_prime(uint8_t *l_prime, uint8_t n_symbs) { + for (int i=0; idlsch_pdu_rel15; + uint8_t rnti_type = params_rel15.rnti_type; + uint8_t dci_format = params_rel15.dci_format; + uint8_t ss_type = params_rel15.search_space_type; + uint8_t N_PRB_oh = ((rnti_type==NFAPI_NR_RNTI_SI)||(rnti_type==NFAPI_NR_RNTI_RA)||(rnti_type==NFAPI_NR_RNTI_P))? 0 : \ + (config.pdsch_config.x_overhead.value); + uint8_t N_PRB_DMRS = (config.pdsch_config.dmrs_type.value == NFAPI_NR_DMRS_TYPE1)?6:4; //This only works for antenna port 1000 + uint8_t mcs_table = config.pdsch_config.mcs_table.value; + uint8_t N_sh_symb = dlsch_rel15->nb_symbols; + uint8_t Imcs = dlsch_rel15->mcs_idx; + uint16_t N_RE_prime = NR_NB_SC_PER_RB*N_sh_symb - N_PRB_DMRS - N_PRB_oh; + LOG_I(MAC, "N_RE_prime %d for %d symbols %d DMRS per PRB and %d overhead\n", N_RE_prime, N_sh_symb, N_PRB_DMRS, N_PRB_oh); + + uint16_t N_RE, Ninfo, Ninfo_prime, C, TBS=0, R; + uint8_t table_idx, Qm, n, scale; + + table_idx = get_table_idx(mcs_table, dci_format, rnti_type, ss_type); + scale = ((table_idx==2)&&((Imcs==20)||(Imcs==26)))?11:10; + + N_RE = min(156, N_RE_prime)*dlsch_rel15->n_prb; + R = nr_get_code_rate(Imcs, table_idx); + Qm = nr_get_Qm(Imcs, table_idx); + Ninfo = (N_RE*R*Qm*dlsch_rel15->nb_layers)>>scale; + + if (Ninfo <= 3824) { + n = max(3, (log2(Ninfo)-6)); + Ninfo_prime = max(24, (Ninfo>>n)<= Ninfo_prime) { + TBS = nr_tbs_table[i]; + break; + } + } + else { + n = log2(Ninfo-24)-5; + Ninfo_prime = max(3840, (ROUNDIDIV((Ninfo-24),(1<8424) { + C = CEILIDIV((Ninfo_prime+24),8424); + TBS = (C<<3)*CEILIDIV((Ninfo_prime+24),(C<<3)) - 24; + } + else + TBS = ((CEILIDIV((Ninfo_prime+24),8))<<3) - 24; + } + } + + dlsch_rel15->coding_rate = R; + dlsch_rel15->modulation_order = Qm; + dlsch_rel15->transport_block_size = TBS; + dlsch_rel15->nb_re_dmrs = N_PRB_DMRS; + dlsch_rel15->nb_mod_symbols = N_RE_prime*dlsch_rel15->n_prb*dlsch_rel15->nb_codewords; + + LOG_I(MAC, "TBS %d : N_RE %d N_PRB_DMRS %d N_sh_symb %d N_PRB_oh %d Ninfo %d Ninfo_prime %d R %d Qm %d table %d scale %d nb_symbols %d\n", + TBS, N_RE, N_PRB_DMRS, N_sh_symb, N_PRB_oh, Ninfo, Ninfo_prime, R, Qm, table_idx, scale, dlsch_rel15->nb_mod_symbols); +} + +uint32_t nr_get_G(uint16_t nb_rb, uint16_t nb_symb_sch,uint8_t nb_re_dmrs,uint16_t length_dmrs, uint8_t Qm, uint8_t Nl) { + uint32_t G; + G = ((NR_NB_SC_PER_RB*nb_symb_sch)-(nb_re_dmrs*length_dmrs))*nb_rb*Qm*Nl; + return(G); +} diff --git a/openair1/PHY/NR_TRANSPORT/nr_transport.h b/openair1/PHY/NR_TRANSPORT/nr_transport.h index 12e1b7b3aa..c1a4052191 100644 --- a/openair1/PHY/NR_TRANSPORT/nr_transport.h +++ b/openair1/PHY/NR_TRANSPORT/nr_transport.h @@ -106,4 +106,11 @@ int nr_generate_pbch(NR_gNB_PBCH *pbch, */ void nr_init_pbch_interleaver(uint8_t *interleaver); +NR_gNB_DLSCH_t *new_gNB_dlsch(unsigned char Kmimo, + unsigned char Mdlharq, + uint32_t Nsoft, + uint8_t abstraction_flag, + NR_DL_FRAME_PARMS *frame_parms, + nfapi_nr_config_request_t *config); + #endif /*__NR_TRANSPORT__H__*/ diff --git a/openair1/PHY/NR_TRANSPORT/nr_transport_common_proto.h b/openair1/PHY/NR_TRANSPORT/nr_transport_common_proto.h index b014748bca..3c318874ac 100644 --- a/openair1/PHY/NR_TRANSPORT/nr_transport_common_proto.h +++ b/openair1/PHY/NR_TRANSPORT/nr_transport_common_proto.h @@ -33,6 +33,15 @@ #ifndef __NR_TRANSPORT_COMMON_PROTO__H__ #define __NR_TRANSPORT_COMMON_PROTO__H__ +#define MAX_NUM_NR_DLSCH_SEGMENTS 32 +#define MAX_NUM_NR_ULSCH_SEGMENTS MAX_NUM_NR_DLSCH_SEGMENTS + +#define MAX_NR_DLSCH_PAYLOAD_BYTES (MAX_NUM_NR_DLSCH_SEGMENTS*1056) +#define MAX_NR_ULSCH_PAYLOAD_BYTES (MAX_NUM_NR_DLSCH_SEGMENTS*1056) + +#define MAX_NUM_NR_CHANNEL_BITS (14*273*12*6) // 14 symbols, 273 RB +#define MAX_NUM_NR_RE (14*273*12) + // Functions below implement minor procedures from 38-214 /** \brief Computes Q based on I_MCS PDSCH and when 'MCS-Table-PDSCH' is set to "256QAM". Implements Table 5.1.3.1-2 from 38.214. diff --git a/openair1/PHY/NR_UE_ESTIMATION/filt16a_32.h b/openair1/PHY/NR_UE_ESTIMATION/filt16a_32.h index 52356a9438..f94b18fb27 100644 --- a/openair1/PHY/NR_UE_ESTIMATION/filt16a_32.h +++ b/openair1/PHY/NR_UE_ESTIMATION/filt16a_32.h @@ -90,7 +90,7 @@ short filt8_l0[8] = { 16384,8192,0,0,0,0,0,0}; short filt8_mr0[8] = { -0,8192,16384,8192,0,-8192,0,0}; +0,0,0,8192,16384,8192,0,-8192}; short filt8_r0[8] = { 0,8192,16384,24576,0,0,0,0}; @@ -98,6 +98,9 @@ short filt8_r0[8] = { short filt8_m0[8] = { 0,8192,16384,8192,0,0,0,0}; +short filt8_mm0[8]= { +0,0,0,8192,16384,8192,0,0}; + short filt8_l1[8] = { 24576,16384,0,0,0,0,0,0}; @@ -109,3 +112,6 @@ short filt8_r1[8] = { short filt8_m1[8] = { 0,0,8192,16384,8192,0,0,0}; + +short filt8_mm1[8]= { +0,0,0,0,8192,16384,8192,0}; diff --git a/openair1/PHY/NR_UE_ESTIMATION/nr_adjust_synch_ue.c b/openair1/PHY/NR_UE_ESTIMATION/nr_adjust_synch_ue.c new file mode 100644 index 0000000000..f1a3ef42ab --- /dev/null +++ b/openair1/PHY/NR_UE_ESTIMATION/nr_adjust_synch_ue.c @@ -0,0 +1,140 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.1 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +#include "PHY/types.h" +#include "PHY/defs_nr_UE.h" +#include "PHY/NR_UE_ESTIMATION/nr_estimation.h" +#include "PHY/impl_defs_top.h" +//#include "openair2/LAYER2/MAC/mac_proto.h" + +#include "common/utils/LOG/vcd_signal_dumper.h" + +#define DEBUG_PHY + +// Adjust location synchronization point to account for drift +// The adjustment is performed once per frame based on the +// last channel estimate of the receiver + +void nr_adjust_synch_ue(NR_DL_FRAME_PARMS *frame_parms, + PHY_VARS_NR_UE *ue, + module_id_t eNB_id, + uint8_t subframe, + unsigned char clear, + short coef) +{ + + static int max_pos_fil = 0; + static int count_max_pos_ok = 0; + static int first_time = 1; + int temp = 0, i, aa, max_val = 0, max_pos = 0; + int diff; + short Re,Im,ncoef; + + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_ADJUST_SYNCH, VCD_FUNCTION_IN); + + ncoef = 32767 - coef; + +//#ifdef DEBUG_PHY + LOG_D(PHY,"AbsSubframe %d.%d: rx_offset (before) = %d\n",ue->proc.proc_rxtx[0].frame_rx%1024,subframe,ue->rx_offset); +//#endif //DEBUG_PHY + + + // we only use channel estimates from tx antenna 0 here + for (i = 0; i < frame_parms->nb_prefix_samples; i++) { + temp = 0; + + for (aa=0; aanb_antennas_rx; aa++) { + Re = ((int16_t*)ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[subframe]].dl_ch_estimates_time[eNB_id][aa])[(i<<1)]; + Im = ((int16_t*)ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[subframe]].dl_ch_estimates_time[eNB_id][aa])[1+(i<<1)]; + temp += (Re*Re/2) + (Im*Im/2); + } + + if (temp > max_val) { + max_pos = i; + max_val = temp; + } + } + + // filter position to reduce jitter + if (clear == 1) + max_pos_fil = max_pos; + else + max_pos_fil = ((max_pos_fil * coef) + (max_pos * ncoef)) >> 15; + + // do not filter to have proactive timing adjustment + //max_pos_fil = max_pos; + + if(subframe == 1) + { + diff = max_pos_fil - (frame_parms->nb_prefix_samples>>3); + + if ( abs(diff) < SYNCH_HYST ) + ue->rx_offset = 0; + else + ue->rx_offset = diff; + + if(abs(diff)<5) + count_max_pos_ok ++; + else + count_max_pos_ok = 0; + + //printf("adjust sync count_max_pos_ok = %d\n",count_max_pos_ok); + + if(count_max_pos_ok > 10 && first_time == 1) + { + first_time = 0; + ue->time_sync_cell = 1; + if (ue->mac_enabled==1) { + LOG_I(PHY,"[UE%d] Sending synch status to higher layers\n",ue->Mod_id); + //mac_resynch(); + //dl_phy_sync_success(ue->Mod_id,ue->proc.proc_rxtx[0].frame_rx,0,1);//ue->common_vars.eNb_id); + ue->UE_mode[0] = PRACH; + } + else { + ue->UE_mode[0] = PUSCH; + } + } + + if ( ue->rx_offset < 0 ) + ue->rx_offset += frame_parms->samples_per_frame; + + if ( ue->rx_offset >= frame_parms->samples_per_frame ) + ue->rx_offset -= frame_parms->samples_per_frame; + + + + #ifdef DEBUG_PHY + LOG_D(PHY,"AbsSubframe %d.%d: ThreadId %d diff =%i rx_offset (final) = %i : clear %d,max_pos = %d,max_pos_fil = %d (peak %d) max_val %d target_pos %d \n", + ue->proc.proc_rxtx[ue->current_thread_id[subframe]].frame_rx, + subframe, + ue->current_thread_id[subframe], + diff, + ue->rx_offset, + clear, + max_pos, + max_pos_fil, + temp,max_val, + (frame_parms->nb_prefix_samples>>3)); + #endif //DEBUG_PHY + + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_ADJUST_SYNCH, VCD_FUNCTION_OUT); + } +} diff --git a/openair1/PHY/NR_UE_ESTIMATION/nr_dl_channel_estimation.c b/openair1/PHY/NR_UE_ESTIMATION/nr_dl_channel_estimation.c index de97f0fb9e..942bf8c0ba 100644 --- a/openair1/PHY/NR_UE_ESTIMATION/nr_dl_channel_estimation.c +++ b/openair1/PHY/NR_UE_ESTIMATION/nr_dl_channel_estimation.c @@ -28,7 +28,7 @@ #include "PHY/NR_REFSIG/refsig_defs_ue.h" #include "filt16a_32.h" #include "T.h" -//#define DEBUG_CH +//#define DEBUG_PDSCH int nr_pbch_channel_estimation(PHY_VARS_NR_UE *ue, @@ -48,7 +48,7 @@ int nr_pbch_channel_estimation(PHY_VARS_NR_UE *ue, //uint16_t Nid_cell = (eNB_offset == 0) ? ue->frame_parms.Nid_cell : ue->measurements.adj_cell_id[eNB_offset-1]; - uint8_t nushift,ssb_index=0, n_hf=0; + uint8_t nushift, ssb_index=0, n_hf=0; int **dl_ch_estimates =ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[Ns>>1]].dl_ch_estimates[eNB_offset]; int **rxdataF=ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[Ns>>1]].rxdataF; @@ -262,7 +262,7 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, unsigned short nb_rb_coreset) { int pilot[200] __attribute__((aligned(16))); - unsigned char aarx; + unsigned char aarx,p; unsigned short k; unsigned int pilot_cnt; int16_t ch[2],*pil,*rxF,*dl_ch,*fl,*fm,*fr; @@ -286,7 +286,7 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, k = coreset_start_subcarrier; -#ifdef DEBUG_CH +#ifdef DEBUG_PDCCH printf("PDCCH Channel Estimation : ThreadId %d, eNB_offset %d ch_offset %d, OFDM size %d, Ncp=%d, l=%d, Ns=%d, k=%d symbol %d\n",ue->current_thread_id[Ns>>1], eNB_offset,ch_offset,ue->frame_parms.ofdm_symbol_size, ue->frame_parms.Ncp,l,Ns,k, symbol); #endif @@ -295,9 +295,11 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, fm = filt16a_m1; fr = filt16a_r1; + // generate pilot nr_pdcch_dmrs_rx(ue,eNB_offset,Ns,ue->nr_gold_pdcch[eNB_offset][Ns][symbol], &pilot[0],2000,nb_rb_coreset); + for (aarx=0; aarxframe_parms.nb_antennas_rx; aarx++) { pil = (int16_t *)&pilot[0]; @@ -309,18 +311,18 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, multadd_complex_vector_real_scalar(dl_ch-(ue->frame_parms.ofdm_symbol_size<<1), ue->ch_est_alpha,dl_ch-(ue->frame_parms.ofdm_symbol_size<<1), 1,ue->frame_parms.ofdm_symbol_size); - //#ifdef DEBUG_CH +#ifdef DEBUG_PDCCH printf("pdcch ch est pilot addr %p RB_DL %d\n",&pilot[0], ue->frame_parms.N_RB_DL); printf("k %d, first_carrier %d\n",k,ue->frame_parms.first_carrier_offset); printf("rxF addr %p\n", rxF); - printf("dl_ch addr %p\n",dl_ch); - //#endif - // if ((ue->frame_parms.N_RB_DL&1)==0) { + printf("dl_ch addr %p\n",dl_ch); +#endif + // if ((ue->frame_parms.N_RB_DL&1)==0) { // Treat first 2 pilots specially (left edge) ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDCCH printf("ch 0 %d\n",((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])); printf("pilot 0 : rxF - > (%d,%d) addr %p ch -> (%d,%d), pil -> (%d,%d) \n",rxF[0],rxF[1],&rxF[0],ch[0],ch[1],pil[0],pil[1]); #endif @@ -335,7 +337,7 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDCCH printf("pilot 1 : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); #endif multadd_real_vector_complex_scalar(fm, @@ -348,7 +350,7 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDCCH printf("pilot 2 : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); #endif @@ -357,7 +359,7 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, dl_ch, 16); -#ifdef DEBUG_CH +#ifdef DEBUG_PDCCH for (int m =0; m<12; m++) printf("data : dl_ch -> (%d,%d)\n",dl_ch[0+2*m],dl_ch[1+2*m]); #endif @@ -376,7 +378,7 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDCCH printf("pilot %d : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",pilot_cnt,rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); #endif multadd_real_vector_complex_scalar(fl, @@ -392,7 +394,7 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDCCH printf("pilot %d : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",pilot_cnt+1,rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); #endif multadd_real_vector_complex_scalar(fm, @@ -405,7 +407,7 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDCCH printf("pilot 1 : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); #endif @@ -424,33 +426,81 @@ int nr_pdcch_channel_estimation(PHY_VARS_NR_UE *ue, //} } + + void (*idft)(int16_t *,int16_t *, int); + + switch (ue->frame_parms.ofdm_symbol_size) { + case 128: + idft = idft128; + break; + + case 256: + idft = idft256; + break; + + case 512: + idft = idft512; + break; + + case 1024: + idft = idft1024; + break; + + case 1536: + idft = idft1536; + break; + + case 2048: + idft = idft2048; + break; + + default: + idft = idft512; + break; + } + + if( (Ns== 2) && (l == 0)) + { + // do ifft of channel estimate + for (aarx=0; aarxframe_parms.nb_antennas_rx; aarx++) + for (p=0; pframe_parms.nb_antenna_ports_eNB; p++) { + if (ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[Ns>>1]].dl_ch_estimates[eNB_offset][(p<<1)+aarx]) + { + LOG_D(PHY,"Channel Impulse Computation Slot %d ThreadId %d Symbol %d \n", Ns, ue->current_thread_id[Ns>>1], l); + idft((int16_t*) &ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[Ns>>1]].dl_ch_estimates[eNB_offset][(p<<1)+aarx][0], + (int16_t*) ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[Ns>>1]].dl_ch_estimates_time[eNB_offset][(p<<1)+aarx],1); + } + } + } return(0); } int nr_pdsch_channel_estimation(PHY_VARS_NR_UE *ue, - uint8_t eNB_id, - uint8_t eNB_offset, - unsigned char Ns, - unsigned char l, - unsigned char symbol, - unsigned short bwp_start_subcarrier, - unsigned short nb_rb_pdsch) + uint8_t eNB_id, + uint8_t eNB_offset, + unsigned char Ns, + unsigned short p, + unsigned char l, + unsigned char symbol, + unsigned short bwp_start_subcarrier, + unsigned short nb_rb_pdsch) { - int pilot[200] __attribute__((aligned(16))); + int pilot[1320] __attribute__((aligned(16))); unsigned char aarx; unsigned short k; unsigned int pilot_cnt; - int16_t ch[2],*pil,*rxF,*dl_ch,*fl,*fm,*fr,*fml,*fmr; + int16_t ch[2],*pil,*rxF,*dl_ch,*fl,*fm,*fr,*fml,*fmr,*fmm; int ch_offset,symbol_offset; - //uint16_t Nid_cell = (eNB_offset == 0) ? ue->frame_parms.Nid_cell : ue->measurements.adj_cell_id[eNB_offset-1]; - uint8_t nushift=0; + uint8_t nushift; int **dl_ch_estimates =ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[Ns>>1]].dl_ch_estimates[eNB_offset]; int **rxdataF=ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[Ns>>1]].rxdataF; + nushift = (p>>1)&1; + ue->frame_parms.nushift = nushift; if (ue->high_speed_flag == 0) // use second channel estimate position for temporary storage ch_offset = ue->frame_parms.ofdm_symbol_size ; @@ -460,38 +510,41 @@ int nr_pdsch_channel_estimation(PHY_VARS_NR_UE *ue, symbol_offset = ue->frame_parms.ofdm_symbol_size*symbol; k = bwp_start_subcarrier; + int re_offset = k; #ifdef DEBUG_CH - printf("PBCH Channel Estimation : ThreadId %d, eNB_offset %d ch_offset %d, OFDM size %d, Ncp=%d, l=%d, Ns=%d, k=%d symbol %d\n",ue->current_thread_id[Ns>>1], eNB_offset,ch_offset,ue->frame_parms.ofdm_symbol_size, + printf("PDSCH Channel Estimation : ThreadId %d, eNB_offset %d ch_offset %d, symbol_offset %d OFDM size %d, Ncp=%d, l=%d, Ns=%d, k=%d symbol %d\n",ue->current_thread_id[Ns], eNB_offset,ch_offset,symbol_offset,ue->frame_parms.ofdm_symbol_size, ue->frame_parms.Ncp,l,Ns,k, symbol); #endif switch (nushift) { - case 0: - fl = filt8_l0; - fm = filt8_m0; - fr = filt8_r0; - fml = filt8_m0; - fmr = filt8_mr0; - break; + case 0: + fl = filt8_l0; + fm = filt8_m0; + fr = filt8_r0; + fmm = filt8_mm0; + fml = filt8_m0; + fmr = filt8_mr0; + break; - case 1: - fl = filt8_l1; - fm = filt8_m1; - fr = filt8_r1; - fml = filt8_ml1; - fmr = filt8_m1; - break; + case 1: + fl = filt8_l1; + fm = filt8_m1; + fr = filt8_r1; + fmm = filt8_mm1; + fml = filt8_ml1; + fmr = filt8_m1; + break; - default: - msg("pdsch_channel_estimation: nushift=%d -> ERROR\n",nushift); - return(-1); - break; - } + default: + msg("pdsch_channel_estimation: nushift=%d -> ERROR\n",nushift); + return(-1); + break; + } // generate pilot - nr_pdsch_dmrs_rx(ue,eNB_offset,Ns,ue->nr_gold_pdsch, &pilot[0],1000,1,nb_rb_pdsch); + nr_pdsch_dmrs_rx(ue,Ns,ue->nr_gold_pdsch[eNB_offset][Ns][0], &pilot[0],1000,0,nb_rb_pdsch); for (aarx=0; aarxframe_parms.nb_antennas_rx; aarx++) { @@ -504,33 +557,35 @@ int nr_pdsch_channel_estimation(PHY_VARS_NR_UE *ue, multadd_complex_vector_real_scalar(dl_ch-(ue->frame_parms.ofdm_symbol_size<<1), ue->ch_est_alpha,dl_ch-(ue->frame_parms.ofdm_symbol_size<<1), 1,ue->frame_parms.ofdm_symbol_size); -#ifdef DEBUG_CH +#ifdef DEBUG_PDSCH printf("ch est pilot addr %p RB_DL %d\n",&pilot[0], ue->frame_parms.N_RB_DL); printf("k %d, first_carrier %d\n",k,ue->frame_parms.first_carrier_offset); - printf("rxF addr %p\n", rxF); - printf("dl_ch addr %p\n",dl_ch); + printf("rxF addr %p p %d\n", rxF,p); + printf("dl_ch addr %p nushift %d\n",dl_ch,nushift); #endif - if ((ue->frame_parms.N_RB_DL&1)==0) { + //if ((ue->frame_parms.N_RB_DL&1)==0) { // Treat first 2 pilots specially (left edge) ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDSCH printf("ch 0 %d\n",((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])); printf("pilot 0 : rxF - > (%d,%d) addr %p ch -> (%d,%d), pil -> (%d,%d) \n",rxF[0],rxF[1],&rxF[0],ch[0],ch[1],pil[0],pil[1]); + printf("data 0 : rxF - > (%d,%d) addr %p ch -> (%d,%d), pil -> (%d,%d) \n",rxF[2],rxF[3],&rxF[2],ch[0],ch[1],pil[0],pil[1]); #endif multadd_real_vector_complex_scalar(fl, ch, dl_ch, 8); pil+=2; - rxF+=4; + re_offset = (re_offset+2)&(ue->frame_parms.ofdm_symbol_size-1); + rxF = (int16_t *)&rxdataF[aarx][(symbol_offset+nushift+re_offset)]; //for (int i= 0; i<8; i++) //printf("dl_ch addr %p %d\n", dl_ch+i, *(dl_ch+i)); ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDSCH printf("pilot 1 : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); #endif multadd_real_vector_complex_scalar(fml, @@ -538,21 +593,36 @@ int nr_pdsch_channel_estimation(PHY_VARS_NR_UE *ue, dl_ch, 8); pil+=2; - rxF+=4; - dl_ch+=4; - k+=4; + re_offset = (re_offset+2)&(ue->frame_parms.ofdm_symbol_size-1); + rxF = (int16_t *)&rxdataF[aarx][(symbol_offset+nushift+re_offset)]; + //printf("dl_ch addr %p\n",dl_ch); + + ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); + ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); +#ifdef DEBUG_PDSCH + printf("pilot 2 : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); +#endif + multadd_real_vector_complex_scalar(fmm, + ch, + dl_ch, + 8); + + //for (int i= 0; i<16; i++) + //printf("dl_ch addr %p %d\n", dl_ch+i, *(dl_ch+i)); + + pil+=2; + re_offset = (re_offset+2)&(ue->frame_parms.ofdm_symbol_size-1); + rxF = (int16_t *)&rxdataF[aarx][(symbol_offset+nushift+re_offset)]; + dl_ch+=8; - for (pilot_cnt=2; pilot_cnt<(6*(nb_rb_pdsch-1)+4); pilot_cnt+=2) { - if ((pilot_cnt%6)==0) - dl_ch+=4; - - if (k >= ue->frame_parms.ofdm_symbol_size){ - k-=ue->frame_parms.ofdm_symbol_size; - rxF = (int16_t *)&rxdataF[aarx][(symbol_offset+k+nushift)];} + for (pilot_cnt=3; pilot_cnt<(6*nb_rb_pdsch-3); pilot_cnt+=2) { + //if ((pilot_cnt%6)==0) + //dl_ch+=4; + //printf("re_offset %d\n",re_offset); ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDSCH printf("pilot %d : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",pilot_cnt,rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); #endif multadd_real_vector_complex_scalar(fm, @@ -561,56 +631,69 @@ int nr_pdsch_channel_estimation(PHY_VARS_NR_UE *ue, 8); pil+=2; - rxF+=4; - dl_ch+=4; - + re_offset = (re_offset+2)&(ue->frame_parms.ofdm_symbol_size-1); + rxF = (int16_t *)&rxdataF[aarx][(symbol_offset+nushift+re_offset)]; + ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDSCH printf("pilot %d : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",pilot_cnt+1,rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); #endif - multadd_real_vector_complex_scalar(fm, + multadd_real_vector_complex_scalar(fmm, ch, dl_ch, 8); pil+=2; - rxF+=4; - dl_ch+=4; - k+=4; + re_offset = (re_offset+2)&(ue->frame_parms.ofdm_symbol_size-1); + rxF = (int16_t *)&rxdataF[aarx][(symbol_offset+nushift+re_offset)]; + dl_ch+=8; } - + // Treat first 2 pilots specially (right edge) + ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); + ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); +#ifdef DEBUG_PDSCH + printf("pilot %d : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",pilot_cnt,rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); +#endif + multadd_real_vector_complex_scalar(fm, + ch, + dl_ch, + 8); + + //for (int i= 0; i<8; i++) + //printf("dl_ch addr %p %d\n", dl_ch+i, *(dl_ch+i)); + + pil+=2; + re_offset = (re_offset+2)&(ue->frame_parms.ofdm_symbol_size-1); + rxF = (int16_t *)&rxdataF[aarx][(symbol_offset+nushift+re_offset)]; + ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH +#ifdef DEBUG_PDSCH printf("ch 0 %d\n",((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])); - printf("pilot 0 : rxF - > (%d,%d) addr %p ch -> (%d,%d), pil -> (%d,%d) \n",rxF[0],rxF[1],&rxF[0],ch[0],ch[1],pil[0],pil[1]); + printf("pilot %d: rxF - > (%d,%d) addr %p ch -> (%d,%d), pil -> (%d,%d) \n",pilot_cnt+1,rxF[0],rxF[1],&rxF[0],ch[0],ch[1],pil[0],pil[1]); #endif multadd_real_vector_complex_scalar(fmr, ch, dl_ch, 8); + pil+=2; - rxF+=4; - //for (int i= 0; i<8; i++) - //printf("dl_ch addr %p %d\n", dl_ch+i, *(dl_ch+i)); - + re_offset = (re_offset+2)&(ue->frame_parms.ofdm_symbol_size-1); + rxF = (int16_t *)&rxdataF[aarx][(symbol_offset+nushift+re_offset)]; + dl_ch+=8; + ch[0] = (int16_t)(((int32_t)pil[0]*rxF[0] - (int32_t)pil[1]*rxF[1])>>15); ch[1] = (int16_t)(((int32_t)pil[0]*rxF[1] + (int32_t)pil[1]*rxF[0])>>15); -#ifdef DEBUG_CH - printf("pilot 1 : rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); +#ifdef DEBUG_PDSCH + printf("pilot %d: rxF - > (%d,%d) ch -> (%d,%d), pil -> (%d,%d) \n",pilot_cnt+2,rxF[0],rxF[1],ch[0],ch[1],pil[0],pil[1]); #endif multadd_real_vector_complex_scalar(fr, ch, dl_ch, 8); - pil+=2; - rxF+=4; - dl_ch+=4; - k+=4; - - } + //} } diff --git a/openair1/PHY/NR_UE_ESTIMATION/nr_estimation.h b/openair1/PHY/NR_UE_ESTIMATION/nr_estimation.h index 25d6755723..ae4295a99b 100644 --- a/openair1/PHY/NR_UE_ESTIMATION/nr_estimation.h +++ b/openair1/PHY/NR_UE_ESTIMATION/nr_estimation.h @@ -62,8 +62,17 @@ int nr_pdsch_channel_estimation(PHY_VARS_NR_UE *ue, uint8_t eNB_id, uint8_t eNB_offset, unsigned char Ns, + unsigned short p, unsigned char l, unsigned char symbol, unsigned short bwp_start_subcarrier, unsigned short nb_rb_pdsch); + +void nr_adjust_synch_ue(NR_DL_FRAME_PARMS *frame_parms, + PHY_VARS_NR_UE *ue, + module_id_t eNB_id, + uint8_t subframe, + unsigned char clear, + short coef); + #endif diff --git a/openair1/PHY/NR_UE_TRANSPORT/dci_nr.c b/openair1/PHY/NR_UE_TRANSPORT/dci_nr.c index dcd69b83ed..f072532d27 100755 --- a/openair1/PHY/NR_UE_TRANSPORT/dci_nr.c +++ b/openair1/PHY/NR_UE_TRANSPORT/dci_nr.c @@ -62,7 +62,8 @@ #ifdef NR_PDCCH_DCI_RUN -static const uint16_t conjugate[8]__attribute__((aligned(32))) = {-1,1,-1,1,-1,1,-1,1}; + +//static const int16_t conjugate[8]__attribute__((aligned(32))) = {-1,1,-1,1,-1,1,-1,1}; void nr_pdcch_demapping_deinterleaving(uint32_t *llr, @@ -130,6 +131,7 @@ void nr_pdcch_demapping_deinterleaving(uint32_t *llr, bundle_j = (c*coreset_interleaver_size_R)+r; f_bundle_j = ((r*coreset_C)+c+n_shift)%((coreset_nbr_rb*coreset_time_dur)/reg_bundle_size_L); if (coreset_interleaved==0) f_bundle_j=bundle_j; + #ifdef NR_PDCCH_DCI_DEBUG printf("\n\t\t<-NR_PDCCH_DCI_DEBUG (nr_pdcch_demapping_deinterleaving)-> [r=%d,c=%d] bundle_j(%d) interleaved at f_bundle_j(%d)\n",r,c,bundle_j,f_bundle_j); #endif @@ -141,6 +143,7 @@ void nr_pdcch_demapping_deinterleaving(uint32_t *llr, index_llr = 9*((uint16_t)floor(f_reg/coreset_time_dur)+((f_reg%coreset_time_dur)*(coreset_nbr_rb))); for (int i=0; i<9; i++){ z[index_z + i] = llr[index_llr + i]; + #ifdef NR_PDCCH_DCI_DEBUG printf("\t\t<-NR_PDCCH_DCI_DEBUG (nr_pdcch_demapping_deinterleaving)-> [reg=%d,bundle_j=%d] z[%d]=(%d,%d) <-> \t[f_reg=%d,fbundle_j=%d] llr[%d]=(%d,%d) \n", reg,bundle_j,(index_z + i),*(int16_t*) &z[index_z + i],*(1 + (int16_t*) &z[index_z + i]), @@ -163,6 +166,7 @@ int32_t nr_pdcch_llr(NR_DL_FRAME_PARMS *frame_parms, int32_t **rxdataF_comp, pdcch_llrp = &pdcch_llr[2 * symbol * coreset_nbr_rb * 9]; + if (!pdcch_llrp) { printf("pdcch_qpsk_llr: llr is null, symbol %d\n", symbol); return (-1); @@ -299,89 +303,6 @@ __m128i mmtmpPD0,mmtmpPD1,mmtmpPD2,mmtmpPD3; #endif -void pdcch_detection_mrc_i(NR_DL_FRAME_PARMS *frame_parms, - int32_t **rxdataF_comp, - int32_t **rxdataF_comp_i, - int32_t **rho, - int32_t **rho_i, - uint8_t symbol) -{ - - uint8_t aatx; - -#if defined(__x86_64__) || defined(__i386__) - __m128i *rxdataF_comp128_0,*rxdataF_comp128_1,*rxdataF_comp128_i0,*rxdataF_comp128_i1,*rho128_0,*rho128_1,*rho128_i0,*rho128_i1; -#elif defined(__arm__) - int16x8_t *rxdataF_comp128_0,*rxdataF_comp128_1,*rxdataF_comp128_i0,*rxdataF_comp128_i1,*rho128_0,*rho128_1,*rho128_i0,*rho128_i1; -#endif - int32_t i; - - if (frame_parms->nb_antennas_rx>1) { - for (aatx=0; aatxnb_antenna_ports_eNB; aatx++) { - //if (frame_parms->mode1_flag && (aatx>0)) break; - -#if defined(__x86_64__) || defined(__i386__) - rxdataF_comp128_0 = (__m128i *)&rxdataF_comp[(aatx<<1)][symbol*frame_parms->N_RB_DL*12]; - rxdataF_comp128_1 = (__m128i *)&rxdataF_comp[(aatx<<1)+1][symbol*frame_parms->N_RB_DL*12]; -#elif defined(__arm__) - rxdataF_comp128_0 = (int16x8_t *)&rxdataF_comp[(aatx<<1)][symbol*frame_parms->N_RB_DL*12]; - rxdataF_comp128_1 = (int16x8_t *)&rxdataF_comp[(aatx<<1)+1][symbol*frame_parms->N_RB_DL*12]; -#endif - // MRC on each re of rb on MF output - for (i=0; iN_RB_DL*3; i++) { -#if defined(__x86_64__) || defined(__i386__) - rxdataF_comp128_0[i] = _mm_adds_epi16(_mm_srai_epi16(rxdataF_comp128_0[i],1),_mm_srai_epi16(rxdataF_comp128_1[i],1)); -#elif defined(__arm__) - rxdataF_comp128_0[i] = vhaddq_s16(rxdataF_comp128_0[i],rxdataF_comp128_1[i]); -#endif - } - } - -#if defined(__x86_64__) || defined(__i386__) - rho128_0 = (__m128i *) &rho[0][symbol*frame_parms->N_RB_DL*12]; - rho128_1 = (__m128i *) &rho[1][symbol*frame_parms->N_RB_DL*12]; -#elif defined(__arm__) - rho128_0 = (int16x8_t *) &rho[0][symbol*frame_parms->N_RB_DL*12]; - rho128_1 = (int16x8_t *) &rho[1][symbol*frame_parms->N_RB_DL*12]; -#endif - for (i=0; iN_RB_DL*3; i++) { -#if defined(__x86_64__) || defined(__i386__) - rho128_0[i] = _mm_adds_epi16(_mm_srai_epi16(rho128_0[i],1),_mm_srai_epi16(rho128_1[i],1)); -#elif defined(__arm__) - rho128_0[i] = vhaddq_s16(rho128_0[i],rho128_1[i]); -#endif - } - -#if defined(__x86_64__) || defined(__i386__) - rho128_i0 = (__m128i *) &rho_i[0][symbol*frame_parms->N_RB_DL*12]; - rho128_i1 = (__m128i *) &rho_i[1][symbol*frame_parms->N_RB_DL*12]; - rxdataF_comp128_i0 = (__m128i *)&rxdataF_comp_i[0][symbol*frame_parms->N_RB_DL*12]; - rxdataF_comp128_i1 = (__m128i *)&rxdataF_comp_i[1][symbol*frame_parms->N_RB_DL*12]; -#elif defined(__arm__) - rho128_i0 = (int16x8_t*) &rho_i[0][symbol*frame_parms->N_RB_DL*12]; - rho128_i1 = (int16x8_t*) &rho_i[1][symbol*frame_parms->N_RB_DL*12]; - rxdataF_comp128_i0 = (int16x8_t *)&rxdataF_comp_i[0][symbol*frame_parms->N_RB_DL*12]; - rxdataF_comp128_i1 = (int16x8_t *)&rxdataF_comp_i[1][symbol*frame_parms->N_RB_DL*12]; - -#endif - // MRC on each re of rb on MF and rho - for (i=0; iN_RB_DL*3; i++) { -#if defined(__x86_64__) || defined(__i386__) - rxdataF_comp128_i0[i] = _mm_adds_epi16(_mm_srai_epi16(rxdataF_comp128_i0[i],1),_mm_srai_epi16(rxdataF_comp128_i1[i],1)); - rho128_i0[i] = _mm_adds_epi16(_mm_srai_epi16(rho128_i0[i],1),_mm_srai_epi16(rho128_i1[i],1)); -#elif defined(__arm__) - rxdataF_comp128_i0[i] = vhaddq_s16(rxdataF_comp128_i0[i],rxdataF_comp128_i1[i]); - rho128_i0[i] = vhaddq_s16(rho128_i0[i],rho128_i1[i]); - -#endif - } - } - -#if defined(__x86_64__) || defined(__i386__) - _mm_empty(); - _m_empty(); -#endif -} #ifdef NR_PDCCH_DCI_RUN @@ -504,6 +425,7 @@ void nr_pdcch_extract_rbs_single(int32_t **rxdataF, if ((c_rb < (frame_parms->N_RB_DL >> 1)) && ((frame_parms->N_RB_DL & 1) == 0)) { //if RB to be treated is lower than middle system bandwidth then rxdataF pointed at (offset + c_br + symbol * ofdm_symbol_size): even case rxF = &rxdataF[aarx][(frame_parms->first_carrier_offset + 12 * c_rb + (symbol * (frame_parms->ofdm_symbol_size)))]; + #ifdef NR_PDCCH_DCI_DEBUG printf("\t\t<-NR_PDCCH_DCI_DEBUG (nr_pdcch_extract_rbs_single)-> in even case c_rb (%d) is lower than half N_RB_DL -> rxF = &rxdataF[aarx = (%d)][(frame_parms->first_carrier_offset + 12 * c_rb + (symbol * (frame_parms->ofdm_symbol_size))) = (%d)]\n", c_rb,aarx,(frame_parms->first_carrier_offset + 12 * c_rb + (symbol * (frame_parms->ofdm_symbol_size)))); @@ -526,6 +448,7 @@ void nr_pdcch_extract_rbs_single(int32_t **rxdataF, if ((c_rb < (frame_parms->N_RB_DL >> 1)) && ((frame_parms->N_RB_DL & 1) != 0)){ //if RB to be treated is lower than middle system bandwidth then rxdataF pointed at (offset + c_br + symbol * ofdm_symbol_size): odd case rxF = &rxdataF[aarx][(frame_parms->first_carrier_offset + 12 * c_rb + (symbol * (frame_parms->ofdm_symbol_size)))]; + #ifdef NR_PDCCH_DCI_DEBUG printf("\t\t<-NR_PDCCH_DCI_DEBUG (nr_pdcch_extract_rbs_single)-> in odd case c_rb (%d) is lower or equal than half N_RB_DL -> rxF = &rxdataF[aarx = (%d)][(frame_parms->first_carrier_offset + 12 * c_rb + (symbol * (frame_parms->ofdm_symbol_size))) = (%d)]\n", c_rb,aarx,(frame_parms->first_carrier_offset + 12 * c_rb + (symbol * (frame_parms->ofdm_symbol_size)))); @@ -544,6 +467,7 @@ void nr_pdcch_extract_rbs_single(int32_t **rxdataF, // if odd number RBs in system bandwidth and first RB to be treated is higher than middle system bandwidth (around DC) // we have to treat the RB in two parts: first part from i=0 to 5, the data is at the end of rxdataF (pointing at the end of the table) rxF = &rxdataF[aarx][(frame_parms->first_carrier_offset + 12 * c_rb + (symbol * (frame_parms->ofdm_symbol_size)))]; + #ifdef NR_PDCCH_DCI_DEBUG printf("\t\t<-NR_PDCCH_DCI_DEBUG (nr_pdcch_extract_rbs_single)-> in odd case c_rb (%d) is half N_RB_DL + 1 we treat DC case -> rxF = &rxdataF[aarx = (%d)][(frame_parms->first_carrier_offset + 12 * c_rb + (symbol * (frame_parms->ofdm_symbol_size))) = (%d)]\n", c_rb,aarx,(frame_parms->first_carrier_offset + 12 * c_rb + (symbol * (frame_parms->ofdm_symbol_size)))); @@ -614,6 +538,7 @@ void nr_pdcch_extract_rbs_single(int32_t **rxdataF, for (i = 0; i < 12; i++) { if ((i != 1) && (i != 5) && (i != 9)) { rxF_ext[j] = rxF[i]; + #ifdef NR_PDCCH_DCI_DEBUG printf("\t\t<-NR_PDCCH_DCI_DEBUG (nr_pdcch_extract_rbs_single)-> RB[c_rb %d] \t RE[re %d] => rxF_ext[%d]=(%d,%d)\t rxF[%d]=(%d,%d)\n", c_rb, i, j, *(short *) &rxF_ext[j],*(1 + (short*) &rxF_ext[j]), i, @@ -759,6 +684,7 @@ void nr_pdcch_channel_compensation(int32_t **rxdataF_ext, rxdataF128+=3; rxdataF_comp128+=3; + #elif defined(__arm__) // to be filled in #endif @@ -855,7 +781,7 @@ int32_t nr_rx_pdcch(PHY_VARS_NR_UE *ue, int do_common; if (searchSpaceType == common) do_common=1; if (searchSpaceType == ue_specific) do_common=0; - uint8_t log2_maxh, aatx, aarx; + uint8_t log2_maxh, aarx; int32_t avgs; int32_t avgP[4]; @@ -940,6 +866,7 @@ int32_t nr_rx_pdcch(PHY_VARS_NR_UE *ue, for (int s = start_symbol; s < (start_symbol + coreset_time_dur); s++) { + #ifdef NR_PDCCH_DCI_DEBUG printf("\t<-NR_PDCCH_DCI_DEBUG (nr_rx_pdcch)-> we enter nr_pdcch_extract_rbs_single(is_secondary_ue=%d) to remove DM-RS PDCCH\n", is_secondary_ue); @@ -985,7 +912,7 @@ int32_t nr_rx_pdcch(PHY_VARS_NR_UE *ue, nr_pdcch_channel_compensation(pdcch_vars[eNB_id]->rxdataF_ext, pdcch_vars[eNB_id]->dl_ch_estimates_ext, pdcch_vars[eNB_id]->rxdataF_comp, - (aatx > 1) ? pdcch_vars[eNB_id]->rho : NULL, + NULL, frame_parms, s, log2_maxh, @@ -1013,7 +940,7 @@ int32_t nr_rx_pdcch(PHY_VARS_NR_UE *ue, #endif nr_pdcch_llr(frame_parms, pdcch_vars[eNB_id]->rxdataF_comp, - (char *) pdcch_vars[eNB_id]->llr, + pdcch_vars[eNB_id]->llr, s, coreset_nbr_rb); @@ -1048,7 +975,7 @@ int32_t nr_rx_pdcch(PHY_VARS_NR_UE *ue, nr_pdcch_unscrambling(pdcch_vars[eNB_id]->crnti, frame_parms, nr_tti_rx, - (uint16_t*) pdcch_vars[eNB_id]->e_rx, + pdcch_vars[eNB_id]->e_rx, coreset_time_dur*coreset_nbr_rb*9*2, // get_nCCE(n_pdcch_symbols, frame_parms, mi) * 72, pdcch_DMRS_scrambling_id, @@ -1130,6 +1057,7 @@ void nr_pdcch_unscrambling(uint16_t crnti, NR_DL_FRAME_PARMS *frame_parms, uint8 reset = 0; } + /* #ifdef NR_PDCCH_DCI_DEBUG if (i%2 == 0) printf("\t\t<-NR_PDCCH_DCI_DEBUG (nr_pdcch_unscrambling)-> unscrambling %d : scrambled_z=%d, => ", @@ -1157,6 +1085,7 @@ void nr_pdcch_unscrambling(uint16_t crnti, NR_DL_FRAME_PARMS *frame_parms, uint8 printf("unscrambled_z=%d\n",z[i]); #endif } + } #endif @@ -1194,7 +1123,7 @@ void nr_dci_decoding_procedure0(int s, uint8_t L2 = (1 << L); unsigned int Yk, nb_candidates = 0, i, m; unsigned int CCEmap_cand; - double *polar_input = malloc (sizeof(double) * 108*L2); + int8_t decoderState=0; // A[p], p is the current active CORESET @@ -1353,12 +1282,8 @@ void nr_dci_decoding_procedure0(int s, CCEmap = CCEmap1; else if (CCEind < 96) CCEmap = CCEmap2; - else { - LOG_E(PHY, "Illegal CCEind %d (Yk %d, m %d, nCCE %d, L2 %d\n",CCEind, Yk, m, nCCE, L2); - //mac_xface->macphy_exit("Illegal CCEind\n"); - return; // not reached - } - + else AssertFatal(1==0,"Illegal CCEind %d (Yk %d, m %d, nCCE %d, L2 %d\n",CCEind, Yk, m, nCCE[p], L2); + switch (L2) { case 1: CCEmap_mask = (1 << (CCEind & 0x1f)); @@ -1387,6 +1312,7 @@ void nr_dci_decoding_procedure0(int s, #endif if (CCEmap_cand == 0) { + #ifdef DEBUG_DCI_DECODING if (do_common == 1) LOG_I(PHY,"[DCI search nPdcch %d - common] Attempting candidate %d Aggregation Level %d DCI length %d at CCE %d/%d (CCEmap %x,CCEmap_cand %x)\n", @@ -1406,6 +1332,7 @@ void nr_dci_decoding_procedure0(int s, printf("\t\t<-NR_PDCCH_DCI_DEBUG (nr_dci_decoding_procedure0: polar decoding)-> polar intput (with coreset_time_dur=%d, coreset_nbr_rb=%d, p=%d, CCEind=%d): \n", coreset_time_dur,coreset_nbr_rb,p,CCEind); #endif + /* int reg_p=0,reg_e=0; for (int m=0; m < (L2*6); m++){ reg_p = (((int)floor(m/coreset_time_dur))+((m%coreset_time_dur)*(L2*6/coreset_time_dur)))*9*2; @@ -1431,14 +1358,16 @@ void nr_dci_decoding_procedure0(int s, } for (j=0;j<27;j++) printf("\t\t<-NR_PDCCH_DCI_DEBUG (nr_dci_decoding_procedure0: polar decoding input)-> polar_hex[%d]=%x\n",j,polar_hex[j]); #endif - + */ + uint32_t dci_estimation[4]={0}; nr_polar_init(&nrPolar_params, 1, sizeof_bits, L2); + t_nrPolar_paramsPtr currentPtrDCI=nr_polar_params(nrPolar_params, 1, sizeof_bits, L2); - decoderState = polar_decoder_int16((int16_t*)&pdcch_vars[eNB_id]->e_rx[CCEind*9*6*2], + decoderState = polar_decoder_int16(&pdcch_vars[eNB_id]->e_rx[CCEind*9*6*2], (uint64_t*)dci_estimation, currentPtrDCI); crc = decoderState; @@ -1455,6 +1384,7 @@ void nr_dci_decoding_procedure0(int s, printf ("\t\t<-NR_PDCCH_DCI_DEBUG (nr_dci_decoding_procedure0)-> crc_scrambled=%d\n",*crc_scrambled); #endif + if (crc == crc_scrambled_values[_C_RNTI_]) { *crc_scrambled =_c_rnti; *format_found=1; @@ -2598,6 +2528,7 @@ uint16_t nr_dci_format_size (PHY_VARS_NR_UE *ue, * */ + // } #ifdef NR_PDCCH_DCI_DEBUG printf("\t\t<-NR_PDCCH_DCI_DEBUG (nr_dci_format_size) dci_fields_sizes[][] = { \n"); @@ -2609,6 +2540,7 @@ uint16_t nr_dci_format_size (PHY_VARS_NR_UE *ue, printf(" }\n"); #endif + return dci_size[format]; } diff --git a/openair1/PHY/NR_UE_TRANSPORT/nr_dlsch_decoding.c b/openair1/PHY/NR_UE_TRANSPORT/nr_dlsch_decoding.c new file mode 100644 index 0000000000..428dfc9e99 --- /dev/null +++ b/openair1/PHY/NR_UE_TRANSPORT/nr_dlsch_decoding.c @@ -0,0 +1,2188 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.0 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +/*! \file PHY/NR_UE_TRANSPORT/nr_dlsch_decoding.c +* \brief Top-level routines for decoding Turbo-coded (DLSCH) transport channels from 36-212, V8.6 2009-03 +* \author R. Knopp +* \date 2011 +* \version 0.1 +* \company Eurecom +* \email: knopp@eurecom.fr +* \note +* \warning +*/ + +#include "PHY/defs_nr_UE.h" +#include "PHY/phy_extern_nr_ue.h" +#include "PHY/CODING/coding_extern.h" +#include "PHY/CODING/coding_defs.h" +#include "PHY/NR_TRANSPORT/nr_transport_common_proto.h" +#include "PHY/NR_TRANSPORT/nr_dlsch.h" +#include "SCHED_NR_UE/defs.h" +#include "SIMULATION/TOOLS/sim.h" +#include "targets/RT/USER/nr-uesoftmodem.h" +#include "PHY/CODING/nrLDPC_decoder/nrLDPC_decoder.h" +#include "PHY/CODING/nrLDPC_decoder/nrLDPC_types.h" +//#define DEBUG_DLSCH_DECODING + +#define OAI_LDPC_MAX_NUM_LLR 27000//26112 // NR_LDPC_NCOL_BG1*NR_LDPC_ZMAX + +static int8_t llrProcBuf[OAI_LDPC_MAX_NUM_LLR] __attribute__ ((aligned(32))); + +static uint64_t nb_total_decod =0; +static uint64_t nb_error_decod =0; + +//extern double cpuf; + +void free_nr_ue_dlsch(NR_UE_DLSCH_t *dlsch) +{ + + int i,r; + + if (dlsch) { + for (i=0; iMdlharq; i++) { + if (dlsch->harq_processes[i]) { + if (dlsch->harq_processes[i]->b) { + free16(dlsch->harq_processes[i]->b,MAX_DLSCH_PAYLOAD_BYTES); + dlsch->harq_processes[i]->b = NULL; + } + + for (r=0; rharq_processes[i]->c[r],1056); + dlsch->harq_processes[i]->c[r] = NULL; + } + + for (r=0; rharq_processes[i]->d[r]) { + free16(dlsch->harq_processes[i]->d[r],(3*8448)*sizeof(short)); + dlsch->harq_processes[i]->d[r] = NULL; + } + + free16(dlsch->harq_processes[i],sizeof(NR_DL_UE_HARQ_t)); + dlsch->harq_processes[i] = NULL; + } + } + + free16(dlsch,sizeof(NR_UE_DLSCH_t)); + dlsch = NULL; + } +} + +NR_UE_DLSCH_t *new_nr_ue_dlsch(uint8_t Kmimo,uint8_t Mdlharq,uint32_t Nsoft,uint8_t max_ldpc_iterations,uint8_t N_RB_DL, uint8_t abstraction_flag) +{ + + NR_UE_DLSCH_t *dlsch; + uint8_t exit_flag = 0,i,r; + + unsigned char bw_scaling =1; + + switch (N_RB_DL) { + case 6: + bw_scaling =16; + break; + + case 25: + bw_scaling =4; + break; + + case 50: + bw_scaling =2; + break; + + default: + bw_scaling =1; + break; + } + + dlsch = (NR_UE_DLSCH_t *)malloc16(sizeof(NR_UE_DLSCH_t)); + + if (dlsch) { + memset(dlsch,0,sizeof(NR_UE_DLSCH_t)); + dlsch->Kmimo = Kmimo; + dlsch->Mdlharq = Mdlharq; + dlsch->Nsoft = Nsoft; + dlsch->max_ldpc_iterations = max_ldpc_iterations; + + for (i=0; iharq_processes[i] = (NR_DL_UE_HARQ_t *)malloc16(sizeof(NR_DL_UE_HARQ_t)); + + if (dlsch->harq_processes[i]) { + memset(dlsch->harq_processes[i],0,sizeof(NR_DL_UE_HARQ_t)); + dlsch->harq_processes[i]->first_tx=1; + dlsch->harq_processes[i]->b = (uint8_t*)malloc16(MAX_DLSCH_PAYLOAD_BYTES/bw_scaling); + + if (dlsch->harq_processes[i]->b) + memset(dlsch->harq_processes[i]->b,0,MAX_DLSCH_PAYLOAD_BYTES/bw_scaling); + else + exit_flag=3; + + if (abstraction_flag == 0) { + for (r=0; rharq_processes[i]->c[r] = (uint8_t*)malloc16(1056); + + if (dlsch->harq_processes[i]->c[r]) + memset(dlsch->harq_processes[i]->c[r],0,1056); + else + exit_flag=2; + + dlsch->harq_processes[i]->d[r] = (short*)malloc16((3*8448)*sizeof(short)); + + if (dlsch->harq_processes[i]->d[r]) + memset(dlsch->harq_processes[i]->d[r],0,(3*8448)*sizeof(short)); + else + exit_flag=2; + } + } + } else { + exit_flag=1; + } + } + + if (exit_flag==0) + return(dlsch); + } + + printf("new_ue_dlsch with size %zu: exit_flag = %u\n",sizeof(NR_DL_UE_HARQ_t), exit_flag); + free_nr_ue_dlsch(dlsch); + + return(NULL); +} + +void nr_dlsch_unscrambling(int16_t* llr, + uint32_t size, + uint8_t q, + uint32_t Nid, + uint32_t n_RNTI) { + + uint8_t reset; + uint32_t x1, x2, s=0; + + reset = 1; + x2 = (n_RNTI<<15) + (q<<14) + Nid; + + for (int i=0; i>(i&0x1f))&1)==1) + llr[i] = -llr[i]; + } + +} + +uint32_t nr_dlsch_decoding(PHY_VARS_NR_UE *phy_vars_ue, + short *dlsch_llr, + NR_DL_FRAME_PARMS *frame_parms, + NR_UE_DLSCH_t *dlsch, + NR_DL_UE_HARQ_t *harq_process, + uint32_t frame, + uint16_t nb_symb_sch, + uint8_t nr_tti_rx, + uint8_t harq_pid, + uint8_t is_crnti, + uint8_t llr8_flag) +{ + +#if UE_TIMING_TRACE + time_stats_t *dlsch_rate_unmatching_stats=&phy_vars_ue->dlsch_rate_unmatching_stats; + time_stats_t *dlsch_turbo_decoding_stats=&phy_vars_ue->dlsch_turbo_decoding_stats; + time_stats_t *dlsch_deinterleaving_stats=&phy_vars_ue->dlsch_deinterleaving_stats; +#endif + uint32_t A,E; + uint32_t G; + uint32_t ret,offset; + int32_t no_iteration_ldpc; + //short dummy_w[MAX_NUM_DLSCH_SEGMENTS][3*(8448+64)]; + uint32_t r,r_offset=0,Kr=8424,Kr_bytes,K_bytes_F,err_flag=0; + uint8_t crc_type; + t_nrLDPC_dec_params decParams; + t_nrLDPC_dec_params* p_decParams = &decParams; + t_nrLDPC_time_stats procTime; + t_nrLDPC_time_stats* p_procTime =&procTime ; + int16_t z [68*384]; + int8_t l [68*384]; + //__m128i l; + int16_t inv_d [68*384]; +// int16_t *p_invd =&inv_d; + uint8_t kb, kc; + uint8_t Ilbrm = 0; + uint32_t Tbslbrm = 950984; + uint16_t nb_rb = 30; //to update + //uint16_t nb_symb_sch = 12; + uint8_t nb_re_dmrs = 6; + uint16_t length_dmrs = 1; + + uint32_t i,j; +// uint32_t k; + + __m128i *pv = (__m128i*)&z; + __m128i *pl = (__m128i*)&l; + + //NR_DL_UE_HARQ_t *harq_process = dlsch->harq_processes[0]; + + if (!dlsch_llr) { + printf("dlsch_decoding.c: NULL dlsch_llr pointer\n"); + return(dlsch->max_ldpc_iterations); + } + + if (!harq_process) { + printf("dlsch_decoding.c: NULL harq_process pointer\n"); + return(dlsch->max_ldpc_iterations); + } + + if (!frame_parms) { + printf("dlsch_decoding.c: NULL frame_parms pointer\n"); + return(dlsch->max_ldpc_iterations); + } + + /*if (nr_tti_rx> (10*frame_parms->ttis_per_subframe-1)) { + printf("dlsch_decoding.c: Illegal subframe index %d\n",nr_tti_rx); + return(dlsch->max_ldpc_iterations); + }*/ + + /*if (harq_process->harq_ack.ack != 2) { + LOG_D(PHY, "[UE %d] DLSCH @ SF%d : ACK bit is %d instead of DTX even before PDSCH is decoded!\n", + phy_vars_ue->Mod_id, nr_tti_rx, harq_process->harq_ack.ack); + }*/ + + // nb_rb = dlsch->nb_rb; + + /* + if (nb_rb > frame_parms->N_RB_DL) { + printf("dlsch_decoding.c: Illegal nb_rb %d\n",nb_rb); + return(max_ldpc_iterations); + }*/ + + /*harq_pid = dlsch->current_harq_pid[phy_vars_ue->current_thread_id[subframe]]; + if (harq_pid >= 8) { + printf("dlsch_decoding.c: Illegal harq_pid %d\n",harq_pid); + return(max_ldpc_iterations); + } + */ + + nb_rb = harq_process->nb_rb; + + harq_process->trials[harq_process->round]++; + + harq_process->TBS = nr_compute_tbs(harq_process->mcs,nb_rb,nb_symb_sch,nb_re_dmrs,length_dmrs, harq_process->Nl); + + A = harq_process->TBS; + ret = dlsch->max_ldpc_iterations; + + harq_process->G = nr_get_G(nb_rb, nb_symb_sch, nb_re_dmrs, length_dmrs, harq_process->Qm,harq_process->Nl); + G = harq_process->G; + + LOG_I(PHY,"DLSCH Decoding, harq_pid %d TBS %d G %d mcs %d Nl %d nb_symb_sch %d nb_rb %d\n",harq_pid,A,G, harq_process->mcs, harq_process->Nl, nb_symb_sch,nb_rb); + + if (harq_process->round == 0) { + // This is a new packet, so compute quantities regarding segmentation + harq_process->B = A+24; + nr_segmentation(NULL, + NULL, + harq_process->B, + &harq_process->C, + &harq_process->K, + &harq_process->Z, + &harq_process->F); + p_decParams->Z = harq_process->Z; + +#ifdef DEBUG_DLSCH_DECODING + printf("dlsch decoding nr segmentation Z %d\n", p_decParams->Z); + if (!frame%100) + printf("K %d C %d Z %d nl %d \n", harq_process->K, harq_process->C, p_decParams->Z, harq_process->Nl); +#endif + } + + kb = harq_process->K/harq_process->Z; + if ( kb==22){ + p_decParams->BG = 1; + p_decParams->R = 13; + kc = 68; + } + else{ + p_decParams->BG = 2; + p_decParams->R = 13; + kc = 52; + } + + p_decParams->numMaxIter = dlsch->max_ldpc_iterations; + Kr = p_decParams->Z*kb; + p_decParams->outMode= 0; + + + err_flag = 0; + r_offset = 0; + + unsigned char bw_scaling =1; + + switch (frame_parms->N_RB_DL) { + + case 106: + bw_scaling =2; + break; + + default: + bw_scaling =1; + break; + } + + if (harq_process->C > MAX_NUM_NR_DLSCH_SEGMENTS/bw_scaling) { + LOG_E(PHY,"Illegal harq_process->C %d > %d\n",harq_process->C,MAX_NUM_NR_DLSCH_SEGMENTS/bw_scaling); + return((1+dlsch->max_ldpc_iterations)); + } +#ifdef DEBUG_DLSCH_DECODING + printf("Segmentation: C %d, K %d\n",harq_process->C,harq_process->K); +#endif + + opp_enabled=1; + + Kr = harq_process->K; + Kr_bytes = Kr>>3; + + K_bytes_F = Kr_bytes-(harq_process->F>>3); + + Tbslbrm = nr_compute_tbs(28,nb_rb,frame_parms->symbols_per_slot,0,0, harq_process->Nl); + + for (r=0; rC; r++) { + + //printf("start rx segment %d\n",r); + +#if UE_TIMING_TRACE + start_meas(dlsch_rate_unmatching_stats); +#endif + +#ifdef DEBUG_DLSCH_DECODING + LOG_D(PHY,"HARQ_PID %d Rate Matching Segment %d (coded bits %d,unpunctured/repeated bits %d, TBS %d, mod_order %d, nb_rb %d, Nl %d, rv %d, round %d)...\n", + harq_pid,r, G, + Kr*3, + harq_process->TBS, + harq_process->Qm, + harq_process->nb_rb, + harq_process->Nl, + harq_process->rvidx, + harq_process->round); +#endif + + if (nr_rate_matching_ldpc_rx(Ilbrm, + Tbslbrm, + p_decParams->BG, + p_decParams->Z, + G, + harq_process->w[r], + dlsch_llr+r_offset, + harq_process->C, + harq_process->rvidx, + (harq_process->round==0)?1:0, + harq_process->Qm, + harq_process->Nl, + r, + &E)==-1) { + +#if UE_TIMING_TRACE + stop_meas(dlsch_rate_unmatching_stats); +#endif + LOG_E(PHY,"dlsch_decoding.c: Problem in rate_matching\n"); + return(dlsch->max_ldpc_iterations); + } else + { +#if UE_TIMING_TRACE + stop_meas(dlsch_rate_unmatching_stats); +#endif + } + r_offset += E; + + //for (int i =0; i<16; i++) + // printf("rx output ratematching w[%d]= %d r_offset %d\n", i,harq_process->w[r][i], r_offset); + +#if UE_TIMING_TRACE + start_meas(dlsch_deinterleaving_stats); +#endif + nr_deinterleaving_ldpc(E, + harq_process->Qm, + harq_process->d[r], + harq_process->w[r]); + + //for (int i =0; i<16; i++) + // printf("rx output interleaving d[%d]= %d r_offset %d\n", i,harq_process->d[r][i], r_offset); + +#if UE_TIMING_TRACE + stop_meas(dlsch_deinterleaving_stats); +#endif +#ifdef DEBUG_DLSCH_DECODING + + if (r==0) { + write_output("decoder_llr.m","decllr",dlsch_llr,G,1,0); + write_output("decoder_in.m","dec",&harq_process->d[0][0],(3*8*Kr_bytes)+12,1,0); + } + + printf("decoder input(segment %d) :",r); + int i; for (i=0;i<(3*8*Kr_bytes)+12;i++) + printf("%d : %d\n",i,harq_process->d[r][i]); + printf("\n"); +#endif + + + // printf("Clearing c, %p\n",harq_process->c[r]); + memset(harq_process->c[r],0,Kr_bytes); + + // printf("done\n"); + if (harq_process->C == 1) + crc_type = CRC24_A; + else + crc_type = CRC24_B; + + if (err_flag == 0) { +/* + LOG_I(PHY, "turbo algo Kr=%d cb_cnt=%d C=%d nbRB=%d crc_type %d TBSInput=%d TBSHarq=%d TBSplus24=%d mcs=%d Qm=%d RIV=%d round=%d maxIter %d\n", + Kr,r,harq_process->C,harq_process->nb_rb,crc_type,A,harq_process->TBS, + harq_process->B,harq_process->mcs,harq_process->Qm,harq_process->rvidx,harq_process->round,dlsch->max_ldpc_iterations); +*/ + +#if UE_TIMING_TRACE + start_meas(dlsch_turbo_decoding_stats); +#endif + + //LOG_E(PHY,"AbsSubframe %d.%d Start turbo segment %d/%d A %d ",frame%1024,nr_tti_rx,r,harq_process->C-1, A); + + //printf("harq process dr iteration %d\n", p_decParams->numMaxIter); + + for (int cnt =0; cnt < (kc-2)*p_decParams->Z; cnt++){ + inv_d[cnt] = (1)*harq_process->d[r][cnt]; + } + + /*for (int cnt =0; cnt < 16; cnt++){ + printf("dr %d inv_d %d \n", harq_process->d[r][cnt], inv_d[cnt]); + } + + printf(" \n"); + + printf("end dr \n"); + for (int cnt =(50*p_decParams->Z-16) ; cnt < 50*p_decParams->Z; cnt++){ + printf("%d ", harq_process->d[r][cnt]); + } + printf(" \n");*/ + + memset(pv,0,2*harq_process->Z*sizeof(int16_t)); + //memset(pl,0,2*p_decParams->Z*sizeof(int8_t)); + memset((pv+K_bytes_F),127,harq_process->F*sizeof(int16_t)); + + for (i=((2*p_decParams->Z)>>3), j = 0; i < K_bytes_F+((2*p_decParams->Z)>>3); i++, j++) + { + pv[i]= _mm_loadu_si128((__m128i*)(&inv_d[8*j])); + } + + for (i=Kr_bytes+((2*p_decParams->Z)>>3),j=Kr_bytes; i < ((kc*p_decParams->Z)>>3); i++, j++) + { + pv[i]= _mm_loadu_si128((__m128i*)(&inv_d[8*j])); + } + + for (i=0, j=0; j < ((kc*p_decParams->Z)>>4); i+=2, j++) + { + + pl[j] = _mm_packs_epi16(pv[i],pv[i+1]); + + } + + no_iteration_ldpc = nrLDPC_decoder(p_decParams, + (int8_t*)&pl[0], + llrProcBuf, + p_procTime); + + if (check_crc(llrProcBuf,harq_process->B,harq_process->F,crc_type)) { + printf("CRC OK\n"); + ret = 2; + } + else { + printf("CRC NOK\n"); + ret = 1+dlsch->max_ldpc_iterations; + } + + + nb_total_decod++; + if (no_iteration_ldpc > dlsch->max_ldpc_iterations){ + nb_error_decod++; + } + + ret=no_iteration_ldpc; + + //if (!nb_total_decod%10000){ + //printf("Error number of iteration LPDC %d %ld/%ld \n", no_iteration_ldpc, nb_error_decod,nb_total_decod);fflush(stdout); + //} + //else + //printf("OK number of iteration LPDC %d\n", no_iteration_ldpc); + + for (int m=0; m < Kr>>3; m ++) + { + harq_process->c[r][m]= (uint8_t) llrProcBuf[m]; + } + +#ifdef DEBUG_DLSCH_DECODING + //printf("output decoder %d %d %d %d %d \n", harq_process->c[r][0], harq_process->c[r][1], harq_process->c[r][2],harq_process->c[r][3], harq_process->c[r][4]); + for (int k=0;k>3;k++) + printf("output decoder [%d] = 0x%02x \n", k, harq_process->c[r][k]); + printf("no_iterations_ldpc %d (ret %d)\n",no_iteration_ldpc,ret); + //write_output("dec_output.m","dec0",harq_process->c[0],Kr_bytes,1,4); +#endif + + +#if UE_TIMING_TRACE + stop_meas(dlsch_turbo_decoding_stats); +#endif + } + + /*printf("Segmentation: C %d r %d, dlsch_rate_unmatching_stats %5.3f dlsch_deinterleaving_stats %5.3f dlsch_turbo_decoding_stats %5.3f \n", + harq_process->C, + r, + dlsch_rate_unmatching_stats->p_time/(cpuf*1000.0), + dlsch_deinterleaving_stats->p_time/(cpuf*1000.0), + dlsch_turbo_decoding_stats->p_time/(cpuf*1000.0));*/ + + + if ((err_flag == 0) && (ret>=(1+dlsch->max_ldpc_iterations))) {// a Code segment is in error so break; + LOG_D(PHY,"AbsSubframe %d.%d CRC failed, segment %d/%d \n",frame%1024,nr_tti_rx,r,harq_process->C-1); + err_flag = 1; + } + } + + int32_t frame_rx_prev = frame; + int32_t tti_rx_prev = nr_tti_rx - 1; + if (tti_rx_prev < 0) { + frame_rx_prev--; + tti_rx_prev += 10*frame_parms->ttis_per_subframe; + } + frame_rx_prev = frame_rx_prev%1024; + + if (err_flag == 1) { +#if UE_DEBUG_TRACE + LOG_I(PHY,"[UE %d] DLSCH: Setting NAK for SFN/SF %d/%d (pid %d, status %d, round %d, TBS %d, mcs %d) Kr %d r %d harq_process->round %d\n", + phy_vars_ue->Mod_id, frame, nr_tti_rx, harq_pid,harq_process->status, harq_process->round,harq_process->TBS,harq_process->mcs,Kr,r,harq_process->round); +#endif + harq_process->harq_ack.ack = 0; + harq_process->harq_ack.harq_id = harq_pid; + harq_process->harq_ack.send_harq_status = 1; + harq_process->errors[harq_process->round]++; + harq_process->round++; + + + // printf("Rate: [UE %d] DLSCH: Setting NACK for subframe %d (pid %d, round %d)\n",phy_vars_ue->Mod_id,subframe,harq_pid,harq_process->round); + if (harq_process->round >= dlsch->Mdlharq) { + harq_process->status = SCH_IDLE; + harq_process->round = 0; + } + if(is_crnti) + { + LOG_D(PHY,"[UE %d] DLSCH: Setting NACK for nr_tti_rx %d (pid %d, pid status %d, round %d/Max %d, TBS %d)\n", + phy_vars_ue->Mod_id,nr_tti_rx,harq_pid,harq_process->status,harq_process->round,dlsch->Mdlharq,harq_process->TBS); + } + + return((1+dlsch->max_ldpc_iterations)); + } else { +#if UE_DEBUG_TRACE + LOG_I(PHY,"[UE %d] DLSCH: Setting ACK for nr_tti_rx %d TBS %d mcs %d nb_rb %d harq_process->round %d\n", + phy_vars_ue->Mod_id,nr_tti_rx,harq_process->TBS,harq_process->mcs,harq_process->nb_rb, harq_process->round); +#endif + + harq_process->status = SCH_IDLE; + harq_process->round = 0; + harq_process->harq_ack.ack = 1; + harq_process->harq_ack.harq_id = harq_pid; + harq_process->harq_ack.send_harq_status = 1; + //LOG_I(PHY,"[UE %d] DLSCH: Setting ACK for SFN/SF %d/%d (pid %d, status %d, round %d, TBS %d, mcs %d)\n", + // phy_vars_ue->Mod_id, frame, subframe, harq_pid, harq_process->status, harq_process->round,harq_process->TBS,harq_process->mcs); + + if(is_crnti) + { + LOG_D(PHY,"[UE %d] DLSCH: Setting ACK for nr_tti_rx %d (pid %d, round %d, TBS %d)\n",phy_vars_ue->Mod_id,nr_tti_rx,harq_pid,harq_process->round,harq_process->TBS); + } + //LOG_D(PHY,"[UE %d] DLSCH: Setting ACK for subframe %d (pid %d, round %d)\n",phy_vars_ue->Mod_id,subframe,harq_pid,harq_process->round); + + } + + // Reassembly of Transport block here + offset = 0; + Kr = harq_process->K; + Kr_bytes = Kr>>3; + + /* + printf("harq_pid %d\n",harq_pid); + printf("F %d, Fbytes %d\n",harq_process->F,harq_process->F>>3); + printf("C %d\n",harq_process->C); + */ + for (r=0; rC; r++) { + + memcpy(harq_process->b+offset, + harq_process->c[r], + Kr_bytes- - (harq_process->F>>3) -((harq_process->C>1)?3:0)); + offset += (Kr_bytes - (harq_process->F>>3) - ((harq_process->C>1)?3:0)); + +#ifdef DEBUG_DLSCH_DECODING + printf("Segment %d : Kr= %d bytes\n",r,Kr_bytes); + printf("copied %d bytes to b sequence (harq_pid %d)\n", + (Kr_bytes - (harq_process->F>>3)-((harq_process->C>1)?3:0)),harq_pid); + printf("b[0] = %x,c[%d] = %x\n", + harq_process->b[offset], + harq_process->F>>3, + harq_process->c[r]); +#endif + + } + + dlsch->last_iteration_cnt = ret; + + return(ret); +} + +#ifdef UE_DLSCH_PARALLELISATION +uint32_t nr_dlsch_decoding_mthread(PHY_VARS_NR_UE *phy_vars_ue, + UE_nr_rxtx_proc_t *proc, + int eNB_id, + short *dlsch_llr, + NR_DL_FRAME_PARMS *frame_parms, + NR_UE_DLSCH_t *dlsch, + NR_DL_UE_HARQ_t *harq_process, + uint32_t frame, + uint16_t nb_symb_sch, + uint8_t nr_tti_rx, + uint8_t harq_pid, + uint8_t is_crnti, + uint8_t llr8_flag) +{ + +#if UE_TIMING_TRACE + time_stats_t *dlsch_rate_unmatching_stats=&phy_vars_ue->dlsch_rate_unmatching_stats; + time_stats_t *dlsch_turbo_decoding_stats=&phy_vars_ue->dlsch_turbo_decoding_stats; + time_stats_t *dlsch_deinterleaving_stats=&phy_vars_ue->dlsch_deinterleaving_stats; +#endif + uint32_t A,E; + uint32_t G; + uint32_t ret,offset; + //short dummy_w[MAX_NUM_DLSCH_SEGMENTS][3*(8448+64)]; + uint32_t r,r_offset=0,Kr,Kr_bytes,err_flag=0,K_bytes_F; + uint8_t crc_type; + //UE_rxtx_proc_t *proc = &phy_vars_ue->proc; + int32_t no_iteration_ldpc; + int Cby2; + /*uint8_t C; + uint8_t Qm; + uint8_t Nl; + uint8_t r_thread; + uint32_t Er, Gp,GpmodC;*/ + t_nrLDPC_dec_params decParams; + t_nrLDPC_dec_params* p_decParams = &decParams; + t_nrLDPC_time_stats procTime; + t_nrLDPC_time_stats* p_procTime =&procTime ; + int16_t z [68*384]; + int8_t l [68*384]; + //__m128i l; + int16_t inv_d [68*384]; + //int16_t *p_invd =&inv_d; + uint8_t kb, kc; + uint8_t Ilbrm = 0; + uint32_t Tbslbrm = 950984; + uint16_t nb_rb = 30; //to update + //uint16_t nb_symb_sch = 12; + uint8_t nb_re_dmrs = 6; + uint16_t length_dmrs = 1; + + uint32_t i,j; + //uint32_t k; + + __m128i *pv = (__m128i*)&z; + __m128i *pl = (__m128i*)&l; + + + if (!dlsch_llr) { + printf("dlsch_decoding.c: NULL dlsch_llr pointer\n"); + return(dlsch->max_ldpc_iterations); + } + + if (!harq_process) { + printf("dlsch_decoding.c: NULL harq_process pointer\n"); + return(dlsch->max_ldpc_iterations); + } + + if (!frame_parms) { + printf("dlsch_decoding.c: NULL frame_parms pointer\n"); + return(dlsch->max_ldpc_iterations); + } + + /* if (nr_tti_rx> (10*frame_parms->ttis_per_subframe-1)) { + printf("dlsch_decoding.c: Illegal subframe index %d\n",nr_tti_rx); + return(dlsch->max_ldpc_iterations); + } + + if (dlsch->harq_ack[nr_tti_rx].ack != 2) { + LOG_D(PHY, "[UE %d] DLSCH @ SF%d : ACK bit is %d instead of DTX even before PDSCH is decoded!\n", + phy_vars_ue->Mod_id, nr_tti_rx, dlsch->harq_ack[nr_tti_rx].ack); + }*/ + + /* + if (nb_rb > frame_parms->N_RB_DL) { + printf("dlsch_decoding.c: Illegal nb_rb %d\n",nb_rb); + return(max_ldpc_iterations); + }*/ + + /*harq_pid = dlsch->current_harq_pid[phy_vars_ue->current_thread_id[subframe]]; + if (harq_pid >= 8) { + printf("dlsch_decoding.c: Illegal harq_pid %d\n",harq_pid); + return(max_ldpc_iterations); + } + */ + + nb_rb = harq_process->nb_rb; + harq_process->trials[harq_process->round]++; + + harq_process->TBS = nr_compute_tbs(harq_process->mcs,nb_rb,nb_symb_sch,nb_re_dmrs,length_dmrs, harq_process->Nl); + + A = harq_process->TBS; //2072 for QPSK 1/3 + + ret = dlsch->max_ldpc_iterations; + + harq_process->G = nr_get_G(nb_rb, nb_symb_sch, nb_re_dmrs, length_dmrs, harq_process->Qm,harq_process->Nl); + + G = harq_process->G; + + proc->decoder_main_available = 1; + proc->decoder_thread_available = 0; + proc->decoder_thread_available1 = 0; + //get_G(frame_parms,nb_rb,dlsch->rb_alloc,mod_order,num_pdcch_symbols,phy_vars_ue->frame,subframe); + + // printf("DLSCH Decoding, harq_pid %d Ndi %d\n",harq_pid,harq_process->Ndi); + + if (harq_process->round == 0) { + // This is a new packet, so compute quantities regarding segmentation + harq_process->B = A+24; + nr_segmentation(NULL, + NULL, + harq_process->B, + &harq_process->C, + &harq_process->K, + &harq_process->Z, + &harq_process->F); + p_decParams->Z = harq_process->Z; + + } + + kb = harq_process->K/harq_process->Z; + if ( kb==22){ + p_decParams->BG = 1; + p_decParams->R = 89; + kc = 68; + } + else{ + p_decParams->BG = 2; + p_decParams->R = 13; + kc = 52; + } + + p_decParams->numMaxIter = 2; + Kr = p_decParams->Z*kb; + p_decParams->outMode= 0; + + /* + else { + printf("dlsch_decoding.c: Ndi>0 not checked yet!!\n"); + return(max_ldpc_iterations); + } + */ + err_flag = 0; + r_offset = 0; + + unsigned char bw_scaling =1; + + switch (frame_parms->N_RB_DL) { + case 106: + bw_scaling =2; + break; + + default: + bw_scaling =1; + break; + } + + if (harq_process->C > MAX_NUM_DLSCH_SEGMENTS/bw_scaling) { + LOG_E(PHY,"Illegal harq_process->C %d > %d\n",harq_process->C,MAX_NUM_DLSCH_SEGMENTS/bw_scaling); + return((1+dlsch->max_ldpc_iterations)); + } +#ifdef DEBUG_DLSCH_DECODING + printf("Segmentation: C %d, Cminus %d, Kminus %d, Kplus %d\n",harq_process->C,harq_process->Cminus,harq_process->Kminus,harq_process->Kplus); +#endif + + opp_enabled=1; +if (harq_process->C>1) { // wakeup worker if more than 1 segment + if (pthread_mutex_lock(&proc->mutex_dlsch_td) != 0) { + LOG_E( PHY, "[SCHED][UE %d][Slot0] error locking mutex for UE dlsch td\n",phy_vars_ue->Mod_id ); + exit_fun("nothing to add"); + } + + /*Qm= harq_process->Qm; + Nl=harq_process->Nl; + r_thread = harq_process->C/2-1; + C= harq_process->C; + + Gp = G/Nl/Qm; + GpmodC = Gp%C; + + + + if (r_thread < (C-(GpmodC))) + Er = Nl*Qm * (Gp/C); + else + Er = Nl*Qm * ((GpmodC==0?0:1) + (Gp/C)); + printf("mthread Er %d\n", Er); + + printf("mthread instance_cnt_dlsch_td %d\n", proc->instance_cnt_dlsch_td);*/ + + proc->instance_cnt_dlsch_td++; + proc->eNB_id = eNB_id; + proc->harq_pid = harq_pid; + proc->llr8_flag = llr8_flag; + //proc->r[0] = 1; + + if (proc->instance_cnt_dlsch_td == 0) + { + LOG_D(PHY,"unblock dlsch td processing thread blocked on instance_cnt_dlsch_td : %d \n", proc->instance_cnt_dlsch_td ); + if (pthread_cond_signal(&proc->cond_dlsch_td) != 0) { + LOG_E( PHY, "[SCHED][UE %d][Slot0] ERROR pthread_cond_signal for UE dlsch td\n", phy_vars_ue->Mod_id); + exit_fun("nothing to add"); + } + if (pthread_mutex_unlock(&proc->mutex_dlsch_td) != 0) { + LOG_E( PHY, "[SCHED][UE %d][Slot0] error unlocking mutex for UE dlsch td \n",phy_vars_ue->Mod_id ); + exit_fun("nothing to add"); + } + + } else + { + LOG_E( PHY, "[SCHED][UE %d] UE dlsch td thread busy (IC %d)!!\n", phy_vars_ue->Mod_id, proc->instance_cnt_dlsch_td); + if (proc->instance_cnt_dlsch_td > 4) + exit_fun("instance_cnt_dlsch_td > 4"); + } + //AssertFatal(pthread_cond_signal(&proc->cond_slot1_dl_processing) ==0 ,""); + AssertFatal(pthread_mutex_unlock(&proc->mutex_dlsch_td) ==0,""); + + if (harq_process->C>2) { + if (pthread_mutex_lock(&proc->mutex_dlsch_td1) != 0) { + LOG_E( PHY, "[SCHED][UE %d][Slot0] error locking mutex for UE dlsch td\n",phy_vars_ue->Mod_id ); + exit_fun("nothing to add"); + } + + proc->instance_cnt_dlsch_td1++; + proc->eNB_id = eNB_id; + proc->harq_pid = harq_pid; + proc->llr8_flag = llr8_flag; + // proc->Er = Er; + + if (proc->instance_cnt_dlsch_td1 == 0) + { + LOG_D(PHY,"unblock slot1 dl processing thread blocked on instance_cnt_dlsch_td : %d \n", proc->instance_cnt_dlsch_td1 ); + if (pthread_cond_signal(&proc->cond_dlsch_td1) != 0) { + LOG_E( PHY, "[SCHED][UE %d][Slot0] ERROR pthread_cond_signal for UE dlsch td\n", phy_vars_ue->Mod_id); + exit_fun("nothing to add"); + } + if (pthread_mutex_unlock(&proc->mutex_dlsch_td1) != 0) { + LOG_E( PHY, "[SCHED][UE %d][Slot0] error unlocking mutex for UE dlsch td \n",phy_vars_ue->Mod_id ); + exit_fun("nothing to add"); + } + + } else + { + LOG_E( PHY, "[SCHED][UE %d] UE dlsch td thread 1 busy (IC %d)!!\n", phy_vars_ue->Mod_id, proc->instance_cnt_dlsch_td1); + if (proc->instance_cnt_dlsch_td1 > 4) + exit_fun("instance_cnt_dlsch_td1 > 4"); + } + + AssertFatal(pthread_mutex_unlock(&proc->mutex_dlsch_td1) ==0,""); + + } + /* + if (pthread_mutex_timedlock(&proc->mutex_td,&wait) != 0) { + printf("[eNB] ERROR pthread_mutex_lock for TD thread (IC %d)\n", proc->instance_cnt_td); + exit_fun( "error locking mutex_fep" ); + return -1; + } + + if (proc->instance_cnt_td==0) { + printf("[UE] TD thread busy\n"); + exit_fun("TD thread busy"); + pthread_mutex_unlock( &proc->mutex_td ); + return -1; + } + + ++proc->instance_cnt_td; + + proc->tdp.UE = phy_vars_ue; + proc->tdp.eNB_id = eNB_id; + proc->tdp.harq_pid = harq_pid; + proc->tdp.llr8_flag = llr8_flag; + + printf("----- 2thread llr flag %d tdp flag %d\n",llr8_flag, proc->tdp.llr8_flag); + + + // wakeup worker to do second half segments + if (pthread_cond_signal(&proc->cond_td) != 0) { + printf("[UE] ERROR pthread_cond_signal for td thread exit\n"); + exit_fun( "ERROR pthread_cond_signal" ); + return (1+dlsch->last_iteration_cnt); + } + + pthread_mutex_unlock( &proc->mutex_td );*/ + + + Cby2 = 1; //harq_process->C/2; + //proc->decoder_main_available = 1; + } + else { + Cby2 = 1; + } + + for (r=0; rK; + Kr_bytes = Kr>>3; + K_bytes_F = Kr_bytes-(harq_process->F>>3); + + Tbslbrm = nr_compute_tbs(28,nb_rb,frame_parms->symbols_per_slot,0,0, harq_process->Nl); + +#if UE_TIMING_TRACE + start_meas(dlsch_rate_unmatching_stats); +#endif + +#ifdef DEBUG_DLSCH_DECODING + LOG_D(PHY,"HARQ_PID %d Rate Matching Segment %d (coded bits %d,unpunctured/repeated bits %d, TBS %d, mod_order %d, nb_rb %d, Nl %d, rv %d, round %d)...\n", + harq_pid,r, G, + Kr*3, + harq_process->TBS, + harq_process->Qm, + harq_process->nb_rb, + harq_process->Nl, + harq_process->rvidx, + harq_process->round); +#endif + +#ifdef DEBUG_DLSCH_DECODING + printf(" in decoding dlsch->harq_processes[harq_pid]->rvidx = %d\n", dlsch->harq_processes[harq_pid]->rvidx); +#endif + if (nr_rate_matching_ldpc_rx(Ilbrm, + Tbslbrm, + p_decParams->BG, + p_decParams->Z, + G, + harq_process->w[r], + dlsch_llr+r_offset, + harq_process->C, + harq_process->rvidx, + (harq_process->round==0)?1:0, + harq_process->Qm, + harq_process->Nl, + r, + &E)==-1) { +#if UE_TIMING_TRACE + stop_meas(dlsch_rate_unmatching_stats); +#endif + LOG_E(PHY,"dlsch_decoding.c: Problem in rate_matching\n"); + return(dlsch->max_ldpc_iterations); + } else + { +#if UE_TIMING_TRACE + stop_meas(dlsch_rate_unmatching_stats); +#endif + } + r_offset += E; + //printf("main thread r_offset %d\n",r_offset); + + /* + printf("Subblock deinterleaving, d %p w %p\n", + harq_process->d[r], + harq_process->w); + */ +#if UE_TIMING_TRACE + start_meas(dlsch_deinterleaving_stats); +#endif + nr_deinterleaving_ldpc(E, + harq_process->Qm, + harq_process->d[r], + harq_process->w[r]); +#if UE_TIMING_TRACE + stop_meas(dlsch_deinterleaving_stats); +#endif +#ifdef DEBUG_DLSCH_DECODING + + if (r==0) { + write_output("decoder_llr.m","decllr",dlsch_llr,G,1,0); + write_output("decoder_in.m","dec",&harq_process->d[0][96],(3*8*Kr_bytes)+12,1,0); + } + + printf("decoder input(segment %d) :",r); + for (int i=0;i<(3*8*Kr_bytes);i++) + printf("%d : %d\n",i,harq_process->d[r][i]); + printf("\n"); +#endif + + + // printf("Clearing c, %p\n",harq_process->c[r]); + memset(harq_process->c[r],0,Kr_bytes); + + // printf("done\n"); + if (harq_process->C == 1) + crc_type = CRC24_A; + else + crc_type = CRC24_B; + + //#ifndef __AVX2__ + + if (err_flag == 0) { +/* + LOG_I(PHY, "turbo algo Kr=%d cb_cnt=%d C=%d nbRB=%d crc_type %d TBSInput=%d TBSHarq=%d TBSplus24=%d mcs=%d Qm=%d RIV=%d round=%d maxIter %d\n", + Kr,r,harq_process->C,harq_process->nb_rb,crc_type,A,harq_process->TBS, + harq_process->B,harq_process->mcs,harq_process->Qm,harq_process->rvidx,harq_process->round,dlsch->max_ldpc_iterations); +*/ + if (llr8_flag) { + AssertFatal (Kr >= 256, "turbo algo issue Kr=%d cb_cnt=%d C=%d nbRB=%d TBSInput=%d TBSHarq=%d TBSplus24=%d mcs=%d Qm=%d RIV=%d round=%d\n", + Kr,r,harq_process->C,harq_process->nb_rb,A,harq_process->TBS,harq_process->B,harq_process->mcs,harq_process->Qm,harq_process->rvidx,harq_process->round); + } +#if UE_TIMING_TRACE + start_meas(dlsch_turbo_decoding_stats); +#endif + LOG_D(PHY,"mthread AbsSubframe %d.%d Start turbo segment %d/%d \n",frame%1024,nr_tti_rx,r,harq_process->C-1); + + memset(pv,0,2*p_decParams->Z*sizeof(int16_t)); + //memset(pl,0,2*p_decParams->Z*sizeof(int8_t)); + memset((pv+K_bytes_F),127,harq_process->F*sizeof(int16_t)); + + + for (i=((2*p_decParams->Z)>>3), j = 0; i < K_bytes_F+((2*p_decParams->Z)>>3); i++, j++) + { + pv[i]= _mm_loadu_si128((__m128i*)(&inv_d[8*j])); + } + + for (i=Kr_bytes+((2*p_decParams->Z)>>3),j=Kr_bytes; i < ((kc*p_decParams->Z)>>3); i++, j++) + { + pv[i]= _mm_loadu_si128((__m128i*)(&inv_d[8*j])); + } + + for (i=0, j=0; j < ((kc*p_decParams->Z)>>4); i+=2, j++) + { + + pl[j] = _mm_packs_epi16(pv[i],pv[i+1]); + + } + + no_iteration_ldpc = nrLDPC_decoder(p_decParams, + (int8_t*)&pl[0], + llrProcBuf, + p_procTime); + + nb_total_decod++; + if (no_iteration_ldpc > 10){ + nb_error_decod++; + ret = 1+dlsch->max_ldpc_iterations; + } + else { + ret=2; + } + //if (!nb_total_decod%10000){ + printf("Error number of iteration LPDC %d %ld/%ld \n", no_iteration_ldpc, nb_error_decod,nb_total_decod);fflush(stdout); + //} + + //else + //printf("OK number of iteration LPDC %d\n", no_iteration_ldpc); + + for (int m=0; m < Kr>>3; m ++) + { + harq_process->c[r][m]= (uint8_t) llrProcBuf[m]; + } + + /*for (int u=0; u < Kr>>3; u ++) + { + ullrProcBuf[u]= (uint8_t) llrProcBuf[u]; + } + + + printf("output unsigned ullrProcBuf \n"); + + for (int j=0; j < Kr>>3; j ++) + { + + printf(" %d \n", ullrProcBuf[j]); + + } + printf(" \n");*/ + //printf("output channel decoder %d %d %d %d %d \n", harq_process->c[r][0], harq_process->c[r][1], harq_process->c[r][2],harq_process->c[r][3], harq_process->c[r][4]); + + //printf("output decoder %d %d %d %d %d \n", harq_process->c[r][0], harq_process->c[r][1], harq_process->c[r][2],harq_process->c[r][3], harq_process->c[r][4]); + + +#if UE_TIMING_TRACE + stop_meas(dlsch_turbo_decoding_stats); +#endif + } + + + if ((err_flag == 0) && (ret>=(1+dlsch->max_ldpc_iterations))) {// a Code segment is in error so break; + LOG_D(PHY,"AbsSubframe %d.%d CRC failed, segment %d/%d \n",frame%1024,nr_tti_rx,r,harq_process->C-1); + err_flag = 1; + } + } + + int32_t frame_rx_prev = frame; + int32_t tti_rx_prev = nr_tti_rx - 1; + if (tti_rx_prev < 0) { + frame_rx_prev--; + tti_rx_prev += 10*frame_parms->ttis_per_subframe; + } + frame_rx_prev = frame_rx_prev%1024; + + if (err_flag == 1) { +#if UE_DEBUG_TRACE + LOG_I(PHY,"[UE %d] DLSCH: Setting NAK for SFN/SF %d/%d (pid %d, status %d, round %d, TBS %d, mcs %d) Kr %d r %d harq_process->round %d\n", + phy_vars_ue->Mod_id, frame, nr_tti_rx, harq_pid,harq_process->status, harq_process->round,harq_process->TBS,harq_process->mcs,Kr,r,harq_process->round); +#endif + harq_process->harq_ack.ack = 0; + harq_process->harq_ack.harq_id = harq_pid; + harq_process->harq_ack.send_harq_status = 1; + harq_process->errors[harq_process->round]++; + harq_process->round++; + + + // printf("Rate: [UE %d] DLSCH: Setting NACK for subframe %d (pid %d, round %d)\n",phy_vars_ue->Mod_id,subframe,harq_pid,harq_process->round); + if (harq_process->round >= dlsch->Mdlharq) { + harq_process->status = SCH_IDLE; + harq_process->round = 0; + } + if(is_crnti) + { + LOG_D(PHY,"[UE %d] DLSCH: Setting NACK for nr_tti_rx %d (pid %d, pid status %d, round %d/Max %d, TBS %d)\n", + phy_vars_ue->Mod_id,nr_tti_rx,harq_pid,harq_process->status,harq_process->round,dlsch->Mdlharq,harq_process->TBS); + } + + return((1+dlsch->max_ldpc_iterations)); + } else { +#if UE_DEBUG_TRACE + LOG_I(PHY,"[UE %d] DLSCH: Setting ACK for nr_tti_rx %d TBS %d mcs %d nb_rb %d\n", + phy_vars_ue->Mod_id,nr_tti_rx,harq_process->TBS,harq_process->mcs,harq_process->nb_rb); +#endif + + harq_process->status = SCH_IDLE; + harq_process->round = 0; + harq_process->harq_ack.ack = 1; + harq_process->harq_ack.harq_id = harq_pid; + harq_process->harq_ack.send_harq_status = 1; + //LOG_I(PHY,"[UE %d] DLSCH: Setting ACK for SFN/SF %d/%d (pid %d, status %d, round %d, TBS %d, mcs %d)\n", + // phy_vars_ue->Mod_id, frame, subframe, harq_pid, harq_process->status, harq_process->round,harq_process->TBS,harq_process->mcs); + + if(is_crnti) + { + LOG_D(PHY,"[UE %d] DLSCH: Setting ACK for nr_tti_rx %d (pid %d, round %d, TBS %d)\n",phy_vars_ue->Mod_id,nr_tti_rx,harq_pid,harq_process->round,harq_process->TBS); + } + //LOG_D(PHY,"[UE %d] DLSCH: Setting ACK for subframe %d (pid %d, round %d)\n",phy_vars_ue->Mod_id,subframe,harq_pid,harq_process->round); + + } + + // Reassembly of Transport block here + offset = 0; + + /* + printf("harq_pid %d\n",harq_pid); + printf("F %d, Fbytes %d\n",harq_process->F,harq_process->F>>3); + printf("C %d\n",harq_process->C); + */ + uint32_t wait = 0; + if (harq_process->C==2){ + while((proc->decoder_thread_available == 0) ) + { + usleep(1); + wait++; + } + } + else if ((harq_process->C==3) ){ + while((proc->decoder_thread_available == 0) || (proc->decoder_thread_available1 == 0)) + { + usleep(1); + wait++; + } + } + + proc->decoder_main_available = 0; + + for (r=0; rC; r++) { + + Kr = harq_process->K; //to check if same K in all segments + Kr_bytes = Kr>>3; + + memcpy(harq_process->b+offset, + harq_process->c[r], + Kr_bytes- - (harq_process->F>>3) -((harq_process->C>1)?3:0)); + offset += (Kr_bytes - (harq_process->F>>3) - ((harq_process->C>1)?3:0)); + +#ifdef DEBUG_DLSCH_DECODING + printf("Segment %d : Kr= %d bytes\n",r,Kr_bytes); + printf("copied %d bytes to b sequence (harq_pid %d)\n", + (Kr_bytes - (harq_process->F>>3)-((harq_process->C>1)?3:0)),harq_pid); + printf("b[0] = %x,c[%d] = %x\n", + harq_process->b[offset], + harq_process->F>>3, + harq_process->c[r]); +#endif + } + + dlsch->last_iteration_cnt = ret; + //proc->decoder_thread_available = 0; + //proc->decoder_main_available = 0; + + //wait for worker to finish + //wait_on_busy_condition(&proc->mutex_td,&proc->cond_td,&proc->instance_cnt_dlsch td,"dlsch td thread"); + + //return( (ret>proc->tdp.ret) ? ret : proc->tdp.ret ); + + return(ret); +} +#endif + +#ifdef UE_DLSCH_PARALLELISATION +#define FIFO_PRIORITY 39 +void *nr_dlsch_decoding_2thread0(void *arg) +{ + static __thread int UE_dlsch_td_retval; + struct nr_rxtx_thread_data *rtd = arg; + UE_nr_rxtx_proc_t *proc = rtd->proc; + PHY_VARS_NR_UE *phy_vars_ue = rtd->UE; + NR_DL_FRAME_PARMS *frame_parms = &phy_vars_ue->frame_parms; + int llr8_flag1; + int32_t no_iteration_ldpc; + t_nrLDPC_dec_params decParams; + t_nrLDPC_dec_params* p_decParams = &decParams; + t_nrLDPC_time_stats procTime; + t_nrLDPC_time_stats* p_procTime =&procTime ; + int16_t z [68*384]; + int8_t l [68*384]; + //__m128i l; + int16_t inv_d [68*384]; + int16_t *p_invd =&inv_d; + uint8_t kb, kc; + uint8_t Ilbrm = 0; + uint32_t Tbslbrm = 950984; + uint16_t nb_rb = 30; //to update + uint16_t nb_symb_sch = 12; + uint8_t nb_re_dmrs = 6; + uint16_t length_dmrs = 1; + + uint32_t i,j; + uint32_t k; + + __m128i *pv = (__m128i*)&z; + __m128i *pl = (__m128i*)&l; + + proc->instance_cnt_dlsch_td=-1; + proc->nr_tti_rx=proc->sub_frame_start; + + proc->decoder_thread_available = 0; + + char threadname[256]; + sprintf(threadname,"UE_thread_dlsch_td_%d", proc->sub_frame_start); + + cpu_set_t cpuset; + CPU_ZERO(&cpuset); + if ( (proc->sub_frame_start+1)%RX_NB_TH == 0 && threads.dlsch_td_one != -1 ) + CPU_SET(threads.dlsch_td_one, &cpuset); + if ( (proc->sub_frame_start+1)%RX_NB_TH == 1 && threads.dlsch_td_two != -1 ) + CPU_SET(threads.dlsch_td_two, &cpuset); + if ( (proc->sub_frame_start+1)%RX_NB_TH == 2 && threads.dlsch_td_three != -1 ) + CPU_SET(threads.dlsch_td_three, &cpuset); + + + +#if UE_TIMING_TRACE + time_stats_t *dlsch_rate_unmatching_stats=&phy_vars_ue->dlsch_rate_unmatching_stats; + time_stats_t *dlsch_turbo_decoding_stats=&phy_vars_ue->dlsch_turbo_decoding_stats; + time_stats_t *dlsch_deinterleaving_stats=&phy_vars_ue->dlsch_deinterleaving_stats; +#endif + uint32_t A,E; + uint32_t G; + uint32_t ret,offset; +// short dummy_w[MAX_NUM_DLSCH_SEGMENTS][3*(8448+64)]; + uint32_t r,r_offset=0,Kr,Kr_bytes,err_flag=0,K_bytes_F; + uint8_t crc_type; + uint8_t C,Cprime; + uint8_t Qm; + uint8_t Nl; + //uint32_t Er; + + init_thread(900000,1000000 , FIFO_PRIORITY-1, &cpuset, threadname); + while (!oai_exit) { + + //proc->decoder_thread_available = 1; + + if (pthread_mutex_lock(&proc->mutex_dlsch_td) != 0) { + LOG_E( PHY, "[SCHED][UE] error locking mutex for UE dlsch td\n" ); + exit_fun("nothing to add"); + } + while (proc->instance_cnt_dlsch_td < 0) { + // most of the time, the thread is waiting here + pthread_cond_wait( &proc->cond_dlsch_td, &proc->mutex_dlsch_td ); + } + if (pthread_mutex_unlock(&proc->mutex_dlsch_td) != 0) { + LOG_E( PHY, "[SCHED][UE] error unlocking mutex for UE dlsch_td \n" ); + exit_fun("nothing to add"); + } + + uint32_t wait = 0; + while(proc->decoder_main_available == 0) + { + usleep(1); + wait++; + } + + //proc->decoder_thread_available = 0; + //PHY_VARS_NR_UE *phy_vars_ue = tdp->UE; + int eNB_id = proc->eNB_id; + int harq_pid = proc->harq_pid; + llr8_flag1 = proc->llr8_flag; + //r_offset = proc->Er; + //UE_rxtx_proc_t *proc = tdp->proc; + int frame = proc->frame_rx; + int subframe = proc->nr_tti_rx; + NR_UE_DLSCH_t *dlsch = phy_vars_ue->dlsch[phy_vars_ue->current_thread_id[subframe]][eNB_id][0]; + NR_DL_UE_HARQ_t *harq_process = dlsch->harq_processes[harq_pid]; + short *dlsch_llr = phy_vars_ue->pdsch_vars[phy_vars_ue->current_thread_id[subframe]][eNB_id]->llr[0]; + //printf("2thread0 llr flag %d tdp flag %d\n",llr8_flag1, tdp->llr8_flag); + + /* + if (nb_rb > frame_parms->N_RB_DL) { + printf("dlsch_decoding.c: Illegal nb_rb %d\n",nb_rb); + return(max_ldpc_iterations); + }*/ + + /*harq_pid = dlsch->current_harq_pid[phy_vars_ue->current_thread_id[subframe]]; + if (harq_pid >= 8) { + printf("dlsch_decoding.c: Illegal harq_pid %d\n",harq_pid); + return(max_ldpc_iterations); + } + */ + + nb_rb = harq_process->nb_rb; + + harq_process->trials[harq_process->round]++; + + harq_process->TBS = nr_compute_tbs(harq_process->mcs,nb_rb,nb_symb_sch,nb_re_dmrs,length_dmrs, harq_process->Nl); + + A = harq_process->TBS; //2072 for QPSK 1/3 + + ret = dlsch->max_ldpc_iterations; + + harq_process->G = nr_get_G(nb_rb, nb_symb_sch, nb_re_dmrs, length_dmrs, harq_process->Qm,harq_process->Nl); + G = harq_process->G; + + if (harq_process->round == 0) { + // This is a new packet, so compute quantities regarding segmentation + harq_process->B = A+24; + + nr_segmentation(NULL, + NULL, + harq_process->B, + &harq_process->C, + &harq_process->K, + &harq_process->Z, + &harq_process->F); + p_decParams->Z = harq_process->Z; + + } + + kb = harq_process->K/harq_process->Z; + if ( kb==22){ + p_decParams->BG = 1; + p_decParams->R = 89; + kc = 68; + } + else{ + p_decParams->BG = 2; + p_decParams->R = 13; + kc = 52; + } + + p_decParams->numMaxIter = 2; + Kr = p_decParams->Z*kb; + p_decParams->outMode= 0; + + /* + else { + printf("dlsch_decoding.c: Ndi>0 not checked yet!!\n"); + return(max_ldpc_iterations); + } + */ + err_flag = 0; + //r_offset = 0; + + /* + unsigned char bw_scaling =1; + + switch (frame_parms->N_RB_DL) { + case 106: + bw_scaling =2; + break; + + default: + bw_scaling =1; + break; + } + + if (harq_process->C > MAX_NUM_DLSCH_SEGMENTS/bw_scaling) { + LOG_E(PHY,"Illegal harq_process->C %d > %d\n",harq_process->C,MAX_NUM_DLSCH_SEGMENTS/bw_scaling); + return((1+dlsch->max_ldpc_iterations)); + }*/ +#ifdef DEBUG_DLSCH_DECODING + printf("Segmentation: C %d, Cminus %d, Kminus %d, Kplus %d\n",harq_process->C,harq_process->Cminus,harq_process->Kminus,harq_process->Kplus); +#endif + + opp_enabled=1; + + Qm= harq_process->Qm; + Nl=harq_process->Nl; + //r_thread = harq_process->C/2-1; + C= harq_process->C; + + Cprime = C; //assume CBGTI not present + + if (r <= Cprime - ((G/(Nl*Qm))%Cprime) - 1) + r_offset = Nl*Qm*(G/(Nl*Qm*Cprime)); + else + r_offset = Nl*Qm*((G/(Nl*Qm*Cprime))+1); + + // printf("thread0 r_offset %d\n",r_offset); + + //for (r=(harq_process->C/2); rC; r++) { + r=1; //(harq_process->C/2); + + + Kr = harq_process->K; + Kr_bytes = Kr>>3; + K_bytes_F = Kr_bytes-(harq_process->F>>3); + + Tbslbrm = nr_compute_tbs(28,nb_rb,frame_parms->symbols_per_slot,0,0, harq_process->Nl); + +#if UE_TIMING_TRACE + start_meas(dlsch_rate_unmatching_stats); +#endif + +#ifdef DEBUG_DLSCH_DECODING + LOG_D(PHY,"HARQ_PID %d Rate Matching Segment %d (coded bits %d,unpunctured/repeated bits %d, TBS %d, mod_order %d, nb_rb %d, Nl %d, rv %d, round %d)...\n", + harq_pid,r, G, + Kr*3, + harq_process->TBS, + harq_process->Qm, + harq_process->nb_rb, + harq_process->Nl, + harq_process->rvidx, + harq_process->round); +#endif + +#ifdef DEBUG_DLSCH_DECODING + printf(" in decoding dlsch->harq_processes[harq_pid]->rvidx = %d\n", dlsch->harq_processes[harq_pid]->rvidx); +#endif + if (nr_rate_matching_ldpc_rx(Ilbrm, + Tbslbrm, + p_decParams->BG, + p_decParams->Z, + G, + harq_process->w[r], + dlsch_llr+r_offset, + harq_process->C, + harq_process->rvidx, + (harq_process->round==0)?1:0, + harq_process->Qm, + harq_process->Nl, + r, + &E)==-1) { +#if UE_TIMING_TRACE + stop_meas(dlsch_rate_unmatching_stats); +#endif + LOG_E(PHY,"dlsch_decoding.c: Problem in rate_matching\n"); + //return(dlsch->max_ldpc_iterations); + } else + { +#if UE_TIMING_TRACE + stop_meas(dlsch_rate_unmatching_stats); +#endif + } + r_offset += E; + + //for (int i =0; i<16; i++) + // printf("rx output ratematching w[%d]= %d r_offset %d\n", i,harq_process->w[r][i], r_offset); + +#if UE_TIMING_TRACE + start_meas(dlsch_deinterleaving_stats); +#endif + nr_deinterleaving_ldpc(E, + harq_process->Qm, + harq_process->d[r], + harq_process->w[r]); +#if UE_TIMING_TRACE + stop_meas(dlsch_deinterleaving_stats); +#endif +#ifdef DEBUG_DLSCH_DECODING + + if (r==0) { + write_output("decoder_llr.m","decllr",dlsch_llr,G,1,0); + write_output("decoder_in.m","dec",&harq_process->d[0][0],(3*8*Kr_bytes)+12,1,0); + } + + printf("decoder input(segment %d) :",r); + int i; for (i=0;i<(3*8*Kr_bytes)+12;i++) + printf("%d : %d\n",i,harq_process->d[r][i]); + printf("\n"); +#endif + + + // printf("Clearing c, %p\n",harq_process->c[r]); + memset(harq_process->c[r],0,Kr_bytes); + + // printf("done\n"); + if (harq_process->C == 1) + crc_type = CRC24_A; + else + crc_type = CRC24_B; + + +#if 1 + if (err_flag == 0) { +/* + LOG_I(PHY, "turbo algo Kr=%d cb_cnt=%d C=%d nbRB=%d crc_type %d TBSInput=%d TBSHarq=%d TBSplus24=%d mcs=%d Qm=%d RIV=%d round=%d maxIter %d\n", + Kr,r,harq_process->C,harq_process->nb_rb,crc_type,A,harq_process->TBS, + harq_process->B,harq_process->mcs,harq_process->Qm,harq_process->rvidx,harq_process->round,dlsch->max_ldpc_iterations); +*/ + if (llr8_flag1) { + AssertFatal (Kr >= 256, "turbo algo issue Kr=%d cb_cnt=%d C=%d nbRB=%d TBSInput=%d TBSHarq=%d TBSplus24=%d mcs=%d Qm=%d RIV=%d round=%d\n", + Kr,r,harq_process->C,harq_process->nb_rb,A,harq_process->TBS,harq_process->B,harq_process->mcs,harq_process->Qm,harq_process->rvidx,harq_process->round); + } +#if UE_TIMING_TRACE + start_meas(dlsch_turbo_decoding_stats); +#endif +// LOG_D(PHY,"AbsSubframe %d.%d Start turbo segment %d/%d \n",frame%1024,subframe,r,harq_process->C-1); + + for (int cnt =0; cnt < (kc-2)*p_decParams->Z; cnt++){ + inv_d[cnt] = (1)*harq_process->d[r][cnt]; + } + + memset(pv,0,2*p_decParams->Z*sizeof(int16_t)); + //memset(pl,0,2*p_decParams->Z*sizeof(int8_t)); + memset((pv+K_bytes_F),127,harq_process->F*sizeof(int16_t)); + + for (i=((2*p_decParams->Z)>>3), j = 0; i < K_bytes_F+((2*p_decParams->Z)>>3); i++, j++) + { + pv[i]= _mm_loadu_si128((__m128i*)(&inv_d[8*j])); + } + + for (i=Kr_bytes+((2*p_decParams->Z)>>3),j=Kr_bytes; i < ((kc*p_decParams->Z)>>3); i++, j++) + { + pv[i]= _mm_loadu_si128((__m128i*)(&inv_d[8*j])); + } + + for (i=0, j=0; j < ((kc*p_decParams->Z)>>4); i+=2, j++) + { + + pl[j] = _mm_packs_epi16(pv[i],pv[i+1]); + + } + + no_iteration_ldpc = nrLDPC_decoder(p_decParams, + (int8_t*)&pl[0], + llrProcBuf, + p_procTime); + + if (no_iteration_ldpc > 10) + printf("Error number of iteration LPDC %d\n", no_iteration_ldpc); + //else + //printf("OK number of iteration LPDC %d\n", no_iteration_ldpc); + + for (int m=0; m < Kr>>3; m ++) + { + harq_process->c[r][m]= (uint8_t) llrProcBuf[m]; + } + + /*for (int u=0; u < Kr>>3; u ++) + { + ullrProcBuf[u]= (uint8_t) llrProcBuf[u]; + } + + + printf("output unsigned ullrProcBuf \n"); + + for (int j=0; j < Kr>>3; j ++) + { + + printf(" %d \n", ullrProcBuf[j]); + + } + printf(" \n");*/ + +#if UE_TIMING_TRACE + stop_meas(dlsch_turbo_decoding_stats); +#endif + } + + if ((err_flag == 0) && (ret>=(1+dlsch->max_ldpc_iterations))) {// a Code segment is in error so break; +// LOG_D(PHY,"AbsSubframe %d.%d CRC failed, segment %d/%d \n",frame%1024,subframe,r,harq_process->C-1); + err_flag = 1; + } + //} + + /*int32_t frame_rx_prev = frame; + int32_t subframe_rx_prev = subframe - 1; + if (subframe_rx_prev < 0) { + frame_rx_prev--; + subframe_rx_prev += 10; + } + frame_rx_prev = frame_rx_prev%1024;*/ +#if 0 + if (err_flag == 1) { +//#if UE_DEBUG_TRACE + LOG_I(PHY,"[UE %d] THREAD 0 DLSCH: Setting NAK for SFN/SF %d/%d (pid %d, status %d, round %d, TBS %d, mcs %d) Kr %d r %d harq_process->round %d\n", + phy_vars_ue->Mod_id, frame, subframe, harq_pid,harq_process->status, harq_process->round,harq_process->TBS,harq_process->mcs,Kr,r,harq_process->round); +//#endif + dlsch->harq_ack[subframe].ack = 0; + dlsch->harq_ack[subframe].harq_id = harq_pid; + dlsch->harq_ack[subframe].send_harq_status = 1; + harq_process->errors[harq_process->round]++; + harq_process->round++; + + + // printf("Rate: [UE %d] DLSCH: Setting NACK for subframe %d (pid %d, round %d)\n",phy_vars_ue->Mod_id,subframe,harq_pid,harq_process->round); + if (harq_process->round >= dlsch->Mdlharq) { + harq_process->status = SCH_IDLE; + harq_process->round = 0; + } +/* if(is_crnti) + { + LOG_D(PHY,"[UE %d] DLSCH: Setting NACK for subframe %d (pid %d, pid status %d, round %d/Max %d, TBS %d)\n", + phy_vars_ue->Mod_id,subframe,harq_pid,harq_process->status,harq_process->round,dlsch->Mdlharq,harq_process->TBS); + }*/ + + + //return((1+dlsch->max_ldpc_iterations)); + } else { +#if UE_DEBUG_TRACE + LOG_I(PHY,"[UE %d] THREAD 0 DLSCH: Setting ACK for subframe %d TBS %d mcs %d nb_rb %d\n", + phy_vars_ue->Mod_id,subframe,harq_process->TBS,harq_process->mcs,harq_process->nb_rb); +#endif + + harq_process->status = SCH_IDLE; + harq_process->round = 0; + dlsch->harq_ack[subframe].ack = 1; + dlsch->harq_ack[subframe].harq_id = harq_pid; + dlsch->harq_ack[subframe].send_harq_status = 1; + //LOG_I(PHY,"[UE %d] DLSCH: Setting ACK for SFN/SF %d/%d (pid %d, status %d, round %d, TBS %d, mcs %d)\n", + // phy_vars_ue->Mod_id, frame, subframe, harq_pid, harq_process->status, harq_process->round,harq_process->TBS,harq_process->mcs); + +/* if(is_crnti) + { + LOG_D(PHY,"[UE %d] DLSCH: Setting ACK for subframe %d (pid %d, round %d, TBS %d)\n",phy_vars_ue->Mod_id,subframe,harq_pid,harq_process->round,harq_process->TBS); + } + LOG_D(PHY,"[UE %d] DLSCH: Setting ACK for subframe %d (pid %d, round %d)\n",phy_vars_ue->Mod_id,subframe,harq_pid,harq_process->round); + + }*/ + + // Reassembly of Transport block here + offset = 0; + + /* + printf("harq_pid %d\n",harq_pid); + printf("F %d, Fbytes %d\n",harq_process->F,harq_process->F>>3); + printf("C %d\n",harq_process->C); + */ + for (r=0; rC; r++) { + if (rCminus) + Kr = harq_process->Kminus; + else + Kr = harq_process->Kplus; + + Kr_bytes = Kr>>3; + + // printf("Segment %d : Kr= %d bytes\n",r,Kr_bytes); + if (r==0) { + memcpy(harq_process->b, + &harq_process->c[0][(harq_process->F>>3)], + Kr_bytes - (harq_process->F>>3)- ((harq_process->C>1)?3:0)); + offset = Kr_bytes - (harq_process->F>>3) - ((harq_process->C>1)?3:0); + // printf("copied %d bytes to b sequence (harq_pid %d)\n", + // Kr_bytes - (harq_process->F>>3),harq_pid); + // printf("b[0] = %x,c[%d] = %x\n", + // harq_process->b[0], + // harq_process->F>>3, + // harq_process->c[0][(harq_process->F>>3)]); + } else { + memcpy(harq_process->b+offset, + harq_process->c[r], + Kr_bytes- ((harq_process->C>1)?3:0)); + offset += (Kr_bytes - ((harq_process->C>1)?3:0)); + } + } + + dlsch->last_iteration_cnt = ret; + + //return(ret); + } +#endif + + proc->decoder_thread_available = 1; + //proc->decoder_main_available = 0; + + if (pthread_mutex_lock(&proc->mutex_dlsch_td) != 0) { + LOG_E( PHY, "[SCHED][UE] error locking mutex for UE RXTX\n" ); + exit_fun("noting to add"); + } + proc->instance_cnt_dlsch_td--; + if (pthread_mutex_unlock(&proc->mutex_dlsch_td) != 0) { + LOG_E( PHY, "[SCHED][UE] error unlocking mutex for UE td1\n" ); + exit_fun("noting to add"); + } + } + + // thread finished + free(arg); + return &UE_dlsch_td_retval; +} +#endif + +#ifdef UE_DLSCH_PARALLELISATION +#define FIFO_PRIORITY 39 +void *nr_dlsch_decoding_2thread1(void *arg) +{ + static __thread int UE_dlsch_td_retval1; + struct nr_rxtx_thread_data *rtd = arg; + UE_nr_rxtx_proc_t *proc = rtd->proc; + PHY_VARS_NR_UE *phy_vars_ue = rtd->UE; + NR_DL_FRAME_PARMS *frame_parms = &phy_vars_ue->frame_parms; + + int llr8_flag1; + int32_t no_iteration_ldpc; + t_nrLDPC_dec_params decParams; + t_nrLDPC_dec_params* p_decParams = &decParams; + t_nrLDPC_time_stats procTime; + t_nrLDPC_time_stats* p_procTime =&procTime ; + int16_t z [68*384]; + int8_t l [68*384]; + //__m128i l; + int16_t inv_d [68*384]; + //int16_t *p_invd =&inv_d; + uint8_t kb, kc; + uint8_t Ilbrm = 0; + uint32_t Tbslbrm = 950984; + uint16_t nb_rb = 30; //to update + uint16_t nb_symb_sch = 12; + uint8_t nb_re_dmrs = 6; + uint16_t length_dmrs = 1; + + uint32_t i,j; + //uint32_t k; + + __m128i *pv = (__m128i*)&z; + __m128i *pl = (__m128i*)&l; + + proc->instance_cnt_dlsch_td1=-1; + proc->nr_tti_rx=proc->sub_frame_start; + + printf("start thread 1\n"); + proc->decoder_thread_available1 = 0; + + char threadname[256]; + sprintf(threadname,"UE_thread_dlsch_td1_%d", proc->sub_frame_start); + + cpu_set_t cpuset; + CPU_ZERO(&cpuset); + if ( (proc->sub_frame_start+1)%RX_NB_TH == 0 && threads.dlsch_td_one != -1 ) + CPU_SET(threads.dlsch_td_one, &cpuset); + if ( (proc->sub_frame_start+1)%RX_NB_TH == 1 && threads.dlsch_td_two != -1 ) + CPU_SET(threads.dlsch_td_two, &cpuset); + if ( (proc->sub_frame_start+1)%RX_NB_TH == 2 && threads.dlsch_td_three != -1 ) + CPU_SET(threads.dlsch_td_three, &cpuset); + + + +#if UE_TIMING_TRACE + time_stats_t *dlsch_rate_unmatching_stats=&phy_vars_ue->dlsch_rate_unmatching_stats; + time_stats_t *dlsch_turbo_decoding_stats=&phy_vars_ue->dlsch_turbo_decoding_stats; + time_stats_t *dlsch_deinterleaving_stats=&phy_vars_ue->dlsch_deinterleaving_stats; +#endif + uint32_t A,E; + uint32_t G; + uint32_t ret,offset; + uint32_t r,r_offset=0,Kr,Kr_bytes,err_flag=0,K_bytes_F; + uint8_t crc_type; + uint8_t C,Cprime; + uint8_t Qm; + uint8_t Nl; + //uint32_t Er; + + init_thread(900000,1000000 , FIFO_PRIORITY-1, &cpuset, threadname); + printf("2thread1 oai_exit %d\n", oai_exit); + while (!oai_exit) { + if (pthread_mutex_lock(&proc->mutex_dlsch_td1) != 0) { + LOG_E( PHY, "[SCHED][UE] error locking mutex for UE dlsch td\n" ); + exit_fun("nothing to add"); + } + while (proc->instance_cnt_dlsch_td1 < 0) { + // most of the time, the thread is waiting here + pthread_cond_wait( &proc->cond_dlsch_td1, &proc->mutex_dlsch_td1 ); + } + if (pthread_mutex_unlock(&proc->mutex_dlsch_td1) != 0) { + LOG_E( PHY, "[SCHED][UE] error unlocking mutex for UE dlsch_td \n" ); + exit_fun("nothing to add"); + } + + //printf("2thread1 main available %d\n", proc->decoder_main_available); + + uint32_t wait = 0; + while(proc->decoder_main_available == 0) + { + usleep(1); + wait++; + } + + //proc->decoder_thread_available1 = 0; + //PHY_VARS_NR_UE *phy_vars_ue = tdp->UE; + int eNB_id = proc->eNB_id; + int harq_pid = proc->harq_pid; + llr8_flag1 = proc->llr8_flag; + //r_offset = proc->Er; + //UE_rxtx_proc_t *proc = tdp->proc; + int frame = proc->frame_rx; + int subframe = proc->nr_tti_rx; + NR_UE_DLSCH_t *dlsch = phy_vars_ue->dlsch[phy_vars_ue->current_thread_id[subframe]][eNB_id][0]; + NR_DL_UE_HARQ_t *harq_process = dlsch->harq_processes[harq_pid]; + short *dlsch_llr = phy_vars_ue->pdsch_vars[phy_vars_ue->current_thread_id[subframe]][eNB_id]->llr[0]; + //printf("2thread0 llr flag %d tdp flag %d\n",llr8_flag1, tdp->llr8_flag); + //printf("2thread1 nr_tti_tx %d subframe %d SF thread id %d r_offset %d\n", proc->nr_tti_rx, subframe, phy_vars_ue->current_thread_id[subframe], r_offset); + + /* + if (nb_rb > frame_parms->N_RB_DL) { + printf("dlsch_decoding.c: Illegal nb_rb %d\n",nb_rb); + return(max_ldpc_iterations); + }*/ + + /*harq_pid = dlsch->current_harq_pid[phy_vars_ue->current_thread_id[subframe]]; + if (harq_pid >= 8) { + printf("dlsch_decoding.c: Illegal harq_pid %d\n",harq_pid); + return(max_ldpc_iterations); + } + */ + + nb_rb = harq_process->nb_rb; + + harq_process->trials[harq_process->round]++; + harq_process->TBS = nr_compute_tbs(harq_process->mcs,nb_rb,nb_symb_sch,nb_re_dmrs,length_dmrs, harq_process->Nl); + + A = harq_process->TBS; //2072 for QPSK 1/3 + + ret = dlsch->max_ldpc_iterations; + + harq_process->G = nr_get_G(nb_rb, nb_symb_sch, nb_re_dmrs, length_dmrs, harq_process->Qm,harq_process->Nl); + + G = harq_process->G; + //get_G(frame_parms,nb_rb,dlsch->rb_alloc,mod_order,num_pdcch_symbols,phy_vars_ue->frame,subframe); + + //printf("DLSCH Decoding, A %d harq_pid %d G %d\n",A, harq_pid,harq_process->G); + + if (harq_process->round == 0) { + // This is a new packet, so compute quantities regarding segmentation + harq_process->B = A+24; + nr_segmentation(NULL, + NULL, + harq_process->B, + &harq_process->C, + &harq_process->K, + &harq_process->Z, + &harq_process->F); + p_decParams->Z = harq_process->Z; + + } + + kb = harq_process->K/harq_process->Z; + if ( kb==22){ + p_decParams->BG = 1; + p_decParams->R = 89; + kc = 68; + } + else{ + p_decParams->BG = 2; + p_decParams->R = 13; + kc = 52; + } + + p_decParams->numMaxIter = 2; + Kr = p_decParams->Z*kb; + p_decParams->outMode= 0; + /* + else { + printf("dlsch_decoding.c: Ndi>0 not checked yet!!\n"); + return(max_ldpc_iterations); + } + */ + err_flag = 0; + //r_offset = 0; + + /* + unsigned char bw_scaling =1; + + switch (frame_parms->N_RB_DL) { + case 106: + bw_scaling =2; + break; + + default: + bw_scaling =1; + break; + } + + if (harq_process->C > MAX_NUM_DLSCH_SEGMENTS/bw_scaling) { + LOG_E(PHY,"Illegal harq_process->C %d > %d\n",harq_process->C,MAX_NUM_DLSCH_SEGMENTS/bw_scaling); + return((1+dlsch->max_ldpc_iterations)); + }*/ +#ifdef DEBUG_DLSCH_DECODING + printf("Segmentation: C %d, Cminus %d, Kminus %d, Kplus %d\n",harq_process->C,harq_process->Cminus,harq_process->Kminus,harq_process->Kplus); +#endif + + opp_enabled=1; + + Qm= harq_process->Qm; + Nl=harq_process->Nl; + //r_thread = harq_process->C/2-1; + C= harq_process->C; + + Cprime = C; //assume CBGTI not present + + if (r <= Cprime - ((G/(Nl*Qm))%Cprime) - 1) + r_offset = Nl*Qm*(G/(Nl*Qm*Cprime)); + else + r_offset = Nl*Qm*((G/(Nl*Qm*Cprime))+1); + + //printf("sub thread r_offset %d\n", r_offset); + + //for (r=(harq_process->C/2); rC; r++) { + r=2; //(harq_process->C/2); + r_offset = r*r_offset; + //printf("thread1 r=%d r_offset %d \n",r, r_offset); + + + Kr = harq_process->K; + Kr_bytes = Kr>>3; + K_bytes_F = Kr_bytes-(harq_process->F>>3); + + Tbslbrm = nr_compute_tbs(28,nb_rb,frame_parms->symbols_per_slot,0,0, harq_process->Nl); + +#if UE_TIMING_TRACE + start_meas(dlsch_rate_unmatching_stats); +#endif + +#ifdef DEBUG_DLSCH_DECODING + LOG_D(PHY,"HARQ_PID %d Rate Matching Segment %d (coded bits %d,unpunctured/repeated bits %d, TBS %d, mod_order %d, nb_rb %d, Nl %d, rv %d, round %d)...\n", + harq_pid,r, G, + Kr*3, + harq_process->TBS, + harq_process->Qm, + harq_process->nb_rb, + harq_process->Nl, + harq_process->rvidx, + harq_process->round); +#endif + +#ifdef DEBUG_DLSCH_DECODING + printf(" in decoding dlsch->harq_processes[harq_pid]->rvidx = %d\n", dlsch->harq_processes[harq_pid]->rvidx); +#endif + if (nr_rate_matching_ldpc_rx(Ilbrm, + Tbslbrm, + p_decParams->BG, + p_decParams->Z, + G, + harq_process->w[r], + dlsch_llr+r_offset, + harq_process->C, + harq_process->rvidx, + (harq_process->round==0)?1:0, + harq_process->Qm, + harq_process->Nl, + r, + &E)==-1) { +#if UE_TIMING_TRACE + stop_meas(dlsch_rate_unmatching_stats); +#endif + LOG_E(PHY,"dlsch_decoding.c: Problem in rate_matching\n"); + //return(dlsch->max_ldpc_iterations); + } else + { +#if UE_TIMING_TRACE + stop_meas(dlsch_rate_unmatching_stats); +#endif + } + r_offset += E; + + /* + printf("Subblock deinterleaving, d %p w %p\n", + harq_process->d[r], + harq_process->w); + */ +#if UE_TIMING_TRACE + start_meas(dlsch_deinterleaving_stats); +#endif + nr_deinterleaving_ldpc(E, + harq_process->Qm, + harq_process->d[r], + harq_process->w[r]); +#if UE_TIMING_TRACE + stop_meas(dlsch_deinterleaving_stats); +#endif +#ifdef DEBUG_DLSCH_DECODING + if (r==0) { + write_output("decoder_llr.m","decllr",dlsch_llr,G,1,0); + write_output("decoder_in.m","dec",&harq_process->d[0][0],(3*8*Kr_bytes)+12,1,0); + } + + printf("decoder input(segment %d) :",r); + int i; for (i=0;i<(3*8*Kr_bytes)+12;i++) + printf("%d : %d\n",i,harq_process->d[r][i]); + printf("\n"); +#endif + + + // printf("Clearing c, %p\n",harq_process->c[r]); + memset(harq_process->c[r],0,Kr_bytes); + + // printf("done\n"); + if (harq_process->C == 1) + crc_type = CRC24_A; + else + crc_type = CRC24_B; + + if (err_flag == 0) { +/* + LOG_I(PHY, "turbo algo Kr=%d cb_cnt=%d C=%d nbRB=%d crc_type %d TBSInput=%d TBSHarq=%d TBSplus24=%d mcs=%d Qm=%d RIV=%d round=%d maxIter %d\n", + Kr,r,harq_process->C,harq_process->nb_rb,crc_type,A,harq_process->TBS, + harq_process->B,harq_process->mcs,harq_process->Qm,harq_process->rvidx,harq_process->round,dlsch->max_ldpc_iterations); +*/ + if (llr8_flag1) { + AssertFatal (Kr >= 256, "turbo algo issue Kr=%d cb_cnt=%d C=%d nbRB=%d TBSInput=%d TBSHarq=%d TBSplus24=%d mcs=%d Qm=%d RIV=%d round=%d\n", + Kr,r,harq_process->C,harq_process->nb_rb,A,harq_process->TBS,harq_process->B,harq_process->mcs,harq_process->Qm,harq_process->rvidx,harq_process->round); + } +#if UE_TIMING_TRACE + start_meas(dlsch_turbo_decoding_stats); +#endif +// LOG_D(PHY,"AbsSubframe %d.%d Start turbo segment %d/%d \n",frame%1024,subframe,r,harq_process->C-1); + + memset(pv,0,2*p_decParams->Z*sizeof(int16_t)); + //memset(pl,0,2*p_decParams->Z*sizeof(int8_t)); + memset((pv+K_bytes_F),127,harq_process->F*sizeof(int16_t)); + + for (i=((2*p_decParams->Z)>>3), j = 0; i < K_bytes_F+((2*p_decParams->Z)>>3); i++, j++) + { + pv[i]= _mm_loadu_si128((__m128i*)(&inv_d[8*j])); + } + + for (i=Kr_bytes+((2*p_decParams->Z)>>3),j=Kr_bytes; i < ((kc*p_decParams->Z)>>3); i++, j++) + { + pv[i]= _mm_loadu_si128((__m128i*)(&inv_d[8*j])); + } + + for (i=0, j=0; j < ((kc*p_decParams->Z)>>4); i+=2, j++) + { + + pl[j] = _mm_packs_epi16(pv[i],pv[i+1]); + + } + + no_iteration_ldpc = nrLDPC_decoder(p_decParams, + (int8_t*)&pl[0], + llrProcBuf, + p_procTime); + + if (no_iteration_ldpc > 10) + printf("Error number of iteration LPDC %d\n", no_iteration_ldpc); + //else + // printf("OK number of iteration LPDC %d\n", no_iteration_ldpc); + + for (int m=0; m < Kr>>3; m ++) + { + harq_process->c[r][m]= (uint8_t) llrProcBuf[m]; + } + + /*for (int u=0; u < Kr>>3; u ++) + { + ullrProcBuf[u]= (uint8_t) llrProcBuf[u]; + } + + + printf("output unsigned ullrProcBuf \n"); + + for (int j=0; j < Kr>>3; j ++) + { + + printf(" %d \n", ullrProcBuf[j]); + + } + printf(" \n");*/ +#endif + //printf("output channel decoder %d %d %d %d %d \n", harq_process->c[r][0], harq_process->c[r][1], harq_process->c[r][2],harq_process->c[r][3], harq_process->c[r][4]); + + //printf("output decoder %d %d %d %d %d \n", harq_process->c[r][0], harq_process->c[r][1], harq_process->c[r][2],harq_process->c[r][3], harq_process->c[r][4]); + +#if UE_TIMING_TRACE + stop_meas(dlsch_turbo_decoding_stats); +#endif + } + + if ((err_flag == 0) && (ret>=(1+dlsch->max_ldpc_iterations))) {// a Code segment is in error so break; +// LOG_D(PHY,"AbsSubframe %d.%d CRC failed, segment %d/%d \n",frame%1024,subframe,r,harq_process->C-1); + err_flag = 1; + } + //} + + /*int32_t frame_rx_prev = frame; + int32_t subframe_rx_prev = subframe - 1; + if (subframe_rx_prev < 0) { + frame_rx_prev--; + subframe_rx_prev += 10; + } + frame_rx_prev = frame_rx_prev%1024;*/ + + + proc->decoder_thread_available1 = 1; + //proc->decoder_main_available = 0; + + //printf("2thread1 proc->instance_cnt_dlsch_td1 %d\n", proc->instance_cnt_dlsch_td1); + + if (pthread_mutex_lock(&proc->mutex_dlsch_td1) != 0) { + LOG_E( PHY, "[SCHED][UE] error locking mutex for UE RXTX\n" ); + exit_fun("noting to add"); + } + proc->instance_cnt_dlsch_td1--; + if (pthread_mutex_unlock(&proc->mutex_dlsch_td1) != 0) { + LOG_E( PHY, "[SCHED][UE] error unlocking mutex for UE td1\n" ); + exit_fun("noting to add"); + } + //printf("end 2thread1 proc->instance_cnt_dlsch_td1 %d\n", proc->instance_cnt_dlsch_td1); + } + + //printf("after 2thread1 after oai exit proc->instance_cnt_dlsch_td %d\n", proc->instance_cnt_dlsch_td1); + // thread finished + free(arg); + return &UE_dlsch_td_retval1; +} +#endif + diff --git a/openair1/PHY/NR_UE_TRANSPORT/nr_dlsch_demodulation.c b/openair1/PHY/NR_UE_TRANSPORT/nr_dlsch_demodulation.c new file mode 100644 index 0000000000..2e00c128b3 --- /dev/null +++ b/openair1/PHY/NR_UE_TRANSPORT/nr_dlsch_demodulation.c @@ -0,0 +1,2236 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.0 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +/*! \file PHY/LTE_TRANSPORT/dlsch_demodulation.c + * \brief Top-level routines for demodulating the PDSCH physical channel from 38-211, V15.2 2018-06 + * \author H.Wang + * \date 2018 + * \version 0.1 + * \company Eurecom + * \note + * \warning + */ +#include "PHY/defs_nr_UE.h" +#include "PHY/phy_extern_nr_ue.h" +#include "nr_transport_proto_ue.h" +//#include "SCHED/defs.h" +//#include "PHY/defs.h" +//#include "extern.h" +#include "PHY/sse_intrin.h" +#include "T.h" + +#ifndef USER_MODE +#define NOCYGWIN_STATIC static +#else +#define NOCYGWIN_STATIC +#endif + +/* dynamic shift for LLR computation for TM3/4 + * set as command line argument, see lte-softmodem.c + * default value: 0 + */ +int16_t nr_dlsch_demod_shift = 0; +//int16_t interf_unaw_shift = 13; + +//#define DEBUG_HARQ + +//#define DEBUG_PHY 1 +//#define DEBUG_DLSCH_DEMOD 1 + + + +// [MCS][i_mod (0,1,2) = (2,4,6)] +//unsigned char offset_mumimo_llr_drange_fix=0; +//inferference-free case +/*unsigned char interf_unaw_shift_tm4_mcs[29]={5, 3, 4, 3, 3, 2, 1, 1, 2, 0, 1, 1, 1, 1, 0, 0, + 1, 1, 1, 1, 0, 2, 1, 0, 1, 0, 1, 0, 0} ;*/ + +//unsigned char interf_unaw_shift_tm1_mcs[29]={5, 5, 4, 3, 3, 3, 2, 2, 4, 4, 2, 3, 3, 3, 1, 1, +// 0, 1, 1, 2, 5, 4, 4, 6, 5, 1, 0, 5, 6} ; // mcs 21, 26, 28 seem to be errorneous + +/* +unsigned char offset_mumimo_llr_drange[29][3]={{8,8,8},{7,7,7},{7,7,7},{7,7,7},{6,6,6},{6,6,6},{6,6,6},{5,5,5},{4,4,4},{1,2,4}, // QPSK +{5,5,4},{5,5,5},{5,5,5},{3,3,3},{2,2,2},{2,2,2},{2,2,2}, // 16-QAM +{2,2,1},{3,3,3},{3,3,3},{3,3,1},{2,2,2},{2,2,2},{0,0,0},{0,0,0},{0,0,0},{0,0,0},{0,0,0},{0,0,0}}; //64-QAM +*/ + /* + //first optimization try + unsigned char offset_mumimo_llr_drange[29][3]={{7, 8, 7},{6, 6, 7},{6, 6, 7},{6, 6, 6},{5, 6, 6},{5, 5, 6},{5, 5, 6},{4, 5, 4},{4, 3, 4},{3, 2, 2},{6, 5, 5},{5, 4, 4},{5, 5, 4},{3, 3, 2},{2, 2, 1},{2, 1, 1},{2, 2, 2},{3, 3, 3},{3, 3, 2},{3, 3, 2},{3, 2, 1},{2, 2, 2},{2, 2, 2},{0, 0, 0},{0, 0, 0},{0, 0, 0},{0, 0, 0},{0, 0, 0}}; + */ + //second optimization try + /* + unsigned char offset_mumimo_llr_drange[29][3]={{5, 8, 7},{4, 6, 8},{3, 6, 7},{7, 7, 6},{4, 7, 8},{4, 7, 4},{6, 6, 6},{3, 6, 6},{3, 6, 6},{1, 3, 4},{1, 1, 0},{3, 3, 2},{3, 4, 1},{4, 0, 1},{4, 2, 2},{3, 1, 2},{2, 1, 0},{2, 1, 1},{1, 0, 1},{1, 0, 1},{0, 0, 0},{1, 0, 0},{0, 0, 0},{0, 1, 0},{1, 0, 0},{0, 0, 0},{0, 0, 0},{0, 0, 0},{0, 0, 0}}; w + */ +//unsigned char offset_mumimo_llr_drange[29][3]= {{0, 6, 5},{0, 4, 5},{0, 4, 5},{0, 5, 4},{0, 5, 6},{0, 5, 3},{0, 4, 4},{0, 4, 4},{0, 3, 3},{0, 1, 2},{1, 1, 0},{1, 3, 2},{3, 4, 1},{2, 0, 0},{2, 2, 2},{1, 1, 1},{2, 1, 0},{2, 1, 1},{1, 0, 1},{1, 0, 1},{0, 0, 0},{1, 0, 0},{0, 0, 0},{0, 1, 0},{1, 0, 0},{0, 0, 0},{0, 0, 0},{0, 0, 0},{0, 0, 0}}; + + +extern void print_shorts(char *s,int16_t *x); + + +int nr_rx_pdsch(PHY_VARS_NR_UE *ue, + PDSCH_t type, + unsigned char eNB_id, + unsigned char eNB_id_i, //if this == ue->n_connected_eNB, we assume MU interference + uint32_t frame, + uint8_t nr_tti_rx, + unsigned char symbol, + unsigned char first_symbol_flag, + RX_type_t rx_type, + unsigned char i_mod, + unsigned char harq_pid) +{ + + NR_UE_COMMON *common_vars = &ue->common_vars; + NR_UE_PDSCH **pdsch_vars; + NR_DL_FRAME_PARMS *frame_parms = &ue->frame_parms; + PHY_NR_MEASUREMENTS *measurements = &ue->measurements; + NR_UE_DLSCH_t **dlsch; + + int avg[4]; +// int avg_0[2]; +// int avg_1[2]; + +#if UE_TIMING_TRACE + uint8_t slot = 0; +#endif + + unsigned char aatx,aarx; + + unsigned short nb_rb = 0, nb_re =0, round; + int avgs = 0;// rb; + NR_DL_UE_HARQ_t *dlsch0_harq,*dlsch1_harq = 0; + + uint8_t beamforming_mode; + uint32_t *rballoc; + + int32_t **rxdataF_comp_ptr; + int32_t **dl_ch_mag_ptr; + int32_t codeword_TB0 = -1; + int32_t codeword_TB1 = -1; + + //to be updated higher layer + unsigned short start_rb = 0; + unsigned short nb_rb_pdsch = 50; + int16_t *pllr_symbol_cw0; + int16_t *pllr_symbol_cw1; + int16_t *pllr_symbol_cw0_deint; + int16_t *pllr_symbol_cw1_deint; + uint32_t llr_offset_symbol; + uint16_t bundle_L = 2; + uint8_t l0 =2; + + switch (type) { + case SI_PDSCH: + pdsch_vars = &ue->pdsch_vars_SI[eNB_id]; + dlsch = &ue->dlsch_SI[eNB_id]; + dlsch0_harq = dlsch[0]->harq_processes[harq_pid]; + beamforming_mode = 0; + break; + + case RA_PDSCH: + pdsch_vars = &ue->pdsch_vars_ra[eNB_id]; + dlsch = &ue->dlsch_ra[eNB_id]; + dlsch0_harq = dlsch[0]->harq_processes[harq_pid]; + beamforming_mode = 0; + break; + + case PDSCH: + pdsch_vars = ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]]; + dlsch = ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id]; + + + //set active for testing -> to be removed + dlsch[0]->harq_processes[harq_pid]->status = ACTIVE; + dlsch[0]->harq_processes[harq_pid]->Qm = 2; + dlsch[0]->harq_processes[harq_pid]->mcs = 9; + dlsch[0]->harq_processes[harq_pid]->Nl=1; + dlsch[0]->harq_processes[harq_pid]->round=0; + dlsch[0]->harq_processes[harq_pid]->nb_rb = nb_rb_pdsch; + frame_parms->nushift = 0; + + //printf("status TB0 = %d, status TB1 = %d \n", dlsch[0]->harq_processes[harq_pid]->status, dlsch[1]->harq_processes[harq_pid]->status); + LOG_D(PHY,"AbsSubframe %d.%d / Sym %d harq_pid %d, harq status %d.%d \n", + frame,nr_tti_rx,symbol,harq_pid, + dlsch[0]->harq_processes[harq_pid]->status, + dlsch[1]->harq_processes[harq_pid]->status); + + if ((dlsch[0]->harq_processes[harq_pid]->status == ACTIVE) && + (dlsch[1]->harq_processes[harq_pid]->status == ACTIVE)){ + codeword_TB0 = dlsch[0]->harq_processes[harq_pid]->codeword; + codeword_TB1 = dlsch[1]->harq_processes[harq_pid]->codeword; + dlsch0_harq = dlsch[codeword_TB0]->harq_processes[harq_pid]; + dlsch1_harq = dlsch[codeword_TB1]->harq_processes[harq_pid]; +#ifdef DEBUG_HARQ + printf("[DEMOD] I am assuming both TBs are active\n"); +#endif + } + else if ((dlsch[0]->harq_processes[harq_pid]->status == ACTIVE) && + (dlsch[1]->harq_processes[harq_pid]->status != ACTIVE) ) { + codeword_TB0 = dlsch[0]->harq_processes[harq_pid]->codeword; + dlsch0_harq = dlsch[0]->harq_processes[harq_pid]; + dlsch1_harq = NULL; + codeword_TB1 = -1; +#ifdef DEBUG_HARQ + printf("[DEMOD] I am assuming only TB0 is active\n"); +#endif + } + else if ((dlsch[0]->harq_processes[harq_pid]->status != ACTIVE) && + (dlsch[1]->harq_processes[harq_pid]->status == ACTIVE) ){ + codeword_TB1 = dlsch[1]->harq_processes[harq_pid]->codeword; + dlsch0_harq = dlsch[1]->harq_processes[harq_pid]; + dlsch1_harq = NULL; + codeword_TB0 = -1; +#ifdef DEBUG_HARQ + printf("[DEMOD] I am assuming only TB1 is active, it is in cw %d\n", dlsch0_harq->codeword); +#endif + } + else { + LOG_E(PHY,"[UE][FATAL] Frame %d nr_tti_rx %d: no active DLSCH\n",ue->proc.proc_rxtx[0].frame_rx,nr_tti_rx); + return(-1); + } + beamforming_mode = ue->transmission_mode[eNB_id]<7?0:ue->transmission_mode[eNB_id]; + break; + + default: + LOG_E(PHY,"[UE][FATAL] Frame %d nr_tti_rx %d: Unknown PDSCH format %d\n",ue->proc.proc_rxtx[0].frame_rx,nr_tti_rx,type); + return(-1); + break; + } +#ifdef DEBUG_HARQ + printf("[DEMOD] MIMO mode = %d\n", dlsch0_harq->mimo_mode); + printf("[DEMOD] cw for TB0 = %d, cw for TB1 = %d\n", codeword_TB0, codeword_TB1); +#endif + + DevAssert(dlsch0_harq); + round = dlsch0_harq->round; + //printf("round = %d\n", round); + + if (eNB_id > 2) { + LOG_W(PHY,"dlsch_demodulation.c: Illegal eNB_id %d\n",eNB_id); + return(-1); + } + + if (!common_vars) { + LOG_W(PHY,"dlsch_demodulation.c: Null common_vars\n"); + return(-1); + } + + if (!dlsch[0]) { + LOG_W(PHY,"dlsch_demodulation.c: Null dlsch_ue pointer\n"); + return(-1); + } + + if (!pdsch_vars) { + LOG_W(PHY,"dlsch_demodulation.c: Null pdsch_vars pointer\n"); + return(-1); + } + + if (!frame_parms) { + LOG_W(PHY,"dlsch_demodulation.c: Null frame_parms\n"); + return(-1); + } + + if (((frame_parms->Ncp == NORMAL) && (symbol>=7)) || + ((frame_parms->Ncp == EXTENDED) && (symbol>=6))) + rballoc = dlsch0_harq->rb_alloc_odd; + else + rballoc = dlsch0_harq->rb_alloc_even; + + + if (dlsch0_harq->mimo_mode>DUALSTREAM_PUSCH_PRECODING) { + LOG_E(PHY,"This transmission mode is not yet supported!\n"); + return(-1); + } + + + if ((dlsch0_harq->mimo_mode==LARGE_CDD) || ((dlsch0_harq->mimo_mode>=DUALSTREAM_UNIFORM_PRECODING1) && (dlsch0_harq->mimo_mode<=DUALSTREAM_PUSCH_PRECODING))) { + DevAssert(dlsch1_harq); + if (eNB_id!=eNB_id_i) { + LOG_E(PHY,"TM3/TM4 requires to set eNB_id==eNB_id_i!\n"); + return(-1); + } + } + +#if UE_TIMING_TRACE + if(symbol > ue->frame_parms.symbols_per_tti>>1) + { + slot = 1; + } +#endif + +#ifdef DEBUG_HARQ + printf("Demod dlsch0_harq->pmi_alloc %d\n", dlsch0_harq->pmi_alloc); +#endif +#if 0 + if (frame_parms->nb_antenna_ports_eNB>1 && beamforming_mode==0) { +#ifdef DEBUG_DLSCH_MOD + LOG_I(PHY,"dlsch: using pmi %x (%p), rb_alloc %x\n",pmi2hex_2Ar1(dlsch0_harq->pmi_alloc),dlsch[0],dlsch0_harq->rb_alloc_even[0]); +#endif + +#if UE_TIMING_TRACE + start_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#endif + nb_rb = dlsch_extract_rbs_dual(common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF, + common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[eNB_id], + pdsch_vars[eNB_id]->rxdataF_ext, + pdsch_vars[eNB_id]->dl_ch_estimates_ext, + dlsch0_harq->pmi_alloc, + pdsch_vars[eNB_id]->pmi_ext, + rballoc, + symbol, + nr_tti_rx, + ue->high_speed_flag, + frame_parms, + dlsch0_harq->mimo_mode); +#ifdef DEBUG_DLSCH_MOD + printf("dlsch: using pmi %lx, rb_alloc %x, pmi_ext ",pmi2hex_2Ar1(dlsch0_harq->pmi_alloc),*rballoc); + for (rb=0;rbpmi_ext[rb]); + printf("\n"); +#endif + + + if (rx_type >= rx_IC_single_stream) { + if (eNB_id_in_connected_eNB) // we are in TM5 + nb_rb = dlsch_extract_rbs_dual(common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF, + common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[eNB_id_i], + pdsch_vars[eNB_id_i]->rxdataF_ext, + pdsch_vars[eNB_id_i]->dl_ch_estimates_ext, + dlsch0_harq->pmi_alloc, + pdsch_vars[eNB_id_i]->pmi_ext, + rballoc, + symbol, + nr_tti_rx, + ue->high_speed_flag, + frame_parms, + dlsch0_harq->mimo_mode); + else + nb_rb = dlsch_extract_rbs_dual(common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF, + common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[eNB_id], + pdsch_vars[eNB_id_i]->rxdataF_ext, + pdsch_vars[eNB_id_i]->dl_ch_estimates_ext, + dlsch0_harq->pmi_alloc, + pdsch_vars[eNB_id_i]->pmi_ext, + rballoc, + symbol, + nr_tti_rx, + ue->high_speed_flag, + frame_parms, + dlsch0_harq->mimo_mode); + } + } else + +#endif + if (beamforming_mode==0) { //else if nb_antennas_ports_eNB==1 && beamforming_mode == 0 + //printf("start nr dlsch extract nr_tti_rx %d thread id %d \n", nr_tti_rx, ue->current_thread_id[nr_tti_rx]); + nb_rb = nr_dlsch_extract_rbs_single(common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF, + common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[eNB_id], + pdsch_vars[eNB_id]->rxdataF_ext, + pdsch_vars[eNB_id]->dl_ch_estimates_ext, + dlsch0_harq->pmi_alloc, + pdsch_vars[eNB_id]->pmi_ext, + rballoc, + symbol, + start_rb, + nb_rb_pdsch, + nr_tti_rx, + ue->high_speed_flag, + frame_parms); + + +/* + if (rx_type==rx_IC_single_stream) { + if (eNB_id_in_connected_eNB) + nb_rb = dlsch_extract_rbs_single(common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF, + common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[eNB_id_i], + pdsch_vars[eNB_id_i]->rxdataF_ext, + pdsch_vars[eNB_id_i]->dl_ch_estimates_ext, + dlsch0_harq->pmi_alloc, + pdsch_vars[eNB_id_i]->pmi_ext, + rballoc, + symbol, + nr_tti_rx, + ue->high_speed_flag, + frame_parms); + else + nb_rb = dlsch_extract_rbs_single(common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF, + common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[eNB_id], + pdsch_vars[eNB_id_i]->rxdataF_ext, + pdsch_vars[eNB_id_i]->dl_ch_estimates_ext, + dlsch0_harq->pmi_alloc, + pdsch_vars[eNB_id_i]->pmi_ext, + rballoc, + symbol, + nr_tti_rx, + ue->high_speed_flag, + frame_parms); + }*/ + } /*else if (beamforming_mode==7) { //else if beamforming_mode == 7 + nb_rb = dlsch_extract_rbs_TM7(common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF, + pdsch_vars[eNB_id]->dl_bf_ch_estimates, + pdsch_vars[eNB_id]->rxdataF_ext, + pdsch_vars[eNB_id]->dl_bf_ch_estimates_ext, + rballoc, + symbol, + nr_tti_rx, + ue->high_speed_flag, + frame_parms); + + } else if(beamforming_mode>7) { + LOG_W(PHY,"dlsch_demodulation: beamforming mode not supported yet.\n"); + }*/ + + //printf("nb_rb = %d, eNB_id %d\n",nb_rb,eNB_id); + if (nb_rb==0) { + LOG_D(PHY,"dlsch_demodulation.c: nb_rb=0\n"); + return(-1); + } + +#if UE_TIMING_TRACE + stop_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#if DISABLE_LOG_X + printf("[AbsSFN %d.%d] Slot%d Symbol %d Flag %d type %d: Pilot/Data extraction %5.2f \n",frame,nr_tti_rx,slot, + symbol,ue->high_speed_flag,type,symbol,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#else + LOG_I(PHY, "[AbsSFN %d.%d] Slot%d Symbol %d Flag %d type %d: Pilot/Data extraction %5.2f \n",frame,nr_tti_rx,slot,symbol, + ue->high_speed_flag,type,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#endif +#endif + +#if UE_TIMING_TRACE + start_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#endif + aatx = frame_parms->nb_antenna_ports_eNB; + aarx = frame_parms->nb_antennas_rx; + + nr_dlsch_scale_channel(pdsch_vars[eNB_id]->dl_ch_estimates_ext, + frame_parms, + dlsch, + symbol, + nb_rb); + + if ((dlsch0_harq->mimo_moden_connected_eNB) && + (dlsch0_harq->dl_power_off==0) + ) // TM5 two-user + { + nr_dlsch_scale_channel(pdsch_vars[eNB_id_i]->dl_ch_estimates_ext, + frame_parms, + dlsch, + symbol, + nb_rb); + } + +#if UE_TIMING_TRACE + stop_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#if DISABLE_LOG_X + printf("[AbsSFN %d.%d] Slot%d Symbol %d: Channel Scale %5.2f \n",frame,nr_tti_rx,slot,symbol,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#else + LOG_I(PHY, "[AbsSFN %d.%d] Slot%d Symbol %d: Channel Scale %5.2f \n",frame,nr_tti_rx,slot,symbol,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#endif +#endif + +#if UE_TIMING_TRACE + start_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#endif + if (first_symbol_flag==1) { + if (beamforming_mode==0){ + if (dlsch0_harq->mimo_modedl_ch_estimates_ext, + frame_parms, + avg, + symbol, + nb_rb); + avgs = 0; + for (aatx=0;aatxnb_antenna_ports_eNB;aatx++) + for (aarx=0;aarxnb_antennas_rx;aarx++) + avgs = cmax(avgs,avg[(aatx<<1)+aarx]); + + pdsch_vars[eNB_id]->log2_maxh = (log2_approx(avgs)/2)+1; + }/* + else if ((dlsch0_harq->mimo_mode == LARGE_CDD) || + ((dlsch0_harq->mimo_mode >=DUALSTREAM_UNIFORM_PRECODING1) && + (dlsch0_harq->mimo_mode <=DUALSTREAM_PUSCH_PRECODING))) + { + dlsch_channel_level_TM34(pdsch_vars[eNB_id]->dl_ch_estimates_ext, + frame_parms, + pdsch_vars[eNB_id]->pmi_ext, + avg_0, + avg_1, + symbol, + nb_rb, + dlsch0_harq->mimo_mode); + + LOG_D(PHY,"Channel Level TM34 avg_0 %d, avg_1 %d, rx_type %d, rx_standard %d, dlsch_demod_shift %d \n", avg_0[0], + avg_1[0], rx_type, rx_standard, dlsch_demod_shift); + if (rx_type>rx_standard) { + avg_0[0] = (log2_approx(avg_0[0])/2) + dlsch_demod_shift;// + 2 ;//+ 4; + avg_1[0] = (log2_approx(avg_1[0])/2) + dlsch_demod_shift;// + 2 ;//+ 4; + pdsch_vars[eNB_id]->log2_maxh0 = cmax(avg_0[0],0); + pdsch_vars[eNB_id]->log2_maxh1 = cmax(avg_1[0],0); + // printf("dlsch_demod_shift %d\n", dlsch_demod_shift); + } + else { + avg_0[0] = (log2_approx(avg_0[0])/2) - 13 + interf_unaw_shift; + avg_1[0] = (log2_approx(avg_1[0])/2) - 13 + interf_unaw_shift; + pdsch_vars[eNB_id]->log2_maxh0 = cmax(avg_0[0],0); + pdsch_vars[eNB_id]->log2_maxh1 = cmax(avg_1[0],0); + } + } + else if (dlsch0_harq->mimo_moden_connected_eNB) && (dlsch0_harq->dl_power_off==0)) { + dlsch_channel_level_TM56(pdsch_vars[eNB_id]->dl_ch_estimates_ext, + frame_parms, + pdsch_vars[eNB_id]->pmi_ext, + avg, + symbol, + nb_rb); + avg[0] = log2_approx(avg[0]) - 13 + offset_mumimo_llr_drange[dlsch0_harq->mcs][(i_mod>>1)-1]; + pdsch_vars[eNB_id]->log2_maxh = cmax(avg[0],0); + + } + else if (dlsch0_harq->dl_power_off==1) { //TM6 + + nr_dlsch_channel_level(pdsch_vars[eNB_id]->dl_ch_estimates_ext, + frame_parms, + avg, + symbol, + nb_rb); + + avgs = 0; + for (aatx=0;aatxnb_antenna_ports_eNB;aatx++) + for (aarx=0;aarxnb_antennas_rx;aarx++) + avgs = cmax(avgs,avg[(aatx<<1)+aarx]); + + pdsch_vars[eNB_id]->log2_maxh = (log2_approx(avgs)/2) + 1; + pdsch_vars[eNB_id]->log2_maxh++; + + } + } + + } + else if (beamforming_mode==7) + dlsch_channel_level_TM7(pdsch_vars[eNB_id]->dl_bf_ch_estimates_ext, + frame_parms, + avg, + symbol, + nb_rb);*/ + } +//#ifdef UE_DEBUG_TRACE + LOG_I(PHY,"[DLSCH] AbsSubframe %d.%d log2_maxh = %d [log2_maxh0 %d log2_maxh1 %d] (%d,%d)\n", + frame%1024,nr_tti_rx, pdsch_vars[eNB_id]->log2_maxh, + pdsch_vars[eNB_id]->log2_maxh0, + pdsch_vars[eNB_id]->log2_maxh1, + avg[0],avgs); + //LOG_D(PHY,"[DLSCH] mimo_mode = %d\n", dlsch0_harq->mimo_mode); +//#endif + + //wait until pdcch is decoded + //proc->channel_level = 1; + } + + /* + uint32_t wait = 0; + while(proc->channel_level == 0) + { + usleep(1); + wait++; + } + */ + +#if T_TRACER + if (type == PDSCH) + { + T(T_UE_PHY_PDSCH_ENERGY, T_INT(eNB_id), T_INT(0), T_INT(frame%1024), T_INT(nr_tti_rx), + T_INT(avg[0]), T_INT(avg[1]), T_INT(avg[2]), T_INT(avg[3])); + } +#endif + +#if UE_TIMING_TRACE + stop_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#if DISABLE_LOG_X + printf("[AbsSFN %d.%d] Slot%d Symbol %d first_symbol_flag %d: Channel Level %5.2f \n",frame,nr_tti_rx,slot,symbol,first_symbol_flag,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#else + LOG_I(PHY, "[AbsSFN %d.%d] Slot%d Symbol %d first_symbol_flag %d: Channel Level %5.2f \n",frame,nr_tti_rx,slot,symbol,first_symbol_flag,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#endif +#endif + + +#if UE_TIMING_TRACE + start_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#endif +// Now channel compensation + if (dlsch0_harq->mimo_moderxdataF_ext, + pdsch_vars[eNB_id]->dl_ch_estimates_ext, + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_magb0, + pdsch_vars[eNB_id]->rxdataF_comp0, + (aatx>1) ? pdsch_vars[eNB_id]->rho : NULL, + frame_parms, + symbol, + first_symbol_flag, + dlsch0_harq->Qm, + nb_rb, + pdsch_vars[eNB_id]->log2_maxh, + measurements); // log2_maxh+I0_shift + /*if (symbol == 5) { + write_output("rxF_comp_d.m","rxF_c_d",&pdsch_vars[eNB_id]->rxdataF_comp0[0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1); + } */ + + /*if ((rx_type==rx_IC_single_stream) && + (eNB_id_in_connected_eNB)) { + nr_dlsch_channel_compensation(pdsch_vars[eNB_id_i]->rxdataF_ext, + pdsch_vars[eNB_id_i]->dl_ch_estimates_ext, + pdsch_vars[eNB_id_i]->dl_ch_mag0, + pdsch_vars[eNB_id_i]->dl_ch_magb0, + pdsch_vars[eNB_id_i]->rxdataF_comp0, + (aatx>1) ? pdsch_vars[eNB_id_i]->rho : NULL, + frame_parms, + symbol, + first_symbol_flag, + i_mod, + nb_rb, + pdsch_vars[eNB_id]->log2_maxh, + measurements); // log2_maxh+I0_shift +#ifdef DEBUG_PHY + if (symbol == 5) { + write_output("rxF_comp_d.m","rxF_c_d",&pdsch_vars[eNB_id]->rxdataF_comp0[0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1); + write_output("rxF_comp_i.m","rxF_c_i",&pdsch_vars[eNB_id_i]->rxdataF_comp0[0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1); + } +#endif + + dlsch_dual_stream_correlation(frame_parms, + symbol, + nb_rb, + pdsch_vars[eNB_id]->dl_ch_estimates_ext, + pdsch_vars[eNB_id_i]->dl_ch_estimates_ext, + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->log2_maxh); + }*/ + } + +#if 0 + else if ((dlsch0_harq->mimo_mode == LARGE_CDD) || ((dlsch0_harq->mimo_mode >=DUALSTREAM_UNIFORM_PRECODING1) && + (dlsch0_harq->mimo_mode <=DUALSTREAM_PUSCH_PRECODING))){ + dlsch_channel_compensation_TM34(frame_parms, + pdsch_vars[eNB_id], + measurements, + eNB_id, + symbol, + dlsch0_harq->Qm, + dlsch1_harq->Qm, + harq_pid, + dlsch0_harq->round, + dlsch0_harq->mimo_mode, + nb_rb, + pdsch_vars[eNB_id]->log2_maxh0, + pdsch_vars[eNB_id]->log2_maxh1); + /* if (symbol == 5) { + write_output("rxF_comp_d00.m","rxF_c_d00",&pdsch_vars[eNB_id]->rxdataF_comp0[0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1);// should be QAM + write_output("rxF_comp_d01.m","rxF_c_d01",&pdsch_vars[eNB_id]->rxdataF_comp0[1][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1);//should be almost 0 + write_output("rxF_comp_d10.m","rxF_c_d10",&pdsch_vars[eNB_id]->rxdataF_comp1[harq_pid][round][0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1);//should be almost 0 + write_output("rxF_comp_d11.m","rxF_c_d11",&pdsch_vars[eNB_id]->rxdataF_comp1[harq_pid][round][1][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1);//should be QAM + } */ + // compute correlation between signal and interference channels (rho12 and rho21) + dlsch_dual_stream_correlation(frame_parms, // this is doing h11'*h12 and h21'*h22 + symbol, + nb_rb, + pdsch_vars[eNB_id]->dl_ch_estimates_ext, + &(pdsch_vars[eNB_id]->dl_ch_estimates_ext[2]), + pdsch_vars[eNB_id]->dl_ch_rho2_ext, + pdsch_vars[eNB_id]->log2_maxh0); + //printf("rho stream1 =%d\n", &pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round] ); + //to be optimized (just take complex conjugate) + dlsch_dual_stream_correlation(frame_parms, // this is doing h12'*h11 and h22'*h21 + symbol, + nb_rb, + &(pdsch_vars[eNB_id]->dl_ch_estimates_ext[2]), + pdsch_vars[eNB_id]->dl_ch_estimates_ext, + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->log2_maxh1); + // printf("rho stream2 =%d\n",&pdsch_vars[eNB_id]->dl_ch_rho2_ext ); + //printf("TM3 log2_maxh : %d\n",pdsch_vars[eNB_id]->log2_maxh); + /* if (symbol == 5) { + write_output("rho0_0.m","rho0_0",&pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round][0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1);// should be QAM + write_output("rho2_0.m","rho2_0",&pdsch_vars[eNB_id]->dl_ch_rho2_ext[0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1);//should be almost 0 + write_output("rho0_1.m.m","rho0_1",&pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round][1][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1);//should be almost 0 + write_output("rho2_1.m","rho2_1",&pdsch_vars[eNB_id]->dl_ch_rho2_ext[1][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1);//should be QAM + } */ + + } else if (dlsch0_harq->mimo_moden_connected_eNB) && (dlsch0_harq->dl_power_off==0)) { + dlsch_channel_compensation_TM56(pdsch_vars[eNB_id]->rxdataF_ext, + pdsch_vars[eNB_id]->dl_ch_estimates_ext, + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_magb0, + pdsch_vars[eNB_id]->rxdataF_comp0, + pdsch_vars[eNB_id]->pmi_ext, + frame_parms, + measurements, + eNB_id, + symbol, + dlsch0_harq->Qm, + nb_rb, + pdsch_vars[eNB_id]->log2_maxh, + dlsch0_harq->dl_power_off); + + for (rb=0; rbpmi_ext[rb]) { + case 0: + pdsch_vars[eNB_id_i]->pmi_ext[rb]=1; + break; + case 1: + pdsch_vars[eNB_id_i]->pmi_ext[rb]=0; + break; + case 2: + pdsch_vars[eNB_id_i]->pmi_ext[rb]=3; + break; + case 3: + pdsch_vars[eNB_id_i]->pmi_ext[rb]=2; + break; + } + // if (rb==0) + // printf("pmi %d, pmi_i %d\n",pdsch_vars[eNB_id]->pmi_ext[rb],pdsch_vars[eNB_id_i]->pmi_ext[rb]); + } + dlsch_channel_compensation_TM56(pdsch_vars[eNB_id_i]->rxdataF_ext, + pdsch_vars[eNB_id_i]->dl_ch_estimates_ext, + pdsch_vars[eNB_id_i]->dl_ch_mag0, + pdsch_vars[eNB_id_i]->dl_ch_magb0, + pdsch_vars[eNB_id_i]->rxdataF_comp0, + pdsch_vars[eNB_id_i]->pmi_ext, + frame_parms, + measurements, + eNB_id_i, + symbol, + i_mod, + nb_rb, + pdsch_vars[eNB_id]->log2_maxh, + dlsch0_harq->dl_power_off); +#ifdef DEBUG_PHY + if (symbol==5) { + write_output("rxF_comp_d.m","rxF_c_d",&pdsch_vars[eNB_id]->rxdataF_comp0[0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1); + write_output("rxF_comp_i.m","rxF_c_i",&pdsch_vars[eNB_id_i]->rxdataF_comp0[0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1); + } +#endif + dlsch_dual_stream_correlation(frame_parms, + symbol, + nb_rb, + pdsch_vars[eNB_id]->dl_ch_estimates_ext, + pdsch_vars[eNB_id_i]->dl_ch_estimates_ext, + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->log2_maxh); + } else if (dlsch0_harq->dl_power_off==1) { + dlsch_channel_compensation_TM56(pdsch_vars[eNB_id]->rxdataF_ext, + pdsch_vars[eNB_id]->dl_ch_estimates_ext, + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_magb0, + pdsch_vars[eNB_id]->rxdataF_comp0, + pdsch_vars[eNB_id]->pmi_ext, + frame_parms, + measurements, + eNB_id, + symbol, + dlsch0_harq->Qm, + nb_rb, + pdsch_vars[eNB_id]->log2_maxh, + 1); + + } + + + } else if (dlsch0_harq->mimo_mode==TM7) { //TM7 + + nr_dlsch_channel_compensation(pdsch_vars[eNB_id]->rxdataF_ext, + pdsch_vars[eNB_id]->dl_bf_ch_estimates_ext, + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_magb0, + pdsch_vars[eNB_id]->rxdataF_comp0, + (aatx>1) ? pdsch_vars[eNB_id]->rho : NULL, + frame_parms, + symbol, + first_symbol_flag, + get_Qm(dlsch0_harq->mcs), + nb_rb, + //9, + pdsch_vars[eNB_id]->log2_maxh, + measurements); // log2_maxh+I0_shift + } +#endif + +#if UE_TIMING_TRACE + stop_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#if DISABLE_LOG_X + printf("[AbsSFN %d.%d] Slot%d Symbol %d log2_maxh %d channel_level %d: Channel Comp %5.2f \n",frame,nr_tti_rx,slot,symbol,pdsch_vars[eNB_id]->log2_maxh,proc->channel_level,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#else + LOG_I(PHY, "[AbsSFN %d.%d] Slot%d Symbol %d log2_maxh %d channel_level %d: Channel Comp %5.2f \n",frame,nr_tti_rx,slot,symbol,pdsch_vars[eNB_id]->log2_maxh,proc->channel_level,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#endif +#endif +// MRC +#if UE_TIMING_TRACE + start_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#endif + +#if 0 + if (frame_parms->nb_antennas_rx > 1) { + if ((dlsch0_harq->mimo_mode == LARGE_CDD) || + ((dlsch0_harq->mimo_mode >=DUALSTREAM_UNIFORM_PRECODING1) && + (dlsch0_harq->mimo_mode <=DUALSTREAM_PUSCH_PRECODING))){ // TM3 or TM4 + if (frame_parms->nb_antenna_ports_eNB == 2) { + dlsch_detection_mrc_TM34(frame_parms, + pdsch_vars[eNB_id], + harq_pid, + dlsch0_harq->round, + symbol, + nb_rb, + 1); + /* if (symbol == 5) { + write_output("rho0_mrc.m","rho0_0",&pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round][0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1);// should be QAM + write_output("rho2_mrc.m","rho2_0",&pdsch_vars[eNB_id]->dl_ch_rho2_ext[0][symbol*frame_parms->N_RB_DL*12],frame_parms->N_RB_DL*12,1,1);//should be almost 0 + } */ + } + } else { + dlsch_detection_mrc(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + pdsch_vars[eNB_id_i]->rxdataF_comp0, + pdsch_vars[eNB_id]->rho, + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_magb0, + pdsch_vars[eNB_id_i]->dl_ch_mag0, + pdsch_vars[eNB_id_i]->dl_ch_magb0, + symbol, + nb_rb, + rx_type==rx_IC_single_stream); + } + } + // printf("Combining"); + if ((dlsch0_harq->mimo_mode == SISO) || + ((dlsch0_harq->mimo_mode >= UNIFORM_PRECODING11) && + (dlsch0_harq->mimo_mode <= PUSCH_PRECODING0)) || + (dlsch0_harq->mimo_mode == TM7)) { + /* + dlsch_siso(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp, + pdsch_vars[eNB_id_i]->rxdataF_comp, + symbol, + nb_rb); + */ + } else if (dlsch0_harq->mimo_mode == ALAMOUTI) { + dlsch_alamouti(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_magb0, + symbol, + nb_rb); + } +#endif + + //printf("start compute LLR\n"); + if ((dlsch0_harq->mimo_mode == LARGE_CDD) || + ((dlsch0_harq->mimo_mode >=DUALSTREAM_UNIFORM_PRECODING1) && + (dlsch0_harq->mimo_mode <=DUALSTREAM_PUSCH_PRECODING))) { + rxdataF_comp_ptr = pdsch_vars[eNB_id]->rxdataF_comp1[harq_pid][round]; + dl_ch_mag_ptr = pdsch_vars[eNB_id]->dl_ch_mag1[harq_pid][round]; + } + else { + rxdataF_comp_ptr = pdsch_vars[eNB_id_i]->rxdataF_comp0; + dl_ch_mag_ptr = pdsch_vars[eNB_id_i]->dl_ch_mag0; + //i_mod should have been passed as a parameter + } + +#if UE_TIMING_TRACE + stop_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#if DISABLE_LOG_X + printf("[AbsSFN %d.%d] Slot%d Symbol %d: Channel Combine %5.2f \n",frame,nr_tti_rx,slot,symbol,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#else + LOG_I(PHY, "[AbsSFN %d.%d] Slot%d Symbol %d: Channel Combine %5.2f \n",frame,nr_tti_rx,slot,symbol,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#endif +#endif + +#if UE_TIMING_TRACE + start_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#endif + //printf("LLR dlsch0_harq->Qm %d rx_type %d cw0 %d cw1 %d symbol %d \n",dlsch0_harq->Qm,rx_type,codeword_TB0,codeword_TB1,symbol); + // compute LLRs + // -> // compute @pointer where llrs should filled for this ofdm-symbol + + pdsch_vars[eNB_id]->llr_offset[l0-1] = 0; + llr_offset_symbol = pdsch_vars[eNB_id]->llr_offset[symbol-1]; + //pllr_symbol_cw0_deint = (int8_t*)pdsch_vars[eNB_id]->llr[0]; + //pllr_symbol_cw1_deint = (int8_t*)pdsch_vars[eNB_id]->llr[1]; + pllr_symbol_cw0 = pdsch_vars[eNB_id]->llr[0]; + pllr_symbol_cw1 = pdsch_vars[eNB_id]->llr[1]; + pllr_symbol_cw0 += llr_offset_symbol; + pllr_symbol_cw1 += llr_offset_symbol; + + nb_re= (symbol==l0)? (nb_rb*6):(nb_rb*12); + pdsch_vars[eNB_id]->llr_offset[symbol] = nb_re*dlsch0_harq->Qm + llr_offset_symbol; + + /*LOG_I(PHY,"compute LLRs [symbol %d] NbRB %d Qm %d LLRs-Length %d LLR-Offset %d @LLR Buff %x @LLR Buff(symb) %x\n", + symbol, + nb_rb,dlsch0_harq->Qm, + pdsch_vars[eNB_id]->llr_length[symbol], + pdsch_vars[eNB_id]->llr_offset[symbol], + (int16_t*)pdsch_vars[eNB_id]->llr[0], + pllr_symbol_cw0);*/ + + /*printf("compute LLRs [symbol %d] NbRB %d Qm %d LLRs-Length %d LLR-Offset %d @LLR Buff %p @LLR Buff(symb) %p\n", + symbol, + nb_rb,dlsch0_harq->Qm, + pdsch_vars[eNB_id]->llr_length[symbol], + pdsch_vars[eNB_id]->llr_offset[symbol], + pdsch_vars[eNB_id]->llr[0], + pllr_symbol_cw0);*/ + + switch (dlsch0_harq->Qm) { + case 2 : + if ((rx_type==rx_standard) || (codeword_TB1 == -1)) { + nr_dlsch_qpsk_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + pllr_symbol_cw0, + symbol, + first_symbol_flag, + nb_rb, + beamforming_mode); + + } else if (codeword_TB0 == -1){ + + nr_dlsch_qpsk_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + pllr_symbol_cw1, + symbol, + first_symbol_flag, + nb_rb, + beamforming_mode); + } + else if (rx_type >= rx_IC_single_stream) { + if (dlsch1_harq->Qm == 2) { + nr_dlsch_qpsk_qpsk_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + rxdataF_comp_ptr, + pdsch_vars[eNB_id]->dl_ch_rho2_ext, + pdsch_vars[eNB_id]->llr[0], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch0_harq->rb_alloc_even,2,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128); + if (rx_type==rx_IC_dual_stream) { + nr_dlsch_qpsk_qpsk_llr(frame_parms, + rxdataF_comp_ptr, + pdsch_vars[eNB_id]->rxdataF_comp0, + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->llr[1], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch1_harq->rb_alloc_even,2,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128_2ndstream); + } + } + else if (dlsch1_harq->Qm == 4) { + nr_dlsch_qpsk_16qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + rxdataF_comp_ptr,//i + dl_ch_mag_ptr,//i + pdsch_vars[eNB_id]->dl_ch_rho2_ext, + pdsch_vars[eNB_id]->llr[0], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch0_harq->rb_alloc_even,2,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128); + if (rx_type==rx_IC_dual_stream) { + nr_dlsch_16qam_qpsk_llr(frame_parms, + rxdataF_comp_ptr, + pdsch_vars[eNB_id]->rxdataF_comp0,//i + dl_ch_mag_ptr, + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->llr[1], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch1_harq->rb_alloc_even,4,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128_2ndstream); + } + } + else { + nr_dlsch_qpsk_64qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + rxdataF_comp_ptr,//i + dl_ch_mag_ptr,//i + pdsch_vars[eNB_id]->dl_ch_rho2_ext, + pdsch_vars[eNB_id]->llr[0], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch0_harq->rb_alloc_even,2,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128); + if (rx_type==rx_IC_dual_stream) { + nr_dlsch_64qam_qpsk_llr(frame_parms, + rxdataF_comp_ptr, + pdsch_vars[eNB_id]->rxdataF_comp0,//i + dl_ch_mag_ptr, + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->llr[1], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch1_harq->rb_alloc_even,6,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128_2ndstream); + } + } + } + break; + case 4 : + if ((rx_type==rx_standard ) || (codeword_TB1 == -1)) { + nr_dlsch_16qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + pdsch_vars[eNB_id]->llr[0], + pdsch_vars[eNB_id]->dl_ch_mag0, + symbol,first_symbol_flag,nb_rb, + pdsch_vars[eNB_id]->llr128, + beamforming_mode); + } else if (codeword_TB0 == -1){ + nr_dlsch_16qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + pdsch_vars[eNB_id]->llr[1], + pdsch_vars[eNB_id]->dl_ch_mag0, + symbol,first_symbol_flag,nb_rb, + pdsch_vars[eNB_id]->llr128_2ndstream, + beamforming_mode); + } + else if (rx_type >= rx_IC_single_stream) { + if (dlsch1_harq->Qm == 2) { + nr_dlsch_16qam_qpsk_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + rxdataF_comp_ptr,//i + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_rho2_ext, + pdsch_vars[eNB_id]->llr[0], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch0_harq->rb_alloc_even,4,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128); + if (rx_type==rx_IC_dual_stream) { + nr_dlsch_qpsk_16qam_llr(frame_parms, + rxdataF_comp_ptr, + pdsch_vars[eNB_id]->rxdataF_comp0,//i + pdsch_vars[eNB_id]->dl_ch_mag0,//i + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->llr[1], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch1_harq->rb_alloc_even,2,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128_2ndstream); + } + } + else if (dlsch1_harq->Qm == 4) { + nr_dlsch_16qam_16qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + rxdataF_comp_ptr,//i + pdsch_vars[eNB_id]->dl_ch_mag0, + dl_ch_mag_ptr,//i + pdsch_vars[eNB_id]->dl_ch_rho2_ext, + pdsch_vars[eNB_id]->llr[0], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch0_harq->rb_alloc_even,4,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128); + if (rx_type==rx_IC_dual_stream) { + nr_dlsch_16qam_16qam_llr(frame_parms, + rxdataF_comp_ptr, + pdsch_vars[eNB_id]->rxdataF_comp0,//i + dl_ch_mag_ptr, + pdsch_vars[eNB_id]->dl_ch_mag0,//i + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->llr[1], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch1_harq->rb_alloc_even,4,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128_2ndstream); + } + } + else { + nr_dlsch_16qam_64qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + rxdataF_comp_ptr,//i + pdsch_vars[eNB_id]->dl_ch_mag0, + dl_ch_mag_ptr,//i + pdsch_vars[eNB_id]->dl_ch_rho2_ext, + pdsch_vars[eNB_id]->llr[0], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch0_harq->rb_alloc_even,4,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128); + if (rx_type==rx_IC_dual_stream) { + nr_dlsch_64qam_16qam_llr(frame_parms, + rxdataF_comp_ptr, + pdsch_vars[eNB_id]->rxdataF_comp0, + dl_ch_mag_ptr, + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->llr[1], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch1_harq->rb_alloc_even,6,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128_2ndstream); + } + } + } + break; + case 6 : + if ((rx_type==rx_standard) || (codeword_TB1 == -1)) { + nr_dlsch_64qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + (int16_t*)pllr_symbol_cw0, + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_magb0, + symbol,first_symbol_flag,nb_rb, + pdsch_vars[eNB_id]->llr_offset[symbol], + beamforming_mode); + } else if (codeword_TB0 == -1){ + nr_dlsch_64qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + pllr_symbol_cw1, + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_magb0, + symbol,first_symbol_flag,nb_rb, + pdsch_vars[eNB_id]->llr_offset[symbol], + beamforming_mode); + } + else if (rx_type >= rx_IC_single_stream) { + if (dlsch1_harq->Qm == 2) { + nr_dlsch_64qam_qpsk_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + rxdataF_comp_ptr,//i + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_rho2_ext, + pdsch_vars[eNB_id]->llr[0], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch0_harq->rb_alloc_even,6,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128); + if (rx_type==rx_IC_dual_stream) { + nr_dlsch_qpsk_64qam_llr(frame_parms, + rxdataF_comp_ptr, + pdsch_vars[eNB_id]->rxdataF_comp0,//i + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->llr[1], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch1_harq->rb_alloc_even,2,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128_2ndstream); + } + } + else if (dlsch1_harq->Qm == 4) { + nr_dlsch_64qam_16qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + rxdataF_comp_ptr,//i + pdsch_vars[eNB_id]->dl_ch_mag0, + dl_ch_mag_ptr,//i + pdsch_vars[eNB_id]->dl_ch_rho2_ext, + pdsch_vars[eNB_id]->llr[0], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch0_harq->rb_alloc_even,6,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128); + if (rx_type==rx_IC_dual_stream) { + nr_dlsch_16qam_64qam_llr(frame_parms, + rxdataF_comp_ptr, + pdsch_vars[eNB_id]->rxdataF_comp0,//i + dl_ch_mag_ptr, + pdsch_vars[eNB_id]->dl_ch_mag0,//i + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pdsch_vars[eNB_id]->llr[1], + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch1_harq->rb_alloc_even,4,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr128_2ndstream); + } + } + else { + nr_dlsch_64qam_64qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + rxdataF_comp_ptr,//i + pdsch_vars[eNB_id]->dl_ch_mag0, + dl_ch_mag_ptr,//i + pdsch_vars[eNB_id]->dl_ch_rho2_ext, + (int16_t*)pllr_symbol_cw0, + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch0_harq->rb_alloc_even,6,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr_offset[symbol]); + if (rx_type==rx_IC_dual_stream) { + nr_dlsch_64qam_64qam_llr(frame_parms, + rxdataF_comp_ptr, + pdsch_vars[eNB_id]->rxdataF_comp0,//i + dl_ch_mag_ptr, + pdsch_vars[eNB_id]->dl_ch_mag0,//i + pdsch_vars[eNB_id]->dl_ch_rho_ext[harq_pid][round], + pllr_symbol_cw1, + symbol,first_symbol_flag,nb_rb, + adjust_G2(frame_parms,dlsch1_harq->rb_alloc_even,6,nr_tti_rx,symbol), + pdsch_vars[eNB_id]->llr_offset[symbol]); + } + } + } + break; + default: + LOG_W(PHY,"rx_dlsch.c : Unknown mod_order!!!!\n"); + return(-1); + break; + } + if (dlsch1_harq) { + switch (get_Qm(dlsch1_harq->mcs)) { + case 2 : + if (rx_type==rx_standard) { + nr_dlsch_qpsk_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + pllr_symbol_cw0, + symbol,first_symbol_flag,nb_rb, + beamforming_mode); + } + break; + case 4: + if (rx_type==rx_standard) { + nr_dlsch_16qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + pdsch_vars[eNB_id]->llr[0], + pdsch_vars[eNB_id]->dl_ch_mag0, + symbol,first_symbol_flag,nb_rb, + pdsch_vars[eNB_id]->llr128, + beamforming_mode); + } + break; + case 6 : + if (rx_type==rx_standard) { + nr_dlsch_64qam_llr(frame_parms, + pdsch_vars[eNB_id]->rxdataF_comp0, + pllr_symbol_cw0, + pdsch_vars[eNB_id]->dl_ch_mag0, + pdsch_vars[eNB_id]->dl_ch_magb0, + symbol,first_symbol_flag,nb_rb, + pdsch_vars[eNB_id]->llr_offset[symbol], + beamforming_mode); + } + break; + default: + LOG_W(PHY,"rx_dlsch.c : Unknown mod_order!!!!\n"); + return(-1); + break; + } + } + + //nr_dlsch_deinterleaving(symbol,bundle_L,(int16_t*)pllr_symbol_cw0,(int16_t*)pllr_symbol_cw0_deint, nb_rb_pdsch); + +#if UE_TIMING_TRACE + stop_meas(&ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot]); +#if DISABLE_LOG_X + printf("[AbsSFN %d.%d] Slot%d Symbol %d: LLR Computation %5.2f \n",frame,nr_tti_rx,slot,symbol,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#else + LOG_I(PHY, "[AbsSFN %d.%d] Slot%d Symbol %d: LLR Computation %5.2f \n",frame,nr_tti_rx,slot,symbol,ue->generic_stat_bis[ue->current_thread_id[nr_tti_rx]][slot].p_time/(cpuf*1000.0)); +#endif +#endif +// Please keep it: useful for debugging +#if 0 + if( (symbol == 13) && (nr_tti_rx==0) && (dlsch0_harq->Qm == 6) /*&& (nb_rb==25)*/) + { + LOG_E(PHY,"Dump Phy Chan Est \n"); + if(1) + { +#if 1 + write_output("rxdataF0.m" , "rxdataF0", &common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF[0][0],14*frame_parms->ofdm_symbol_size,1,1); + //write_output("rxdataF1.m" , "rxdataF1", &common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF[0][0],14*frame_parms->ofdm_symbol_size,1,1); + write_output("dl_ch_estimates00.m", "dl_ch_estimates00", &common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[eNB_id][0][0],14*frame_parms->ofdm_symbol_size,1,1); + //write_output("dl_ch_estimates01.m", "dl_ch_estimates01", &common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[eNB_id][1][0],14*frame_parms->ofdm_symbol_size,1,1); + //write_output("dl_ch_estimates10.m", "dl_ch_estimates10", &common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[eNB_id][2][0],14*frame_parms->ofdm_symbol_size,1,1); + //write_output("dl_ch_estimates11.m", "dl_ch_estimates11", &common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[eNB_id][3][0],14*frame_parms->ofdm_symbol_size,1,1); + + + //write_output("rxdataF_ext00.m" , "rxdataF_ext00", &pdsch_vars[eNB_id]->rxdataF_ext[0][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("rxdataF_ext01.m" , "rxdataF_ext01", &pdsch_vars[eNB_id]->rxdataF_ext[1][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("rxdataF_ext10.m" , "rxdataF_ext10", &pdsch_vars[eNB_id]->rxdataF_ext[2][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("rxdataF_ext11.m" , "rxdataF_ext11", &pdsch_vars[eNB_id]->rxdataF_ext[3][0],14*frame_parms->N_RB_DL*12,1,1); + write_output("dl_ch_estimates_ext00.m", "dl_ch_estimates_ext00", &pdsch_vars[eNB_id]->dl_ch_estimates_ext[0][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("dl_ch_estimates_ext01.m", "dl_ch_estimates_ext01", &pdsch_vars[eNB_id]->dl_ch_estimates_ext[1][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("dl_ch_estimates_ext10.m", "dl_ch_estimates_ext10", &pdsch_vars[eNB_id]->dl_ch_estimates_ext[2][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("dl_ch_estimates_ext11.m", "dl_ch_estimates_ext11", &pdsch_vars[eNB_id]->dl_ch_estimates_ext[3][0],14*frame_parms->N_RB_DL*12,1,1); + write_output("rxdataF_comp00.m","rxdataF_comp00", &pdsch_vars[eNB_id]->rxdataF_comp0[0][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("rxdataF_comp01.m","rxdataF_comp01", &pdsch_vars[eNB_id]->rxdataF_comp0[1][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("rxdataF_comp10.m","rxdataF_comp10", &pdsch_vars[eNB_id]->rxdataF_comp1[harq_pid][round][0][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("rxdataF_comp11.m","rxdataF_comp11", &pdsch_vars[eNB_id]->rxdataF_comp1[harq_pid][round][1][0],14*frame_parms->N_RB_DL*12,1,1); +#endif + write_output("llr0.m","llr0", &pdsch_vars[eNB_id]->llr[0][0],(14*nb_rb*12*dlsch1_harq->Qm) - 4*(nb_rb*4*dlsch1_harq->Qm),1,0); + //write_output("llr1.m","llr1", &pdsch_vars[eNB_id]->llr[1][0],(14*nb_rb*12*dlsch1_harq->Qm) - 4*(nb_rb*4*dlsch1_harq->Qm),1,0); + + + AssertFatal(0," "); + } + + } +#endif + +#if T_TRACER + T(T_UE_PHY_PDSCH_IQ, T_INT(eNB_id), T_INT(ue->Mod_id), T_INT(frame%1024), + T_INT(nr_tti_rx), T_INT(nb_rb), + T_INT(frame_parms->N_RB_UL), T_INT(frame_parms->symbols_per_tti), + T_BUFFER(&pdsch_vars[eNB_id]->rxdataF_comp0[eNB_id][0], + 2 * /* ulsch[UE_id]->harq_processes[harq_pid]->nb_rb */ frame_parms->N_RB_UL *12*frame_parms->symbols_per_tti*2)); +#endif + return(0); + +} + +void nr_dlsch_deinterleaving(uint8_t symbol, + uint16_t L, + uint16_t *llr, + uint16_t *llr_deint, + uint16_t nb_rb_pdsch) +{ + + uint32_t bundle_idx, N_bundle, R, C, r,c; + int32_t m,k; + uint8_t nb_re; + + R=2; + N_bundle = nb_rb_pdsch/L; + C=N_bundle/R; + + uint32_t *bundle_deint = malloc(N_bundle*sizeof(uint32_t)); + + printf("N_bundle %d L %d nb_rb_pdsch %d\n",N_bundle, L,nb_rb_pdsch); + + if (symbol==2) + nb_re = 6; + else + nb_re = 12; + + + AssertFatal(llr!=NULL,"nr_dlsch_deinterleaving: FATAL llr is Null\n"); + + + for (c =0; c< C; c++){ + for (r=0; rnb_antenna_ports_eNB; aatx++) { + if (mod_order == 4) { + QAM_amp128 = _mm_set1_epi16(QAM16_n1); // 2/sqrt(10) + QAM_amp128b = _mm_setzero_si128(); + } else if (mod_order == 6) { + QAM_amp128 = _mm_set1_epi16(QAM64_n1); // + QAM_amp128b = _mm_set1_epi16(QAM64_n2); + } + + // printf("comp: rxdataF_comp %p, symbol %d\n",rxdataF_comp[0],symbol); + + for (aarx=0; aarxnb_antennas_rx; aarx++) { + + dl_ch128 = (__m128i *)&dl_ch_estimates_ext[(aatx<<1)+aarx][symbol*nb_rb*12]; + dl_ch_mag128 = (__m128i *)&dl_ch_mag[(aatx<<1)+aarx][symbol*nb_rb*12]; + dl_ch_mag128b = (__m128i *)&dl_ch_magb[(aatx<<1)+aarx][symbol*nb_rb*12]; + rxdataF128 = (__m128i *)&rxdataF_ext[aarx][symbol*nb_rb*12]; + rxdataF_comp128 = (__m128i *)&rxdataF_comp[(aatx<<1)+aarx][symbol*nb_rb*12]; + + + for (rb=0; rb2) { + // get channel amplitude if not QPSK + + mmtmpD0 = _mm_madd_epi16(dl_ch128[0],dl_ch128[0]); + mmtmpD0 = _mm_srai_epi32(mmtmpD0,(output_shift-2)); + + mmtmpD1 = _mm_madd_epi16(dl_ch128[1],dl_ch128[1]); + mmtmpD1 = _mm_srai_epi32(mmtmpD1,(output_shift-2)); + + mmtmpD0 = _mm_packs_epi32(mmtmpD0,mmtmpD1); + + // store channel magnitude here in a new field of dlsch + + dl_ch_mag128[0] = _mm_unpacklo_epi16(mmtmpD0,mmtmpD0); + dl_ch_mag128b[0] = dl_ch_mag128[0]; + dl_ch_mag128[0] = _mm_mulhi_epi16(dl_ch_mag128[0],QAM_amp128); + dl_ch_mag128[0] = _mm_slli_epi16(dl_ch_mag128[0],1); + //print_ints("Re(ch):",(int16_t*)&mmtmpD0); + //print_shorts("QAM_amp:",(int16_t*)&QAM_amp128); + //print_shorts("mag:",(int16_t*)&dl_ch_mag128[0]); + dl_ch_mag128[1] = _mm_unpackhi_epi16(mmtmpD0,mmtmpD0); + dl_ch_mag128b[1] = dl_ch_mag128[1]; + dl_ch_mag128[1] = _mm_mulhi_epi16(dl_ch_mag128[1],QAM_amp128); + dl_ch_mag128[1] = _mm_slli_epi16(dl_ch_mag128[1],1); + + if (pilots==0) { + mmtmpD0 = _mm_madd_epi16(dl_ch128[2],dl_ch128[2]); + mmtmpD0 = _mm_srai_epi32(mmtmpD0,output_shift); + mmtmpD1 = _mm_packs_epi32(mmtmpD0,mmtmpD0); + + dl_ch_mag128[2] = _mm_unpacklo_epi16(mmtmpD1,mmtmpD1); + dl_ch_mag128b[2] = dl_ch_mag128[2]; + + dl_ch_mag128[2] = _mm_mulhi_epi16(dl_ch_mag128[2],QAM_amp128); + dl_ch_mag128[2] = _mm_slli_epi16(dl_ch_mag128[2],1); + } + + dl_ch_mag128b[0] = _mm_mulhi_epi16(dl_ch_mag128b[0],QAM_amp128b); + dl_ch_mag128b[0] = _mm_slli_epi16(dl_ch_mag128b[0],1); + + + dl_ch_mag128b[1] = _mm_mulhi_epi16(dl_ch_mag128b[1],QAM_amp128b); + dl_ch_mag128b[1] = _mm_slli_epi16(dl_ch_mag128b[1],1); + + if (pilots==0) { + dl_ch_mag128b[2] = _mm_mulhi_epi16(dl_ch_mag128b[2],QAM_amp128b); + dl_ch_mag128b[2] = _mm_slli_epi16(dl_ch_mag128b[2],1); + } + } + + // multiply by conjugated channel + mmtmpD0 = _mm_madd_epi16(dl_ch128[0],rxdataF128[0]); + // print_ints("re",&mmtmpD0); + + // mmtmpD0 contains real part of 4 consecutive outputs (32-bit) + mmtmpD1 = _mm_shufflelo_epi16(dl_ch128[0],_MM_SHUFFLE(2,3,0,1)); + mmtmpD1 = _mm_shufflehi_epi16(mmtmpD1,_MM_SHUFFLE(2,3,0,1)); + mmtmpD1 = _mm_sign_epi16(mmtmpD1,*(__m128i*)&conjugate[0]); + // print_ints("im",&mmtmpD1); + mmtmpD1 = _mm_madd_epi16(mmtmpD1,rxdataF128[0]); + // mmtmpD1 contains imag part of 4 consecutive outputs (32-bit) + mmtmpD0 = _mm_srai_epi32(mmtmpD0,output_shift); + // print_ints("re(shift)",&mmtmpD0); + mmtmpD1 = _mm_srai_epi32(mmtmpD1,output_shift); + // print_ints("im(shift)",&mmtmpD1); + mmtmpD2 = _mm_unpacklo_epi32(mmtmpD0,mmtmpD1); + mmtmpD3 = _mm_unpackhi_epi32(mmtmpD0,mmtmpD1); + // print_ints("c0",&mmtmpD2); + // print_ints("c1",&mmtmpD3); + rxdataF_comp128[0] = _mm_packs_epi32(mmtmpD2,mmtmpD3); + // print_shorts("rx:",rxdataF128); + // print_shorts("ch:",dl_ch128); + // print_shorts("pack:",rxdataF_comp128); + + // multiply by conjugated channel + mmtmpD0 = _mm_madd_epi16(dl_ch128[1],rxdataF128[1]); + // mmtmpD0 contains real part of 4 consecutive outputs (32-bit) + mmtmpD1 = _mm_shufflelo_epi16(dl_ch128[1],_MM_SHUFFLE(2,3,0,1)); + mmtmpD1 = _mm_shufflehi_epi16(mmtmpD1,_MM_SHUFFLE(2,3,0,1)); + mmtmpD1 = _mm_sign_epi16(mmtmpD1,*(__m128i*)conjugate); + mmtmpD1 = _mm_madd_epi16(mmtmpD1,rxdataF128[1]); + // mmtmpD1 contains imag part of 4 consecutive outputs (32-bit) + mmtmpD0 = _mm_srai_epi32(mmtmpD0,output_shift); + mmtmpD1 = _mm_srai_epi32(mmtmpD1,output_shift); + mmtmpD2 = _mm_unpacklo_epi32(mmtmpD0,mmtmpD1); + mmtmpD3 = _mm_unpackhi_epi32(mmtmpD0,mmtmpD1); + + rxdataF_comp128[1] = _mm_packs_epi32(mmtmpD2,mmtmpD3); + // print_shorts("rx:",rxdataF128+1); + // print_shorts("ch:",dl_ch128+1); + // print_shorts("pack:",rxdataF_comp128+1); + + if (pilots==0) { + // multiply by conjugated channel + mmtmpD0 = _mm_madd_epi16(dl_ch128[2],rxdataF128[2]); + // mmtmpD0 contains real part of 4 consecutive outputs (32-bit) + mmtmpD1 = _mm_shufflelo_epi16(dl_ch128[2],_MM_SHUFFLE(2,3,0,1)); + mmtmpD1 = _mm_shufflehi_epi16(mmtmpD1,_MM_SHUFFLE(2,3,0,1)); + mmtmpD1 = _mm_sign_epi16(mmtmpD1,*(__m128i*)conjugate); + mmtmpD1 = _mm_madd_epi16(mmtmpD1,rxdataF128[2]); + // mmtmpD1 contains imag part of 4 consecutive outputs (32-bit) + mmtmpD0 = _mm_srai_epi32(mmtmpD0,output_shift); + mmtmpD1 = _mm_srai_epi32(mmtmpD1,output_shift); + mmtmpD2 = _mm_unpacklo_epi32(mmtmpD0,mmtmpD1); + mmtmpD3 = _mm_unpackhi_epi32(mmtmpD0,mmtmpD1); + + rxdataF_comp128[2] = _mm_packs_epi32(mmtmpD2,mmtmpD3); + // print_shorts("rx:",rxdataF128+2); + // print_shorts("ch:",dl_ch128+2); + // print_shorts("pack:",rxdataF_comp128+2); + + dl_ch128+=3; + dl_ch_mag128+=3; + dl_ch_mag128b+=3; + rxdataF128+=3; + rxdataF_comp128+=3; + } else { // we have a smaller PDSCH in symbols with pilots so skip last group of 4 REs and increment less + dl_ch128+=2; + dl_ch_mag128+=2; + dl_ch_mag128b+=2; + rxdataF128+=2; + rxdataF_comp128+=2; + } + + } + } + } + + if (rho) { + + + for (aarx=0; aarxnb_antennas_rx; aarx++) { + rho128 = (__m128i *)&rho[aarx][symbol*frame_parms->N_RB_DL*12]; + dl_ch128 = (__m128i *)&dl_ch_estimates_ext[aarx][symbol*frame_parms->N_RB_DL*12]; + dl_ch128_2 = (__m128i *)&dl_ch_estimates_ext[2+aarx][symbol*frame_parms->N_RB_DL*12]; + + for (rb=0; rbrx_correlation[0][aarx] = signal_energy(&rho[aarx][symbol*nb_rb*12],rb*12); + } + } + } + + _mm_empty(); + _m_empty(); + +#elif defined(__arm__) + + + unsigned short rb; + unsigned char aatx,aarx,symbol_mod,pilots=0; + + int16x4_t *dl_ch128,*dl_ch128_2,*rxdataF128; + int32x4_t mmtmpD0,mmtmpD1,mmtmpD0b,mmtmpD1b; + int16x8_t *dl_ch_mag128,*dl_ch_mag128b,mmtmpD2,mmtmpD3,mmtmpD4; + int16x8_t QAM_amp128,QAM_amp128b; + int16x4x2_t *rxdataF_comp128,*rho128; + + int16_t conj[4]__attribute__((aligned(16))) = {1,-1,1,-1}; + int32x4_t output_shift128 = vmovq_n_s32(-(int32_t)output_shift); + + symbol_mod = (symbol>=(7-frame_parms->Ncp)) ? symbol-(7-frame_parms->Ncp) : symbol; + + if ((symbol_mod == 0) || (symbol_mod == (4-frame_parms->Ncp))) { + if (frame_parms->nb_antenna_ports_eNB==1) { // 10 out of 12 so don't reduce size + nb_rb=1+(5*nb_rb/6); + } + else { + pilots=1; + } + } + + for (aatx=0; aatxnb_antenna_ports_eNB; aatx++) { + if (mod_order == 4) { + QAM_amp128 = vmovq_n_s16(QAM16_n1); // 2/sqrt(10) + QAM_amp128b = vmovq_n_s16(0); + } else if (mod_order == 6) { + QAM_amp128 = vmovq_n_s16(QAM64_n1); // + QAM_amp128b = vmovq_n_s16(QAM64_n2); + } + // printf("comp: rxdataF_comp %p, symbol %d\n",rxdataF_comp[0],symbol); + + for (aarx=0; aarxnb_antennas_rx; aarx++) { + dl_ch128 = (int16x4_t*)&dl_ch_estimates_ext[(aatx<<1)+aarx][symbol*frame_parms->N_RB_DL*12]; + dl_ch_mag128 = (int16x8_t*)&dl_ch_mag[(aatx<<1)+aarx][symbol*frame_parms->N_RB_DL*12]; + dl_ch_mag128b = (int16x8_t*)&dl_ch_magb[(aatx<<1)+aarx][symbol*frame_parms->N_RB_DL*12]; + rxdataF128 = (int16x4_t*)&rxdataF_ext[aarx][symbol*frame_parms->N_RB_DL*12]; + rxdataF_comp128 = (int16x4x2_t*)&rxdataF_comp[(aatx<<1)+aarx][symbol*frame_parms->N_RB_DL*12]; + + for (rb=0; rb2) { + // get channel amplitude if not QPSK + mmtmpD0 = vmull_s16(dl_ch128[0], dl_ch128[0]); + // mmtmpD0 = [ch0*ch0,ch1*ch1,ch2*ch2,ch3*ch3]; + mmtmpD0 = vqshlq_s32(vqaddq_s32(mmtmpD0,vrev64q_s32(mmtmpD0)),output_shift128); + // mmtmpD0 = [ch0*ch0 + ch1*ch1,ch0*ch0 + ch1*ch1,ch2*ch2 + ch3*ch3,ch2*ch2 + ch3*ch3]>>output_shift128 on 32-bits + mmtmpD1 = vmull_s16(dl_ch128[1], dl_ch128[1]); + mmtmpD1 = vqshlq_s32(vqaddq_s32(mmtmpD1,vrev64q_s32(mmtmpD1)),output_shift128); + mmtmpD2 = vcombine_s16(vmovn_s32(mmtmpD0),vmovn_s32(mmtmpD1)); + // mmtmpD2 = [ch0*ch0 + ch1*ch1,ch0*ch0 + ch1*ch1,ch2*ch2 + ch3*ch3,ch2*ch2 + ch3*ch3,ch4*ch4 + ch5*ch5,ch4*ch4 + ch5*ch5,ch6*ch6 + ch7*ch7,ch6*ch6 + ch7*ch7]>>output_shift128 on 16-bits + mmtmpD0 = vmull_s16(dl_ch128[2], dl_ch128[2]); + mmtmpD0 = vqshlq_s32(vqaddq_s32(mmtmpD0,vrev64q_s32(mmtmpD0)),output_shift128); + mmtmpD1 = vmull_s16(dl_ch128[3], dl_ch128[3]); + mmtmpD1 = vqshlq_s32(vqaddq_s32(mmtmpD1,vrev64q_s32(mmtmpD1)),output_shift128); + mmtmpD3 = vcombine_s16(vmovn_s32(mmtmpD0),vmovn_s32(mmtmpD1)); + if (pilots==0) { + mmtmpD0 = vmull_s16(dl_ch128[4], dl_ch128[4]); + mmtmpD0 = vqshlq_s32(vqaddq_s32(mmtmpD0,vrev64q_s32(mmtmpD0)),output_shift128); + mmtmpD1 = vmull_s16(dl_ch128[5], dl_ch128[5]); + mmtmpD1 = vqshlq_s32(vqaddq_s32(mmtmpD1,vrev64q_s32(mmtmpD1)),output_shift128); + mmtmpD4 = vcombine_s16(vmovn_s32(mmtmpD0),vmovn_s32(mmtmpD1)); + } + + dl_ch_mag128b[0] = vqdmulhq_s16(mmtmpD2,QAM_amp128b); + dl_ch_mag128b[1] = vqdmulhq_s16(mmtmpD3,QAM_amp128b); + dl_ch_mag128[0] = vqdmulhq_s16(mmtmpD2,QAM_amp128); + dl_ch_mag128[1] = vqdmulhq_s16(mmtmpD3,QAM_amp128); + + if (pilots==0) { + dl_ch_mag128b[2] = vqdmulhq_s16(mmtmpD4,QAM_amp128b); + dl_ch_mag128[2] = vqdmulhq_s16(mmtmpD4,QAM_amp128); + } + } + + mmtmpD0 = vmull_s16(dl_ch128[0], rxdataF128[0]); + //mmtmpD0 = [Re(ch[0])Re(rx[0]) Im(ch[0])Im(ch[0]) Re(ch[1])Re(rx[1]) Im(ch[1])Im(ch[1])] + mmtmpD1 = vmull_s16(dl_ch128[1], rxdataF128[1]); + //mmtmpD1 = [Re(ch[2])Re(rx[2]) Im(ch[2])Im(ch[2]) Re(ch[3])Re(rx[3]) Im(ch[3])Im(ch[3])] + mmtmpD0 = vcombine_s32(vpadd_s32(vget_low_s32(mmtmpD0),vget_high_s32(mmtmpD0)), + vpadd_s32(vget_low_s32(mmtmpD1),vget_high_s32(mmtmpD1))); + //mmtmpD0 = [Re(ch[0])Re(rx[0])+Im(ch[0])Im(ch[0]) Re(ch[1])Re(rx[1])+Im(ch[1])Im(ch[1]) Re(ch[2])Re(rx[2])+Im(ch[2])Im(ch[2]) Re(ch[3])Re(rx[3])+Im(ch[3])Im(ch[3])] + + mmtmpD0b = vmull_s16(vrev32_s16(vmul_s16(dl_ch128[0],*(int16x4_t*)conj)), rxdataF128[0]); + //mmtmpD0 = [-Im(ch[0])Re(rx[0]) Re(ch[0])Im(rx[0]) -Im(ch[1])Re(rx[1]) Re(ch[1])Im(rx[1])] + mmtmpD1b = vmull_s16(vrev32_s16(vmul_s16(dl_ch128[1],*(int16x4_t*)conj)), rxdataF128[1]); + //mmtmpD0 = [-Im(ch[2])Re(rx[2]) Re(ch[2])Im(rx[2]) -Im(ch[3])Re(rx[3]) Re(ch[3])Im(rx[3])] + mmtmpD1 = vcombine_s32(vpadd_s32(vget_low_s32(mmtmpD0b),vget_high_s32(mmtmpD0b)), + vpadd_s32(vget_low_s32(mmtmpD1b),vget_high_s32(mmtmpD1b))); + //mmtmpD1 = [-Im(ch[0])Re(rx[0])+Re(ch[0])Im(rx[0]) -Im(ch[1])Re(rx[1])+Re(ch[1])Im(rx[1]) -Im(ch[2])Re(rx[2])+Re(ch[2])Im(rx[2]) -Im(ch[3])Re(rx[3])+Re(ch[3])Im(rx[3])] + + mmtmpD0 = vqshlq_s32(mmtmpD0,output_shift128); + mmtmpD1 = vqshlq_s32(mmtmpD1,output_shift128); + rxdataF_comp128[0] = vzip_s16(vmovn_s32(mmtmpD0),vmovn_s32(mmtmpD1)); + mmtmpD0 = vmull_s16(dl_ch128[2], rxdataF128[2]); + mmtmpD1 = vmull_s16(dl_ch128[3], rxdataF128[3]); + mmtmpD0 = vcombine_s32(vpadd_s32(vget_low_s32(mmtmpD0),vget_high_s32(mmtmpD0)), + vpadd_s32(vget_low_s32(mmtmpD1),vget_high_s32(mmtmpD1))); + mmtmpD0b = vmull_s16(vrev32_s16(vmul_s16(dl_ch128[2],*(int16x4_t*)conj)), rxdataF128[2]); + mmtmpD1b = vmull_s16(vrev32_s16(vmul_s16(dl_ch128[3],*(int16x4_t*)conj)), rxdataF128[3]); + mmtmpD1 = vcombine_s32(vpadd_s32(vget_low_s32(mmtmpD0b),vget_high_s32(mmtmpD0b)), + vpadd_s32(vget_low_s32(mmtmpD1b),vget_high_s32(mmtmpD1b))); + mmtmpD0 = vqshlq_s32(mmtmpD0,output_shift128); + mmtmpD1 = vqshlq_s32(mmtmpD1,output_shift128); + rxdataF_comp128[1] = vzip_s16(vmovn_s32(mmtmpD0),vmovn_s32(mmtmpD1)); + + if (pilots==0) { + mmtmpD0 = vmull_s16(dl_ch128[4], rxdataF128[4]); + mmtmpD1 = vmull_s16(dl_ch128[5], rxdataF128[5]); + mmtmpD0 = vcombine_s32(vpadd_s32(vget_low_s32(mmtmpD0),vget_high_s32(mmtmpD0)), + vpadd_s32(vget_low_s32(mmtmpD1),vget_high_s32(mmtmpD1))); + + mmtmpD0b = vmull_s16(vrev32_s16(vmul_s16(dl_ch128[4],*(int16x4_t*)conj)), rxdataF128[4]); + mmtmpD1b = vmull_s16(vrev32_s16(vmul_s16(dl_ch128[5],*(int16x4_t*)conj)), rxdataF128[5]); + mmtmpD1 = vcombine_s32(vpadd_s32(vget_low_s32(mmtmpD0b),vget_high_s32(mmtmpD0b)), + vpadd_s32(vget_low_s32(mmtmpD1b),vget_high_s32(mmtmpD1b))); + + + mmtmpD0 = vqshlq_s32(mmtmpD0,output_shift128); + mmtmpD1 = vqshlq_s32(mmtmpD1,output_shift128); + rxdataF_comp128[2] = vzip_s16(vmovn_s32(mmtmpD0),vmovn_s32(mmtmpD1)); + + + dl_ch128+=6; + dl_ch_mag128+=3; + dl_ch_mag128b+=3; + rxdataF128+=6; + rxdataF_comp128+=3; + + } else { // we have a smaller PDSCH in symbols with pilots so skip last group of 4 REs and increment less + dl_ch128+=4; + dl_ch_mag128+=2; + dl_ch_mag128b+=2; + rxdataF128+=4; + rxdataF_comp128+=2; + } + } + } + } + + if (rho) { + for (aarx=0; aarxnb_antennas_rx; aarx++) { + rho128 = (int16x4x2_t*)&rho[aarx][symbol*frame_parms->N_RB_DL*12]; + dl_ch128 = (int16x4_t*)&dl_ch_estimates_ext[aarx][symbol*frame_parms->N_RB_DL*12]; + dl_ch128_2 = (int16x4_t*)&dl_ch_estimates_ext[2+aarx][symbol*frame_parms->N_RB_DL*12]; + for (rb=0; rbrx_correlation[0][aarx] = signal_energy(&rho[aarx][symbol*frame_parms->N_RB_DL*12],rb*12); + } + } + } +#endif +} + + +void nr_dlsch_scale_channel(int **dl_ch_estimates_ext, + NR_DL_FRAME_PARMS *frame_parms, + NR_UE_DLSCH_t **dlsch_ue, + uint8_t symbol, + unsigned short nb_rb) +{ + +#if defined(__x86_64__)||defined(__i386__) + + short rb, ch_amp; + unsigned char aatx,aarx,pilots=0,symbol_mod; + __m128i *dl_ch128, ch_amp128; + + + if (symbol==2){ + nb_rb = nb_rb>>1; + pilots=1; + } + + // Determine scaling amplitude based the symbol + +ch_amp = 1024*8; //((pilots) ? (dlsch_ue[0]->sqrt_rho_b) : (dlsch_ue[0]->sqrt_rho_a)); + + LOG_D(PHY,"Scaling PDSCH Chest in OFDM symbol %d by %d, pilots %d nb_rb %d NCP %d symbol %d\n",symbol_mod,ch_amp,pilots,nb_rb,frame_parms->Ncp,symbol); + // printf("Scaling PDSCH Chest in OFDM symbol %d by %d\n",symbol_mod,ch_amp); + + ch_amp128 = _mm_set1_epi16(ch_amp); // Q3.13 + + for (aatx=0; aatxnb_antenna_ports_eNB; aatx++) { + for (aarx=0; aarxnb_antennas_rx; aarx++) { + + dl_ch128=(__m128i *)&dl_ch_estimates_ext[(aatx<<1)+aarx][symbol*nb_rb*12]; + + for (rb=0;rb>x; + //printf("nb_rb*nre = %d = %d * 2^(%d)\n",nb_rb*nre,y,x); + + for (aatx=0; aatxnb_antenna_ports_eNB; aatx++) + for (aarx=0; aarxnb_antennas_rx; aarx++) { + //clear average level + avg128D = _mm_setzero_si128(); + // 5 is always a symbol with no pilots for both normal and extended prefix + + dl_ch128=(__m128i *)&dl_ch_estimates_ext[(aatx<<1)+aarx][symbol*nb_rb*12]; + + for (rb=0;rbNcp-1)))&&(frame_parms->nb_antenna_ports_eNB!=1)) { + dl_ch128+=2; + } + else {*/ + avg128D = _mm_add_epi32(avg128D,_mm_srai_epi16(_mm_madd_epi16(dl_ch128[2],dl_ch128[2]),x)); + //avg128D = _mm_add_epi32(avg128D,_mm_madd_epi16(dl_ch128[2],_mm_srai_epi16(_mm_mulhi_epi16(dl_ch128[2], coeff128),15))); + dl_ch128+=3; + // } + /* + if (rb==0) { + print_shorts("dl_ch128",&dl_ch128[0]); + print_shorts("dl_ch128",&dl_ch128[1]); + print_shorts("dl_ch128",&dl_ch128[2]); + } + */ + } + + avg[(aatx<<1)+aarx] =(((int32_t*)&avg128D)[0] + + ((int32_t*)&avg128D)[1] + + ((int32_t*)&avg128D)[2] + + ((int32_t*)&avg128D)[3])/y; + // printf("Channel level : %d\n",avg[(aatx<<1)+aarx]); + } + + _mm_empty(); + _m_empty(); + +#elif defined(__arm__) + + short rb; + unsigned char aatx,aarx,nre=12,symbol_mod; + int32x4_t avg128D; + int16x4_t *dl_ch128; + + symbol_mod = (symbol>=(7-frame_parms->Ncp)) ? symbol-(7-frame_parms->Ncp) : symbol; + + for (aatx=0; aatxnb_antenna_ports_eNB; aatx++) + for (aarx=0; aarxnb_antennas_rx; aarx++) { + //clear average level + avg128D = vdupq_n_s32(0); + // 5 is always a symbol with no pilots for both normal and extended prefix + + dl_ch128=(int16x4_t *)&dl_ch_estimates_ext[(aatx<<1)+aarx][symbol*frame_parms->N_RB_DL*12]; + + for (rb=0; rbNcp-1)))&&(frame_parms->nb_antenna_ports_eNB!=1)) { + dl_ch128+=4; + } else { + avg128D = vqaddq_s32(avg128D,vmull_s16(dl_ch128[4],dl_ch128[4])); + avg128D = vqaddq_s32(avg128D,vmull_s16(dl_ch128[5],dl_ch128[5])); + dl_ch128+=6; + } + + /* + if (rb==0) { + print_shorts("dl_ch128",&dl_ch128[0]); + print_shorts("dl_ch128",&dl_ch128[1]); + print_shorts("dl_ch128",&dl_ch128[2]); + } + */ + } + + if (symbol==2) //assume start symbol 2 + nre=6; + else + nre=12; + + avg[(aatx<<1)+aarx] = (((int32_t*)&avg128D)[0] + + ((int32_t*)&avg128D)[1] + + ((int32_t*)&avg128D)[2] + + ((int32_t*)&avg128D)[3])/(nb_rb*nre); + + // printf("Channel level : %d\n",avg[(aatx<<1)+aarx]); + } + + +#endif +} + +//============================================================================================== +// Extraction functions +//============================================================================================== + +unsigned short nr_dlsch_extract_rbs_single(int **rxdataF, + int **dl_ch_estimates, + int **rxdataF_ext, + int **dl_ch_estimates_ext, + unsigned short pmi, + unsigned char *pmi_ext, + unsigned int *rb_alloc, + unsigned char symbol, + unsigned short start_rb, + unsigned short nb_rb_pdsch, + unsigned char nr_tti_rx, + uint32_t high_speed_flag, + NR_DL_FRAME_PARMS *frame_parms) { + + + + unsigned short k,rb; + unsigned char i,aarx,l; //,nsymb,sss_symb,pss_symb=0; + int *dl_ch0,*dl_ch0_ext,*rxF,*rxF_ext; + + + + unsigned char pilots=0,j=0; + + //symbol_mod = (symbol>=(7-frame_parms->Ncp)) ? symbol-(7-frame_parms->Ncp) : symbol; + pilots = (symbol==2) ? 1 : 0; //to updated from config + l=symbol; + //nsymb = (frame_parms->Ncp==NORMAL) ? 14:12; + k = frame_parms->first_carrier_offset + 516; //0 + + /* if (frame_parms->frame_type == TDD) { // TDD + sss_symb = nsymb-1; + pss_symb = 2; + } else { + sss_symb = (nsymb>>1)-2; + pss_symb = (nsymb>>1)-1; + }*/ + + + for (aarx=0; aarxnb_antennas_rx; aarx++) { + + if (high_speed_flag == 1) + dl_ch0 = &dl_ch_estimates[aarx][(2*(frame_parms->ofdm_symbol_size))]; + else + dl_ch0 = &dl_ch_estimates[aarx][0]; + + dl_ch0_ext = &dl_ch_estimates_ext[aarx][symbol*(nb_rb_pdsch*12)]; + + rxF_ext = &rxdataF_ext[aarx][symbol*(nb_rb_pdsch*12)]; + rxF = &rxdataF[aarx][(k+(symbol*(frame_parms->ofdm_symbol_size)))]; + + //printf("extract single addr %p pilots %d symbol %d, l %d RB_DL %d nushift %d k %d\n", rxF, pilots, symbol, l, frame_parms->N_RB_DL,frame_parms->nushift,k); + + //if ((frame_parms->N_RB_DL&1) == 0){ // even number of RBs + + for (rb=start_rb;rb=frame_parms->ofdm_symbol_size) { + rxF = &rxdataF[aarx][(symbol*(frame_parms->ofdm_symbol_size))]; + k=k-(frame_parms->ofdm_symbol_size); + } + + //*pmi_ext = (pmi>>((rb>>2)<<1))&3; + //memcpy(dl_ch0_ext,dl_ch0,12*sizeof(int)); + + /* + printf("rb %d\n",rb); + for (i=0;i<12;i++) + printf("(%d %d)",((short *)dl_ch0)[i<<1],((short*)dl_ch0)[1+(i<<1)]); + printf("\n"); + */ + if (pilots==0) { + for (i=0; i<12; i++) { + rxF_ext[i]=rxF[i]; + dl_ch0_ext[i]=dl_ch0[i]; + /*printf("rb %ld %p: (%d,%d)\n",(rxF+i-&rxdataF[aarx][( (symbol*(frame_parms->ofdm_symbol_size)))]),rxF, + ((short*)&rxF[i])[0],((short*)&rxF[i])[1]); + printf("rb %ld %p: dl_ch (%d,%d)\n",(rxF+i-&rxdataF[aarx][( (symbol*(frame_parms->ofdm_symbol_size)))]),rxF, + ((short*)&dl_ch0[i])[0],((short*)&dl_ch0[i])[1]); + printf("re count %d\n",rb*12+i);*/ + } + + dl_ch0_ext+=12; + rxF_ext+=12; + } else { + j=0; + + for (i=0; i<12; i++) { + if ((i&1)!=frame_parms->nushift){ + rxF_ext[j]=rxF[i]; + //printf("extract rb %d, re %d => rxF (%d,%d) rxF ext (%d,%d) addr %p\n",rb,i,*(short *)&rxF[i],*(1+(short*)&rxF[i]),*(short *)&rxF_ext[j],*(1+(short*)&rxF_ext[j]),rxF_ext); + dl_ch0_ext[j]=dl_ch0[i]; + //printf("rb %ld %p: dl_ch (%d,%d)\n",(rxF+i-&rxdataF[aarx][( (symbol*(frame_parms->ofdm_symbol_size)))]),rxF, + //((short*)&dl_ch0[i])[0],((short*)&dl_ch0[i])[1]); + j++; + } + } + + dl_ch0_ext+=6; + rxF_ext+=6; + } + + dl_ch0+=12; + rxF+=12; + k+=12; + } + } + + + return(nb_rb_pdsch/frame_parms->nb_antennas_rx); +} + +//============================================================================================== + +#ifdef USER_MODE + + +void dump_dlsch2(PHY_VARS_UE *ue,uint8_t eNB_id,uint8_t nr_tti_rx,unsigned int *coded_bits_per_codeword,int round, unsigned char harq_pid) +{ + unsigned int nsymb = (ue->frame_parms.Ncp == 0) ? 14 : 12; + char fname[32],vname[32]; + int N_RB_DL=ue->frame_parms.N_RB_DL; + + sprintf(fname,"dlsch%d_rxF_r%d_ext0.m",eNB_id,round); + sprintf(vname,"dl%d_rxF_r%d_ext0",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->rxdataF_ext[0],12*N_RB_DL*nsymb,1,1); + + if (ue->frame_parms.nb_antennas_rx >1) { + sprintf(fname,"dlsch%d_rxF_r%d_ext1.m",eNB_id,round); + sprintf(vname,"dl%d_rxF_r%d_ext1",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->rxdataF_ext[1],12*N_RB_DL*nsymb,1,1); + } + + sprintf(fname,"dlsch%d_ch_r%d_ext00.m",eNB_id,round); + sprintf(vname,"dl%d_ch_r%d_ext00",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dl_ch_estimates_ext[0],12*N_RB_DL*nsymb,1,1); + + if (ue->transmission_mode[eNB_id]==7){ + sprintf(fname,"dlsch%d_bf_ch_r%d.m",eNB_id,round); + sprintf(vname,"dl%d_bf_ch_r%d",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dl_bf_ch_estimates[0],512*nsymb,1,1); + //write_output(fname,vname,phy_vars_ue->lte_ue_pdsch_vars[eNB_id]->dl_bf_ch_estimates[0],512,1,1); + + sprintf(fname,"dlsch%d_bf_ch_r%d_ext00.m",eNB_id,round); + sprintf(vname,"dl%d_bf_ch_r%d_ext00",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dl_bf_ch_estimates_ext[0],12*N_RB_DL*nsymb,1,1); + } + + if (ue->frame_parms.nb_antennas_rx == 2) { + sprintf(fname,"dlsch%d_ch_r%d_ext01.m",eNB_id,round); + sprintf(vname,"dl%d_ch_r%d_ext01",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dl_ch_estimates_ext[1],12*N_RB_DL*nsymb,1,1); + } + + if (ue->frame_parms.nb_antenna_ports_eNB == 2) { + sprintf(fname,"dlsch%d_ch_r%d_ext10.m",eNB_id,round); + sprintf(vname,"dl%d_ch_r%d_ext10",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dl_ch_estimates_ext[2],12*N_RB_DL*nsymb,1,1); + + if (ue->frame_parms.nb_antennas_rx == 2) { + sprintf(fname,"dlsch%d_ch_r%d_ext11.m",eNB_id,round); + sprintf(vname,"dl%d_ch_r%d_ext11",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dl_ch_estimates_ext[3],12*N_RB_DL*nsymb,1,1); + } + } + + sprintf(fname,"dlsch%d_rxF_r%d_uespec0.m",eNB_id,round); + sprintf(vname,"dl%d_rxF_r%d_uespec0",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->rxdataF_uespec_pilots[0],12*N_RB_DL,1,1); + + /* + write_output("dlsch%d_ch_ext01.m","dl01_ch0_ext",pdsch_vars[eNB_id]->dl_ch_estimates_ext[1],12*N_RB_DL*nsymb,1,1); + write_output("dlsch%d_ch_ext10.m","dl10_ch0_ext",pdsch_vars[eNB_id]->dl_ch_estimates_ext[2],12*N_RB_DL*nsymb,1,1); + write_output("dlsch%d_ch_ext11.m","dl11_ch0_ext",pdsch_vars[eNB_id]->dl_ch_estimates_ext[3],12*N_RB_DL*nsymb,1,1); + */ + sprintf(fname,"dlsch%d_r%d_rho.m",eNB_id,round); + sprintf(vname,"dl_rho_r%d_%d",eNB_id,round); + + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dl_ch_rho_ext[harq_pid][round][0],12*N_RB_DL*nsymb,1,1); + + sprintf(fname,"dlsch%d_r%d_rho2.m",eNB_id,round); + sprintf(vname,"dl_rho2_r%d_%d",eNB_id,round); + + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dl_ch_rho2_ext[0],12*N_RB_DL*nsymb,1,1); + + sprintf(fname,"dlsch%d_rxF_r%d_comp0.m",eNB_id,round); + sprintf(vname,"dl%d_rxF_r%d_comp0",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->rxdataF_comp0[0],12*N_RB_DL*nsymb,1,1); + if (ue->frame_parms.nb_antenna_ports_eNB == 2) { + sprintf(fname,"dlsch%d_rxF_r%d_comp1.m",eNB_id,round); + sprintf(vname,"dl%d_rxF_r%d_comp1",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->rxdataF_comp1[harq_pid][round][0],12*N_RB_DL*nsymb,1,1); + } + + sprintf(fname,"dlsch%d_rxF_r%d_llr.m",eNB_id,round); + sprintf(vname,"dl%d_r%d_llr",eNB_id,round); + write_output(fname,vname, ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->llr[0],coded_bits_per_codeword[0],1,0); + sprintf(fname,"dlsch%d_r%d_mag1.m",eNB_id,round); + sprintf(vname,"dl%d_r%d_mag1",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dl_ch_mag0[0],12*N_RB_DL*nsymb,1,1); + sprintf(fname,"dlsch%d_r%d_mag2.m",eNB_id,round); + sprintf(vname,"dl%d_r%d_mag2",eNB_id,round); + write_output(fname,vname,ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dl_ch_magb0[0],12*N_RB_DL*nsymb,1,1); + + // printf("log2_maxh = %d\n",ue->pdsch_vars[eNB_id]->log2_maxh); +} +#endif + +#ifdef DEBUG_DLSCH_DEMOD +/* +void print_bytes(char *s,__m128i *x) +{ + + char *tempb = (char *)x; + + printf("%s : %d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d\n",s, + tempb[0],tempb[1],tempb[2],tempb[3],tempb[4],tempb[5],tempb[6],tempb[7], + tempb[8],tempb[9],tempb[10],tempb[11],tempb[12],tempb[13],tempb[14],tempb[15] + ); + +} + +void print_shorts(char *s,__m128i *x) +{ + + short *tempb = (short *)x; + printf("%s : %d,%d,%d,%d,%d,%d,%d,%d\n",s, + tempb[0],tempb[1],tempb[2],tempb[3],tempb[4],tempb[5],tempb[6],tempb[7]); + +} + +void print_shorts2(char *s,__m64 *x) +{ + + short *tempb = (short *)x; + printf("%s : %d,%d,%d,%d\n",s, + tempb[0],tempb[1],tempb[2],tempb[3]); + +} + +void print_ints(char *s,__m128i *x) +{ + + int *tempb = (int *)x; + printf("%s : %d,%d,%d,%d\n",s, + tempb[0],tempb[1],tempb[2],tempb[3]); + +}*/ +#endif diff --git a/openair1/PHY/NR_UE_TRANSPORT/nr_dlsch_llr_computation.c b/openair1/PHY/NR_UE_TRANSPORT/nr_dlsch_llr_computation.c new file mode 100644 index 0000000000..9527bf10e9 --- /dev/null +++ b/openair1/PHY/NR_UE_TRANSPORT/nr_dlsch_llr_computation.c @@ -0,0 +1,8641 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.1 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +/*! \file PHY/NR_UE_TRANSPORT/nr_dlsch_llr_computation.c + * \brief Top-level routines for LLR computation of the PDSCH physical channel + * \author H. WANG + * \date 2018 + * \version 0.1 + * \company Eurecom + * \email: + * \note + * \warning + */ + +#include "PHY/defs_nr_UE.h" +#include "PHY/phy_extern_nr_ue.h" +#include "nr_transport_proto_ue.h" +#include "PHY/TOOLS/tools_defs.h" +#include "PHY/sse_intrin.h" + +//#define DEBUG_LLR_SIC + + +int16_t zeros[8] __attribute__ ((aligned(16))) = {0,0,0,0,0,0,0,0}; +int16_t ones[8] __attribute__ ((aligned(16))) = {0xffff,0xffff,0xffff,0xffff,0xffff,0xffff,0xffff,0xffff}; +#if defined(__x86_64__) || defined(__i386__) +__m128i rho_rpi __attribute__ ((aligned(16))); +__m128i rho_rmi __attribute__ ((aligned(16))); +__m128i rho_rpi_1_1 __attribute__ ((aligned(16))); +__m128i rho_rpi_1_3 __attribute__ ((aligned(16))); +__m128i rho_rpi_1_5 __attribute__ ((aligned(16))); +__m128i rho_rpi_1_7 __attribute__ ((aligned(16))); +__m128i rho_rpi_3_1 __attribute__ ((aligned(16))); +__m128i rho_rpi_3_3 __attribute__ ((aligned(16))); +__m128i rho_rpi_3_5 __attribute__ ((aligned(16))); +__m128i rho_rpi_3_7 __attribute__ ((aligned(16))); +__m128i rho_rpi_5_1 __attribute__ ((aligned(16))); +__m128i rho_rpi_5_3 __attribute__ ((aligned(16))); +__m128i rho_rpi_5_5 __attribute__ ((aligned(16))); +__m128i rho_rpi_5_7 __attribute__ ((aligned(16))); +__m128i rho_rpi_7_1 __attribute__ ((aligned(16))); +__m128i rho_rpi_7_3 __attribute__ ((aligned(16))); +__m128i rho_rpi_7_5 __attribute__ ((aligned(16))); +__m128i rho_rpi_7_7 __attribute__ ((aligned(16))); +__m128i rho_rmi_1_1 __attribute__ ((aligned(16))); +__m128i rho_rmi_1_3 __attribute__ ((aligned(16))); +__m128i rho_rmi_1_5 __attribute__ ((aligned(16))); +__m128i rho_rmi_1_7 __attribute__ ((aligned(16))); +__m128i rho_rmi_3_1 __attribute__ ((aligned(16))); +__m128i rho_rmi_3_3 __attribute__ ((aligned(16))); +__m128i rho_rmi_3_5 __attribute__ ((aligned(16))); +__m128i rho_rmi_3_7 __attribute__ ((aligned(16))); +__m128i rho_rmi_5_1 __attribute__ ((aligned(16))); +__m128i rho_rmi_5_3 __attribute__ ((aligned(16))); +__m128i rho_rmi_5_5 __attribute__ ((aligned(16))); +__m128i rho_rmi_5_7 __attribute__ ((aligned(16))); +__m128i rho_rmi_7_1 __attribute__ ((aligned(16))); +__m128i rho_rmi_7_3 __attribute__ ((aligned(16))); +__m128i rho_rmi_7_5 __attribute__ ((aligned(16))); +__m128i rho_rmi_7_7 __attribute__ ((aligned(16))); + +__m128i psi_r_m7_m7 __attribute__ ((aligned(16))); +__m128i psi_r_m7_m5 __attribute__ ((aligned(16))); +__m128i psi_r_m7_m3 __attribute__ ((aligned(16))); +__m128i psi_r_m7_m1 __attribute__ ((aligned(16))); +__m128i psi_r_m7_p1 __attribute__ ((aligned(16))); +__m128i psi_r_m7_p3 __attribute__ ((aligned(16))); +__m128i psi_r_m7_p5 __attribute__ ((aligned(16))); +__m128i psi_r_m7_p7 __attribute__ ((aligned(16))); +__m128i psi_r_m5_m7 __attribute__ ((aligned(16))); +__m128i psi_r_m5_m5 __attribute__ ((aligned(16))); +__m128i psi_r_m5_m3 __attribute__ ((aligned(16))); +__m128i psi_r_m5_m1 __attribute__ ((aligned(16))); +__m128i psi_r_m5_p1 __attribute__ ((aligned(16))); +__m128i psi_r_m5_p3 __attribute__ ((aligned(16))); +__m128i psi_r_m5_p5 __attribute__ ((aligned(16))); +__m128i psi_r_m5_p7 __attribute__ ((aligned(16))); +__m128i psi_r_m3_m7 __attribute__ ((aligned(16))); +__m128i psi_r_m3_m5 __attribute__ ((aligned(16))); +__m128i psi_r_m3_m3 __attribute__ ((aligned(16))); +__m128i psi_r_m3_m1 __attribute__ ((aligned(16))); +__m128i psi_r_m3_p1 __attribute__ ((aligned(16))); +__m128i psi_r_m3_p3 __attribute__ ((aligned(16))); +__m128i psi_r_m3_p5 __attribute__ ((aligned(16))); +__m128i psi_r_m3_p7 __attribute__ ((aligned(16))); +__m128i psi_r_m1_m7 __attribute__ ((aligned(16))); +__m128i psi_r_m1_m5 __attribute__ ((aligned(16))); +__m128i psi_r_m1_m3 __attribute__ ((aligned(16))); +__m128i psi_r_m1_m1 __attribute__ ((aligned(16))); +__m128i psi_r_m1_p1 __attribute__ ((aligned(16))); +__m128i psi_r_m1_p3 __attribute__ ((aligned(16))); +__m128i psi_r_m1_p5 __attribute__ ((aligned(16))); +__m128i psi_r_m1_p7 __attribute__ ((aligned(16))); +__m128i psi_r_p1_m7 __attribute__ ((aligned(16))); +__m128i psi_r_p1_m5 __attribute__ ((aligned(16))); +__m128i psi_r_p1_m3 __attribute__ ((aligned(16))); +__m128i psi_r_p1_m1 __attribute__ ((aligned(16))); +__m128i psi_r_p1_p1 __attribute__ ((aligned(16))); +__m128i psi_r_p1_p3 __attribute__ ((aligned(16))); +__m128i psi_r_p1_p5 __attribute__ ((aligned(16))); +__m128i psi_r_p1_p7 __attribute__ ((aligned(16))); +__m128i psi_r_p3_m7 __attribute__ ((aligned(16))); +__m128i psi_r_p3_m5 __attribute__ ((aligned(16))); +__m128i psi_r_p3_m3 __attribute__ ((aligned(16))); +__m128i psi_r_p3_m1 __attribute__ ((aligned(16))); +__m128i psi_r_p3_p1 __attribute__ ((aligned(16))); +__m128i psi_r_p3_p3 __attribute__ ((aligned(16))); +__m128i psi_r_p3_p5 __attribute__ ((aligned(16))); +__m128i psi_r_p3_p7 __attribute__ ((aligned(16))); +__m128i psi_r_p5_m7 __attribute__ ((aligned(16))); +__m128i psi_r_p5_m5 __attribute__ ((aligned(16))); +__m128i psi_r_p5_m3 __attribute__ ((aligned(16))); +__m128i psi_r_p5_m1 __attribute__ ((aligned(16))); +__m128i psi_r_p5_p1 __attribute__ ((aligned(16))); +__m128i psi_r_p5_p3 __attribute__ ((aligned(16))); +__m128i psi_r_p5_p5 __attribute__ ((aligned(16))); +__m128i psi_r_p5_p7 __attribute__ ((aligned(16))); +__m128i psi_r_p7_m7 __attribute__ ((aligned(16))); +__m128i psi_r_p7_m5 __attribute__ ((aligned(16))); +__m128i psi_r_p7_m3 __attribute__ ((aligned(16))); +__m128i psi_r_p7_m1 __attribute__ ((aligned(16))); +__m128i psi_r_p7_p1 __attribute__ ((aligned(16))); +__m128i psi_r_p7_p3 __attribute__ ((aligned(16))); +__m128i psi_r_p7_p5 __attribute__ ((aligned(16))); +__m128i psi_r_p7_p7 __attribute__ ((aligned(16))); + +__m128i psi_i_m7_m7 __attribute__ ((aligned(16))); +__m128i psi_i_m7_m5 __attribute__ ((aligned(16))); +__m128i psi_i_m7_m3 __attribute__ ((aligned(16))); +__m128i psi_i_m7_m1 __attribute__ ((aligned(16))); +__m128i psi_i_m7_p1 __attribute__ ((aligned(16))); +__m128i psi_i_m7_p3 __attribute__ ((aligned(16))); +__m128i psi_i_m7_p5 __attribute__ ((aligned(16))); +__m128i psi_i_m7_p7 __attribute__ ((aligned(16))); +__m128i psi_i_m5_m7 __attribute__ ((aligned(16))); +__m128i psi_i_m5_m5 __attribute__ ((aligned(16))); +__m128i psi_i_m5_m3 __attribute__ ((aligned(16))); +__m128i psi_i_m5_m1 __attribute__ ((aligned(16))); +__m128i psi_i_m5_p1 __attribute__ ((aligned(16))); +__m128i psi_i_m5_p3 __attribute__ ((aligned(16))); +__m128i psi_i_m5_p5 __attribute__ ((aligned(16))); +__m128i psi_i_m5_p7 __attribute__ ((aligned(16))); +__m128i psi_i_m3_m7 __attribute__ ((aligned(16))); +__m128i psi_i_m3_m5 __attribute__ ((aligned(16))); +__m128i psi_i_m3_m3 __attribute__ ((aligned(16))); +__m128i psi_i_m3_m1 __attribute__ ((aligned(16))); +__m128i psi_i_m3_p1 __attribute__ ((aligned(16))); +__m128i psi_i_m3_p3 __attribute__ ((aligned(16))); +__m128i psi_i_m3_p5 __attribute__ ((aligned(16))); +__m128i psi_i_m3_p7 __attribute__ ((aligned(16))); +__m128i psi_i_m1_m7 __attribute__ ((aligned(16))); +__m128i psi_i_m1_m5 __attribute__ ((aligned(16))); +__m128i psi_i_m1_m3 __attribute__ ((aligned(16))); +__m128i psi_i_m1_m1 __attribute__ ((aligned(16))); +__m128i psi_i_m1_p1 __attribute__ ((aligned(16))); +__m128i psi_i_m1_p3 __attribute__ ((aligned(16))); +__m128i psi_i_m1_p5 __attribute__ ((aligned(16))); +__m128i psi_i_m1_p7 __attribute__ ((aligned(16))); +__m128i psi_i_p1_m7 __attribute__ ((aligned(16))); +__m128i psi_i_p1_m5 __attribute__ ((aligned(16))); +__m128i psi_i_p1_m3 __attribute__ ((aligned(16))); +__m128i psi_i_p1_m1 __attribute__ ((aligned(16))); +__m128i psi_i_p1_p1 __attribute__ ((aligned(16))); +__m128i psi_i_p1_p3 __attribute__ ((aligned(16))); +__m128i psi_i_p1_p5 __attribute__ ((aligned(16))); +__m128i psi_i_p1_p7 __attribute__ ((aligned(16))); +__m128i psi_i_p3_m7 __attribute__ ((aligned(16))); +__m128i psi_i_p3_m5 __attribute__ ((aligned(16))); +__m128i psi_i_p3_m3 __attribute__ ((aligned(16))); +__m128i psi_i_p3_m1 __attribute__ ((aligned(16))); +__m128i psi_i_p3_p1 __attribute__ ((aligned(16))); +__m128i psi_i_p3_p3 __attribute__ ((aligned(16))); +__m128i psi_i_p3_p5 __attribute__ ((aligned(16))); +__m128i psi_i_p3_p7 __attribute__ ((aligned(16))); +__m128i psi_i_p5_m7 __attribute__ ((aligned(16))); +__m128i psi_i_p5_m5 __attribute__ ((aligned(16))); +__m128i psi_i_p5_m3 __attribute__ ((aligned(16))); +__m128i psi_i_p5_m1 __attribute__ ((aligned(16))); +__m128i psi_i_p5_p1 __attribute__ ((aligned(16))); +__m128i psi_i_p5_p3 __attribute__ ((aligned(16))); +__m128i psi_i_p5_p5 __attribute__ ((aligned(16))); +__m128i psi_i_p5_p7 __attribute__ ((aligned(16))); +__m128i psi_i_p7_m7 __attribute__ ((aligned(16))); +__m128i psi_i_p7_m5 __attribute__ ((aligned(16))); +__m128i psi_i_p7_m3 __attribute__ ((aligned(16))); +__m128i psi_i_p7_m1 __attribute__ ((aligned(16))); +__m128i psi_i_p7_p1 __attribute__ ((aligned(16))); +__m128i psi_i_p7_p3 __attribute__ ((aligned(16))); +__m128i psi_i_p7_p5 __attribute__ ((aligned(16))); +__m128i psi_i_p7_p7 __attribute__ ((aligned(16))); + +__m128i a_r_m7_m7 __attribute__ ((aligned(16))); +__m128i a_r_m7_m5 __attribute__ ((aligned(16))); +__m128i a_r_m7_m3 __attribute__ ((aligned(16))); +__m128i a_r_m7_m1 __attribute__ ((aligned(16))); +__m128i a_r_m7_p1 __attribute__ ((aligned(16))); +__m128i a_r_m7_p3 __attribute__ ((aligned(16))); +__m128i a_r_m7_p5 __attribute__ ((aligned(16))); +__m128i a_r_m7_p7 __attribute__ ((aligned(16))); +__m128i a_r_m5_m7 __attribute__ ((aligned(16))); +__m128i a_r_m5_m5 __attribute__ ((aligned(16))); +__m128i a_r_m5_m3 __attribute__ ((aligned(16))); +__m128i a_r_m5_m1 __attribute__ ((aligned(16))); +__m128i a_r_m5_p1 __attribute__ ((aligned(16))); +__m128i a_r_m5_p3 __attribute__ ((aligned(16))); +__m128i a_r_m5_p5 __attribute__ ((aligned(16))); +__m128i a_r_m5_p7 __attribute__ ((aligned(16))); +__m128i a_r_m3_m7 __attribute__ ((aligned(16))); +__m128i a_r_m3_m5 __attribute__ ((aligned(16))); +__m128i a_r_m3_m3 __attribute__ ((aligned(16))); +__m128i a_r_m3_m1 __attribute__ ((aligned(16))); +__m128i a_r_m3_p1 __attribute__ ((aligned(16))); +__m128i a_r_m3_p3 __attribute__ ((aligned(16))); +__m128i a_r_m3_p5 __attribute__ ((aligned(16))); +__m128i a_r_m3_p7 __attribute__ ((aligned(16))); +__m128i a_r_m1_m7 __attribute__ ((aligned(16))); +__m128i a_r_m1_m5 __attribute__ ((aligned(16))); +__m128i a_r_m1_m3 __attribute__ ((aligned(16))); +__m128i a_r_m1_m1 __attribute__ ((aligned(16))); +__m128i a_r_m1_p1 __attribute__ ((aligned(16))); +__m128i a_r_m1_p3 __attribute__ ((aligned(16))); +__m128i a_r_m1_p5 __attribute__ ((aligned(16))); +__m128i a_r_m1_p7 __attribute__ ((aligned(16))); +__m128i a_r_p1_m7 __attribute__ ((aligned(16))); +__m128i a_r_p1_m5 __attribute__ ((aligned(16))); +__m128i a_r_p1_m3 __attribute__ ((aligned(16))); +__m128i a_r_p1_m1 __attribute__ ((aligned(16))); +__m128i a_r_p1_p1 __attribute__ ((aligned(16))); +__m128i a_r_p1_p3 __attribute__ ((aligned(16))); +__m128i a_r_p1_p5 __attribute__ ((aligned(16))); +__m128i a_r_p1_p7 __attribute__ ((aligned(16))); +__m128i a_r_p3_m7 __attribute__ ((aligned(16))); +__m128i a_r_p3_m5 __attribute__ ((aligned(16))); +__m128i a_r_p3_m3 __attribute__ ((aligned(16))); +__m128i a_r_p3_m1 __attribute__ ((aligned(16))); +__m128i a_r_p3_p1 __attribute__ ((aligned(16))); +__m128i a_r_p3_p3 __attribute__ ((aligned(16))); +__m128i a_r_p3_p5 __attribute__ ((aligned(16))); +__m128i a_r_p3_p7 __attribute__ ((aligned(16))); +__m128i a_r_p5_m7 __attribute__ ((aligned(16))); +__m128i a_r_p5_m5 __attribute__ ((aligned(16))); +__m128i a_r_p5_m3 __attribute__ ((aligned(16))); +__m128i a_r_p5_m1 __attribute__ ((aligned(16))); +__m128i a_r_p5_p1 __attribute__ ((aligned(16))); +__m128i a_r_p5_p3 __attribute__ ((aligned(16))); +__m128i a_r_p5_p5 __attribute__ ((aligned(16))); +__m128i a_r_p5_p7 __attribute__ ((aligned(16))); +__m128i a_r_p7_m7 __attribute__ ((aligned(16))); +__m128i a_r_p7_m5 __attribute__ ((aligned(16))); +__m128i a_r_p7_m3 __attribute__ ((aligned(16))); +__m128i a_r_p7_m1 __attribute__ ((aligned(16))); +__m128i a_r_p7_p1 __attribute__ ((aligned(16))); +__m128i a_r_p7_p3 __attribute__ ((aligned(16))); +__m128i a_r_p7_p5 __attribute__ ((aligned(16))); +__m128i a_r_p7_p7 __attribute__ ((aligned(16))); + +__m128i a_i_m7_m7 __attribute__ ((aligned(16))); +__m128i a_i_m7_m5 __attribute__ ((aligned(16))); +__m128i a_i_m7_m3 __attribute__ ((aligned(16))); +__m128i a_i_m7_m1 __attribute__ ((aligned(16))); +__m128i a_i_m7_p1 __attribute__ ((aligned(16))); +__m128i a_i_m7_p3 __attribute__ ((aligned(16))); +__m128i a_i_m7_p5 __attribute__ ((aligned(16))); +__m128i a_i_m7_p7 __attribute__ ((aligned(16))); +__m128i a_i_m5_m7 __attribute__ ((aligned(16))); +__m128i a_i_m5_m5 __attribute__ ((aligned(16))); +__m128i a_i_m5_m3 __attribute__ ((aligned(16))); +__m128i a_i_m5_m1 __attribute__ ((aligned(16))); +__m128i a_i_m5_p1 __attribute__ ((aligned(16))); +__m128i a_i_m5_p3 __attribute__ ((aligned(16))); +__m128i a_i_m5_p5 __attribute__ ((aligned(16))); +__m128i a_i_m5_p7 __attribute__ ((aligned(16))); +__m128i a_i_m3_m7 __attribute__ ((aligned(16))); +__m128i a_i_m3_m5 __attribute__ ((aligned(16))); +__m128i a_i_m3_m3 __attribute__ ((aligned(16))); +__m128i a_i_m3_m1 __attribute__ ((aligned(16))); +__m128i a_i_m3_p1 __attribute__ ((aligned(16))); +__m128i a_i_m3_p3 __attribute__ ((aligned(16))); +__m128i a_i_m3_p5 __attribute__ ((aligned(16))); +__m128i a_i_m3_p7 __attribute__ ((aligned(16))); +__m128i a_i_m1_m7 __attribute__ ((aligned(16))); +__m128i a_i_m1_m5 __attribute__ ((aligned(16))); +__m128i a_i_m1_m3 __attribute__ ((aligned(16))); +__m128i a_i_m1_m1 __attribute__ ((aligned(16))); +__m128i a_i_m1_p1 __attribute__ ((aligned(16))); +__m128i a_i_m1_p3 __attribute__ ((aligned(16))); +__m128i a_i_m1_p5 __attribute__ ((aligned(16))); +__m128i a_i_m1_p7 __attribute__ ((aligned(16))); +__m128i a_i_p1_m7 __attribute__ ((aligned(16))); +__m128i a_i_p1_m5 __attribute__ ((aligned(16))); +__m128i a_i_p1_m3 __attribute__ ((aligned(16))); +__m128i a_i_p1_m1 __attribute__ ((aligned(16))); +__m128i a_i_p1_p1 __attribute__ ((aligned(16))); +__m128i a_i_p1_p3 __attribute__ ((aligned(16))); +__m128i a_i_p1_p5 __attribute__ ((aligned(16))); +__m128i a_i_p1_p7 __attribute__ ((aligned(16))); +__m128i a_i_p3_m7 __attribute__ ((aligned(16))); +__m128i a_i_p3_m5 __attribute__ ((aligned(16))); +__m128i a_i_p3_m3 __attribute__ ((aligned(16))); +__m128i a_i_p3_m1 __attribute__ ((aligned(16))); +__m128i a_i_p3_p1 __attribute__ ((aligned(16))); +__m128i a_i_p3_p3 __attribute__ ((aligned(16))); +__m128i a_i_p3_p5 __attribute__ ((aligned(16))); +__m128i a_i_p3_p7 __attribute__ ((aligned(16))); +__m128i a_i_p5_m7 __attribute__ ((aligned(16))); +__m128i a_i_p5_m5 __attribute__ ((aligned(16))); +__m128i a_i_p5_m3 __attribute__ ((aligned(16))); +__m128i a_i_p5_m1 __attribute__ ((aligned(16))); +__m128i a_i_p5_p1 __attribute__ ((aligned(16))); +__m128i a_i_p5_p3 __attribute__ ((aligned(16))); +__m128i a_i_p5_p5 __attribute__ ((aligned(16))); +__m128i a_i_p5_p7 __attribute__ ((aligned(16))); +__m128i a_i_p7_m7 __attribute__ ((aligned(16))); +__m128i a_i_p7_m5 __attribute__ ((aligned(16))); +__m128i a_i_p7_m3 __attribute__ ((aligned(16))); +__m128i a_i_p7_m1 __attribute__ ((aligned(16))); +__m128i a_i_p7_p1 __attribute__ ((aligned(16))); +__m128i a_i_p7_p3 __attribute__ ((aligned(16))); +__m128i a_i_p7_p5 __attribute__ ((aligned(16))); +__m128i a_i_p7_p7 __attribute__ ((aligned(16))); + +__m128i psi_a_m7_m7 __attribute__ ((aligned(16))); +__m128i psi_a_m7_m5 __attribute__ ((aligned(16))); +__m128i psi_a_m7_m3 __attribute__ ((aligned(16))); +__m128i psi_a_m7_m1 __attribute__ ((aligned(16))); +__m128i psi_a_m7_p1 __attribute__ ((aligned(16))); +__m128i psi_a_m7_p3 __attribute__ ((aligned(16))); +__m128i psi_a_m7_p5 __attribute__ ((aligned(16))); +__m128i psi_a_m7_p7 __attribute__ ((aligned(16))); +__m128i psi_a_m5_m7 __attribute__ ((aligned(16))); +__m128i psi_a_m5_m5 __attribute__ ((aligned(16))); +__m128i psi_a_m5_m3 __attribute__ ((aligned(16))); +__m128i psi_a_m5_m1 __attribute__ ((aligned(16))); +__m128i psi_a_m5_p1 __attribute__ ((aligned(16))); +__m128i psi_a_m5_p3 __attribute__ ((aligned(16))); +__m128i psi_a_m5_p5 __attribute__ ((aligned(16))); +__m128i psi_a_m5_p7 __attribute__ ((aligned(16))); +__m128i psi_a_m3_m7 __attribute__ ((aligned(16))); +__m128i psi_a_m3_m5 __attribute__ ((aligned(16))); +__m128i psi_a_m3_m3 __attribute__ ((aligned(16))); +__m128i psi_a_m3_m1 __attribute__ ((aligned(16))); +__m128i psi_a_m3_p1 __attribute__ ((aligned(16))); +__m128i psi_a_m3_p3 __attribute__ ((aligned(16))); +__m128i psi_a_m3_p5 __attribute__ ((aligned(16))); +__m128i psi_a_m3_p7 __attribute__ ((aligned(16))); +__m128i psi_a_m1_m7 __attribute__ ((aligned(16))); +__m128i psi_a_m1_m5 __attribute__ ((aligned(16))); +__m128i psi_a_m1_m3 __attribute__ ((aligned(16))); +__m128i psi_a_m1_m1 __attribute__ ((aligned(16))); +__m128i psi_a_m1_p1 __attribute__ ((aligned(16))); +__m128i psi_a_m1_p3 __attribute__ ((aligned(16))); +__m128i psi_a_m1_p5 __attribute__ ((aligned(16))); +__m128i psi_a_m1_p7 __attribute__ ((aligned(16))); +__m128i psi_a_p1_m7 __attribute__ ((aligned(16))); +__m128i psi_a_p1_m5 __attribute__ ((aligned(16))); +__m128i psi_a_p1_m3 __attribute__ ((aligned(16))); +__m128i psi_a_p1_m1 __attribute__ ((aligned(16))); +__m128i psi_a_p1_p1 __attribute__ ((aligned(16))); +__m128i psi_a_p1_p3 __attribute__ ((aligned(16))); +__m128i psi_a_p1_p5 __attribute__ ((aligned(16))); +__m128i psi_a_p1_p7 __attribute__ ((aligned(16))); +__m128i psi_a_p3_m7 __attribute__ ((aligned(16))); +__m128i psi_a_p3_m5 __attribute__ ((aligned(16))); +__m128i psi_a_p3_m3 __attribute__ ((aligned(16))); +__m128i psi_a_p3_m1 __attribute__ ((aligned(16))); +__m128i psi_a_p3_p1 __attribute__ ((aligned(16))); +__m128i psi_a_p3_p3 __attribute__ ((aligned(16))); +__m128i psi_a_p3_p5 __attribute__ ((aligned(16))); +__m128i psi_a_p3_p7 __attribute__ ((aligned(16))); +__m128i psi_a_p5_m7 __attribute__ ((aligned(16))); +__m128i psi_a_p5_m5 __attribute__ ((aligned(16))); +__m128i psi_a_p5_m3 __attribute__ ((aligned(16))); +__m128i psi_a_p5_m1 __attribute__ ((aligned(16))); +__m128i psi_a_p5_p1 __attribute__ ((aligned(16))); +__m128i psi_a_p5_p3 __attribute__ ((aligned(16))); +__m128i psi_a_p5_p5 __attribute__ ((aligned(16))); +__m128i psi_a_p5_p7 __attribute__ ((aligned(16))); +__m128i psi_a_p7_m7 __attribute__ ((aligned(16))); +__m128i psi_a_p7_m5 __attribute__ ((aligned(16))); +__m128i psi_a_p7_m3 __attribute__ ((aligned(16))); +__m128i psi_a_p7_m1 __attribute__ ((aligned(16))); +__m128i psi_a_p7_p1 __attribute__ ((aligned(16))); +__m128i psi_a_p7_p3 __attribute__ ((aligned(16))); +__m128i psi_a_p7_p5 __attribute__ ((aligned(16))); +__m128i psi_a_p7_p7 __attribute__ ((aligned(16))); + +__m128i a_sq_m7_m7 __attribute__ ((aligned(16))); +__m128i a_sq_m7_m5 __attribute__ ((aligned(16))); +__m128i a_sq_m7_m3 __attribute__ ((aligned(16))); +__m128i a_sq_m7_m1 __attribute__ ((aligned(16))); +__m128i a_sq_m7_p1 __attribute__ ((aligned(16))); +__m128i a_sq_m7_p3 __attribute__ ((aligned(16))); +__m128i a_sq_m7_p5 __attribute__ ((aligned(16))); +__m128i a_sq_m7_p7 __attribute__ ((aligned(16))); +__m128i a_sq_m5_m7 __attribute__ ((aligned(16))); +__m128i a_sq_m5_m5 __attribute__ ((aligned(16))); +__m128i a_sq_m5_m3 __attribute__ ((aligned(16))); +__m128i a_sq_m5_m1 __attribute__ ((aligned(16))); +__m128i a_sq_m5_p1 __attribute__ ((aligned(16))); +__m128i a_sq_m5_p3 __attribute__ ((aligned(16))); +__m128i a_sq_m5_p5 __attribute__ ((aligned(16))); +__m128i a_sq_m5_p7 __attribute__ ((aligned(16))); +__m128i a_sq_m3_m7 __attribute__ ((aligned(16))); +__m128i a_sq_m3_m5 __attribute__ ((aligned(16))); +__m128i a_sq_m3_m3 __attribute__ ((aligned(16))); +__m128i a_sq_m3_m1 __attribute__ ((aligned(16))); +__m128i a_sq_m3_p1 __attribute__ ((aligned(16))); +__m128i a_sq_m3_p3 __attribute__ ((aligned(16))); +__m128i a_sq_m3_p5 __attribute__ ((aligned(16))); +__m128i a_sq_m3_p7 __attribute__ ((aligned(16))); +__m128i a_sq_m1_m7 __attribute__ ((aligned(16))); +__m128i a_sq_m1_m5 __attribute__ ((aligned(16))); +__m128i a_sq_m1_m3 __attribute__ ((aligned(16))); +__m128i a_sq_m1_m1 __attribute__ ((aligned(16))); +__m128i a_sq_m1_p1 __attribute__ ((aligned(16))); +__m128i a_sq_m1_p3 __attribute__ ((aligned(16))); +__m128i a_sq_m1_p5 __attribute__ ((aligned(16))); +__m128i a_sq_m1_p7 __attribute__ ((aligned(16))); +__m128i a_sq_p1_m7 __attribute__ ((aligned(16))); +__m128i a_sq_p1_m5 __attribute__ ((aligned(16))); +__m128i a_sq_p1_m3 __attribute__ ((aligned(16))); +__m128i a_sq_p1_m1 __attribute__ ((aligned(16))); +__m128i a_sq_p1_p1 __attribute__ ((aligned(16))); +__m128i a_sq_p1_p3 __attribute__ ((aligned(16))); +__m128i a_sq_p1_p5 __attribute__ ((aligned(16))); +__m128i a_sq_p1_p7 __attribute__ ((aligned(16))); +__m128i a_sq_p3_m7 __attribute__ ((aligned(16))); +__m128i a_sq_p3_m5 __attribute__ ((aligned(16))); +__m128i a_sq_p3_m3 __attribute__ ((aligned(16))); +__m128i a_sq_p3_m1 __attribute__ ((aligned(16))); +__m128i a_sq_p3_p1 __attribute__ ((aligned(16))); +__m128i a_sq_p3_p3 __attribute__ ((aligned(16))); +__m128i a_sq_p3_p5 __attribute__ ((aligned(16))); +__m128i a_sq_p3_p7 __attribute__ ((aligned(16))); +__m128i a_sq_p5_m7 __attribute__ ((aligned(16))); +__m128i a_sq_p5_m5 __attribute__ ((aligned(16))); +__m128i a_sq_p5_m3 __attribute__ ((aligned(16))); +__m128i a_sq_p5_m1 __attribute__ ((aligned(16))); +__m128i a_sq_p5_p1 __attribute__ ((aligned(16))); +__m128i a_sq_p5_p3 __attribute__ ((aligned(16))); +__m128i a_sq_p5_p5 __attribute__ ((aligned(16))); +__m128i a_sq_p5_p7 __attribute__ ((aligned(16))); +__m128i a_sq_p7_m7 __attribute__ ((aligned(16))); +__m128i a_sq_p7_m5 __attribute__ ((aligned(16))); +__m128i a_sq_p7_m3 __attribute__ ((aligned(16))); +__m128i a_sq_p7_m1 __attribute__ ((aligned(16))); +__m128i a_sq_p7_p1 __attribute__ ((aligned(16))); +__m128i a_sq_p7_p3 __attribute__ ((aligned(16))); +__m128i a_sq_p7_p5 __attribute__ ((aligned(16))); +__m128i a_sq_p7_p7 __attribute__ ((aligned(16))); + +__m128i bit_met_m7_m7 __attribute__ ((aligned(16))); +__m128i bit_met_m7_m5 __attribute__ ((aligned(16))); +__m128i bit_met_m7_m3 __attribute__ ((aligned(16))); +__m128i bit_met_m7_m1 __attribute__ ((aligned(16))); +__m128i bit_met_m7_p1 __attribute__ ((aligned(16))); +__m128i bit_met_m7_p3 __attribute__ ((aligned(16))); +__m128i bit_met_m7_p5 __attribute__ ((aligned(16))); +__m128i bit_met_m7_p7 __attribute__ ((aligned(16))); +__m128i bit_met_m5_m7 __attribute__ ((aligned(16))); +__m128i bit_met_m5_m5 __attribute__ ((aligned(16))); +__m128i bit_met_m5_m3 __attribute__ ((aligned(16))); +__m128i bit_met_m5_m1 __attribute__ ((aligned(16))); +__m128i bit_met_m5_p1 __attribute__ ((aligned(16))); +__m128i bit_met_m5_p3 __attribute__ ((aligned(16))); +__m128i bit_met_m5_p5 __attribute__ ((aligned(16))); +__m128i bit_met_m5_p7 __attribute__ ((aligned(16))); +__m128i bit_met_m3_m7 __attribute__ ((aligned(16))); +__m128i bit_met_m3_m5 __attribute__ ((aligned(16))); +__m128i bit_met_m3_m3 __attribute__ ((aligned(16))); +__m128i bit_met_m3_m1 __attribute__ ((aligned(16))); +__m128i bit_met_m3_p1 __attribute__ ((aligned(16))); +__m128i bit_met_m3_p3 __attribute__ ((aligned(16))); +__m128i bit_met_m3_p5 __attribute__ ((aligned(16))); +__m128i bit_met_m3_p7 __attribute__ ((aligned(16))); +__m128i bit_met_m1_m7 __attribute__ ((aligned(16))); +__m128i bit_met_m1_m5 __attribute__ ((aligned(16))); +__m128i bit_met_m1_m3 __attribute__ ((aligned(16))); +__m128i bit_met_m1_m1 __attribute__ ((aligned(16))); +__m128i bit_met_m1_p1 __attribute__ ((aligned(16))); +__m128i bit_met_m1_p3 __attribute__ ((aligned(16))); +__m128i bit_met_m1_p5 __attribute__ ((aligned(16))); +__m128i bit_met_m1_p7 __attribute__ ((aligned(16))); +__m128i bit_met_p1_m7 __attribute__ ((aligned(16))); +__m128i bit_met_p1_m5 __attribute__ ((aligned(16))); +__m128i bit_met_p1_m3 __attribute__ ((aligned(16))); +__m128i bit_met_p1_m1 __attribute__ ((aligned(16))); +__m128i bit_met_p1_p1 __attribute__ ((aligned(16))); +__m128i bit_met_p1_p3 __attribute__ ((aligned(16))); +__m128i bit_met_p1_p5 __attribute__ ((aligned(16))); +__m128i bit_met_p1_p7 __attribute__ ((aligned(16))); +__m128i bit_met_p3_m7 __attribute__ ((aligned(16))); +__m128i bit_met_p3_m5 __attribute__ ((aligned(16))); +__m128i bit_met_p3_m3 __attribute__ ((aligned(16))); +__m128i bit_met_p3_m1 __attribute__ ((aligned(16))); +__m128i bit_met_p3_p1 __attribute__ ((aligned(16))); +__m128i bit_met_p3_p3 __attribute__ ((aligned(16))); +__m128i bit_met_p3_p5 __attribute__ ((aligned(16))); +__m128i bit_met_p3_p7 __attribute__ ((aligned(16))); +__m128i bit_met_p5_m7 __attribute__ ((aligned(16))); +__m128i bit_met_p5_m5 __attribute__ ((aligned(16))); +__m128i bit_met_p5_m3 __attribute__ ((aligned(16))); +__m128i bit_met_p5_m1 __attribute__ ((aligned(16))); +__m128i bit_met_p5_p1 __attribute__ ((aligned(16))); +__m128i bit_met_p5_p3 __attribute__ ((aligned(16))); +__m128i bit_met_p5_p5 __attribute__ ((aligned(16))); +__m128i bit_met_p5_p7 __attribute__ ((aligned(16))); +__m128i bit_met_p7_m7 __attribute__ ((aligned(16))); +__m128i bit_met_p7_m5 __attribute__ ((aligned(16))); +__m128i bit_met_p7_m3 __attribute__ ((aligned(16))); +__m128i bit_met_p7_m1 __attribute__ ((aligned(16))); +__m128i bit_met_p7_p1 __attribute__ ((aligned(16))); +__m128i bit_met_p7_p3 __attribute__ ((aligned(16))); +__m128i bit_met_p7_p5 __attribute__ ((aligned(16))); +__m128i bit_met_p7_p7 __attribute__ ((aligned(16))); + +__m128i y0_p_1_1 __attribute__ ((aligned(16))); +__m128i y0_p_1_3 __attribute__ ((aligned(16))); +__m128i y0_p_1_5 __attribute__ ((aligned(16))); +__m128i y0_p_1_7 __attribute__ ((aligned(16))); +__m128i y0_p_3_1 __attribute__ ((aligned(16))); +__m128i y0_p_3_3 __attribute__ ((aligned(16))); +__m128i y0_p_3_5 __attribute__ ((aligned(16))); +__m128i y0_p_3_7 __attribute__ ((aligned(16))); +__m128i y0_p_5_1 __attribute__ ((aligned(16))); +__m128i y0_p_5_3 __attribute__ ((aligned(16))); +__m128i y0_p_5_5 __attribute__ ((aligned(16))); +__m128i y0_p_5_7 __attribute__ ((aligned(16))); +__m128i y0_p_7_1 __attribute__ ((aligned(16))); +__m128i y0_p_7_3 __attribute__ ((aligned(16))); +__m128i y0_p_7_5 __attribute__ ((aligned(16))); +__m128i y0_p_7_7 __attribute__ ((aligned(16))); +__m128i y0_m_1_1 __attribute__ ((aligned(16))); +__m128i y0_m_1_3 __attribute__ ((aligned(16))); +__m128i y0_m_1_5 __attribute__ ((aligned(16))); +__m128i y0_m_1_7 __attribute__ ((aligned(16))); +__m128i y0_m_3_1 __attribute__ ((aligned(16))); +__m128i y0_m_3_3 __attribute__ ((aligned(16))); +__m128i y0_m_3_5 __attribute__ ((aligned(16))); +__m128i y0_m_3_7 __attribute__ ((aligned(16))); +__m128i y0_m_5_1 __attribute__ ((aligned(16))); +__m128i y0_m_5_3 __attribute__ ((aligned(16))); +__m128i y0_m_5_5 __attribute__ ((aligned(16))); +__m128i y0_m_5_7 __attribute__ ((aligned(16))); +__m128i y0_m_7_1 __attribute__ ((aligned(16))); +__m128i y0_m_7_3 __attribute__ ((aligned(16))); +__m128i y0_m_7_5 __attribute__ ((aligned(16))); +__m128i y0_m_7_7 __attribute__ ((aligned(16))); + +__m128i xmm0 __attribute__ ((aligned(16))); +__m128i xmm1 __attribute__ ((aligned(16))); +__m128i xmm2 __attribute__ ((aligned(16))); +__m128i xmm3 __attribute__ ((aligned(16))); +__m128i xmm4 __attribute__ ((aligned(16))); +__m128i xmm5 __attribute__ ((aligned(16))); +__m128i xmm6 __attribute__ ((aligned(16))); +__m128i xmm7 __attribute__ ((aligned(16))); +__m128i xmm8 __attribute__ ((aligned(16))); + +__m128i y0r __attribute__ ((aligned(16))); +__m128i y0i __attribute__ ((aligned(16))); +__m128i y1r __attribute__ ((aligned(16))); +__m128i y1i __attribute__ ((aligned(16))); +__m128i y2r __attribute__ ((aligned(16))); +__m128i y2i __attribute__ ((aligned(16))); + +__m128i logmax_num_re0 __attribute__ ((aligned(16))); +__m128i logmax_num_im0 __attribute__ ((aligned(16))); +__m128i logmax_den_re0 __attribute__ ((aligned(16))); +__m128i logmax_den_im0 __attribute__ ((aligned(16))); +__m128i logmax_num_re1 __attribute__ ((aligned(16))); +__m128i logmax_num_im1 __attribute__ ((aligned(16))); +__m128i logmax_den_re1 __attribute__ ((aligned(16))); +__m128i logmax_den_im1 __attribute__ ((aligned(16))); + +__m128i tmp_result __attribute__ ((aligned(16))); +__m128i tmp_result2 __attribute__ ((aligned(16))); +__m128i tmp_result3 __attribute__ ((aligned(16))); +__m128i tmp_result4 __attribute__ ((aligned(16))); + + +//============================================================================================== +// Auxiliary Makros + +// calculates psi_a = psi_r*a_r + psi_i*a_i +#define prodsum_psi_a_epi16(psi_r,a_r,psi_i,a_i,psi_a) tmp_result = _mm_mulhi_epi16(psi_r,a_r); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result2 = _mm_mulhi_epi16(psi_i,a_i); tmp_result2 = _mm_slli_epi16(tmp_result2,1); psi_a = _mm_adds_epi16(tmp_result,tmp_result2); + +// calculate interference magnitude +#define interference_abs_epi16(psi,int_ch_mag,int_mag,c1,c2) tmp_result = _mm_cmplt_epi16(psi,int_ch_mag); tmp_result2 = _mm_xor_si128(tmp_result,(*(__m128i*)&ones[0])); tmp_result = _mm_and_si128(tmp_result,c1); tmp_result2 = _mm_and_si128(tmp_result2,c2); int_mag = _mm_or_si128(tmp_result,tmp_result2); + +// calculate interference magnitude +// tmp_result = ones in shorts corr. to interval 2<=x<=4, tmp_result2 interval < 2, tmp_result3 interval 46 +#define interference_abs_64qam_epi16(psi,int_ch_mag,int_two_ch_mag,int_three_ch_mag,a,c1,c3,c5,c7) tmp_result = _mm_cmplt_epi16(psi,int_two_ch_mag); tmp_result3 = _mm_xor_si128(tmp_result,(*(__m128i*)&ones[0])); tmp_result2 = _mm_cmplt_epi16(psi,int_ch_mag); tmp_result = _mm_xor_si128(tmp_result,tmp_result2); tmp_result4 = _mm_cmpgt_epi16(psi,int_three_ch_mag); tmp_result3 = _mm_xor_si128(tmp_result3,tmp_result4); tmp_result = _mm_and_si128(tmp_result,c3); tmp_result2 = _mm_and_si128(tmp_result2,c1); tmp_result3 = _mm_and_si128(tmp_result3,c5); tmp_result4 = _mm_and_si128(tmp_result4,c7); tmp_result = _mm_or_si128(tmp_result,tmp_result2); tmp_result3 = _mm_or_si128(tmp_result3,tmp_result4); a = _mm_or_si128(tmp_result,tmp_result3); + +// calculates a_sq = int_ch_mag*(a_r^2 + a_i^2)*scale_factor +#define square_a_epi16(a_r,a_i,int_ch_mag,scale_factor,a_sq) tmp_result = _mm_mulhi_epi16(a_r,a_r); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result = _mm_mulhi_epi16(tmp_result,scale_factor); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result = _mm_mulhi_epi16(tmp_result,int_ch_mag); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result2 = _mm_mulhi_epi16(a_i,a_i); tmp_result2 = _mm_slli_epi16(tmp_result2,1); tmp_result2 = _mm_mulhi_epi16(tmp_result2,scale_factor); tmp_result2 = _mm_slli_epi16(tmp_result2,1); tmp_result2 = _mm_mulhi_epi16(tmp_result2,int_ch_mag); tmp_result2 = _mm_slli_epi16(tmp_result2,1); a_sq = _mm_adds_epi16(tmp_result,tmp_result2); + +// calculates a_sq = int_ch_mag*(a_r^2 + a_i^2)*scale_factor for 64-QAM +#define square_a_64qam_epi16(a_r,a_i,int_ch_mag,scale_factor,a_sq) tmp_result = _mm_mulhi_epi16(a_r,a_r); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result = _mm_mulhi_epi16(tmp_result,scale_factor); tmp_result = _mm_slli_epi16(tmp_result,3); tmp_result = _mm_mulhi_epi16(tmp_result,int_ch_mag); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result2 = _mm_mulhi_epi16(a_i,a_i); tmp_result2 = _mm_slli_epi16(tmp_result2,1); tmp_result2 = _mm_mulhi_epi16(tmp_result2,scale_factor); tmp_result2 = _mm_slli_epi16(tmp_result2,3); tmp_result2 = _mm_mulhi_epi16(tmp_result2,int_ch_mag); tmp_result2 = _mm_slli_epi16(tmp_result2,1); a_sq = _mm_adds_epi16(tmp_result,tmp_result2); + +#elif defined(__arm__) + +#endif + +//============================================================================================== +// SINGLE-STREAM +//============================================================================================== + +//---------------------------------------------------------------------------------------------- +// QPSK +//---------------------------------------------------------------------------------------------- + +int nr_dlsch_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, + int32_t **rxdataF_comp, + int16_t *dlsch_llr, + uint8_t symbol, + uint8_t first_symbol_flag, + uint16_t nb_rb, + uint8_t beamforming_mode) +{ + + uint32_t *rxF = (uint32_t*)&rxdataF_comp[0][((int32_t)symbol*nb_rb*12)]; + uint32_t *llr32; + int i,len; + + llr32 = (uint32_t*)dlsch_llr; + if (!llr32) { + LOG_E(PHY,"nr_dlsch_qpsk_llr: llr is null, symbol %d, llr32=%p\n",symbol, llr32); + return(-1); + } + + if (symbol ==2) //to update from config + len = nb_rb*6; + else + len = nb_rb*12; + /* + LOG_I(PHY,"dlsch_qpsk_llr: [symb %d / Length %d]: @LLR Buff %x, @LLR Buff(symb) %x \n", + symbol, + len, + dlsch_llr, + llr32); + */ + for (i=0; i>=2; // length in quad words (4 REs) + // printf("len>>=2=%d\n", len); + len+=(len_mod4==0 ? 0 : 1); + // printf("len+=%d\n", len); + for (i=0; iN_RB_DL*12)]; + ch_magb = (__m128i*)&dl_ch_magb[0][(symbol*frame_parms->N_RB_DL*12)]; +#elif defined(__arm__) + ch_mag = (int16x8_t*)&dl_ch_mag[0][(symbol*frame_parms->N_RB_DL*12)]; + ch_magb = (int16x8_t*)&dl_ch_magb[0][(symbol*frame_parms->N_RB_DL*12)]; +#endif + + if (symbol ==2) //to update from config + len = nb_rb*6; + else + len = nb_rb*12; + +// printf("nr_dlsch_64qam_llr: symbol %d,nb_rb %d, len %d,pbch_pss_sss_adjust %d\n",symbol,nb_rb,len,pbch_pss_sss_adjust); + +/* LOG_I(PHY,"nr_dlsch_64qam_llr [symb %d / FirstSym %d / Length %d]: @LLR Buff %x \n", + symbol, + first_symbol_flag, + len, + dlsch_llr, + pllr_symbol);*/ + + llr2 = llr; + llr += (len*6); + + len_mod4 =len&3; + len2=len>>2; // length in quad words (4 REs) + len2+=((len_mod4==0)?0:1); + + for (i=0; iN_RB_DL*12]; + int16_t rho_rho_amp_x0[2*frame_parms->N_RB_DL*12]; + uint16_t amp_tmp; + uint16_t *llr32=(uint16_t*)dlsch_llr; + int i, len, nsymb, len2; + uint8_t symbol, symbol_mod; + int len_acc=0; + uint16_t *sic_data; + uint16_t pbch_pss_sss_adjust; + unsigned char len_mod4=0; + uint16_t *llr2; + __m128i *ch_mag,*ch_magb; + + nsymb = (frame_parms->Ncp==0) ? 14:12; + + for (symbol=num_pdcch_symbols; symbolN_RB_DL*12)]); + int16_t *rho_1=(int16_t*)(&rho_i[0][((int16_t)symbol*frame_parms->N_RB_DL*12)]); + ch_mag = (__m128i*)(&dl_ch_mag[0][((int16_t)symbol*frame_parms->N_RB_DL*12)]); + ch_magb = (__m128i*)(&dl_ch_magb[0][((int16_t)symbol*frame_parms->N_RB_DL*12)]); + sic_data = (uint16_t*)(&sic_buffer[0][((int16_t)len_acc)]); + + symbol_mod = (symbol>=(7-frame_parms->Ncp)) ? symbol-(7-frame_parms->Ncp) : symbol; + + pbch_pss_sss_adjust=adjust_G2(frame_parms,&rb_alloc,6,subframe,symbol); + + if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) { + amp_tmp = 0x1fff;//dlsch0->sqrt_rho_b; already taken into account + if (frame_parms->nb_antenna_ports_eNB!=1) + len = nb_rb*8 - (2*pbch_pss_sss_adjust/3); + else + len = nb_rb*10 - (5*pbch_pss_sss_adjust/6); + } else { + amp_tmp = 0x1fff; //dlsch0->sqrt_rho_a; already taken into account + len = nb_rb*12 - pbch_pss_sss_adjust; + } + + if (mod_order_0==6) + amp_tmp=amp_tmp<<1; // to compensate for >> 1 shift in modulation + + len_acc+=len; + + multadd_complex_vector_real_scalar((int16_t *)sic_data, + amp_tmp, + (int16_t *)rho_amp_x0, //this is in Q13 + 1, + len); + + mult_cpx_vector((int16_t *)rho_1, //Q15 + (int16_t *)rho_amp_x0, //Q13 + (int16_t*)rho_rho_amp_x0, + len, + 13); + + sub_cpx_vector16((int16_t *)rxF, + (int16_t *)rho_rho_amp_x0, + //(int16_t *)clean_x1, + (int16_t *)rxF, + len*2); + + llr2 = llr32; + llr32 += (len*6); + + len_mod4 =len&3; + len2=len>>2; // length in quad words (4 REs) + len2+=(len_mod4?0:1); + + + + for (i=0; iN_RB_DL*12)]; + int16_t *rxF_i=(int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rho=(int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *llr16; + int len; + uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol; + + if (first_symbol_flag == 1) { + llr16 = (int16_t*)dlsch_llr; + } else { + llr16 = (int16_t*)(*llr16p); + } + + AssertFatal(llr16!=NULL,"nr_dlsch_qpsk_qpsk_llr: llr is null, symbol %d\n",symbol); + + if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) { + // if symbol has pilots + if (frame_parms->nb_antenna_ports_eNB!=1) + // in 2 antenna ports we have 8 REs per symbol per RB + len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3); + else + // for 1 antenna port we have 10 REs per symbol per RB + len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6); + } else { + // symbol has no pilots + len = (nb_rb*12) - pbch_pss_sss_adjust; + } + + // printf("nr_dlsch_qpsk_qpsk_llr: symbol %d,nb_rb %d, len %d,pbch_pss_sss_adjust %d\n",symbol,nb_rb,len,pbch_pss_sss_adjust); + // printf("qpsk_qpsk: len %d, llr16 %p\n",len,llr16); + qpsk_qpsk((short *)rxF, + (short *)rxF_i, + (short *)llr16, + (short *)rho, + len); + + llr16 += (len<<1); + *llr16p = (short *)llr16; + + return(0); +} + +//__m128i ONE_OVER_SQRT_8 __attribute__((aligned(16))); + +void qpsk_qpsk(short *stream0_in, + short *stream1_in, + short *stream0_out, + short *rho01, + int length + ) +{ + + /* + This function computes the LLRs of stream 0 (s_0) in presence of the interfering stream 1 (s_1) assuming that both symbols are QPSK. It can be used for both MU-MIMO interference-aware receiver or for SU-MIMO receivers. + + Parameters: + stream0_in = Matched filter output y0' = (h0*g0)*y0 + stream1_in = Matched filter output y1' = (h0*g1)*y0 + stream0_out = LLRs + rho01 = Correlation between the two effective channels \rho_{10} = (h1*g1)*(h0*g0) + length = number of resource elements + */ + +#if defined(__x86_64__) || defined(__i386__) + __m128i *rho01_128i = (__m128i *)rho01; + __m128i *stream0_128i_in = (__m128i *)stream0_in; + __m128i *stream1_128i_in = (__m128i *)stream1_in; + __m128i *stream0_128i_out = (__m128i *)stream0_out; + __m128i ONE_OVER_SQRT_8 = _mm_set1_epi16(23170); //round(2^16/sqrt(8)) +#elif defined(__arm__) + int16x8_t *rho01_128i = (int16x8_t *)rho01; + int16x8_t *stream0_128i_in = (int16x8_t *)stream0_in; + int16x8_t *stream1_128i_in = (int16x8_t *)stream1_in; + int16x8_t *stream0_128i_out = (int16x8_t *)stream0_out; + int16x8_t ONE_OVER_SQRT_8 = vdupq_n_s16(23170); //round(2^16/sqrt(8)) +#endif + + int i; + + + for (i=0; i>2; i+=2) { + // in each iteration, we take 8 complex samples +#if defined(__x86_64__) || defined(__i386__) + xmm0 = rho01_128i[i]; // 4 symbols + xmm1 = rho01_128i[i+1]; + + // put (rho_r + rho_i)/2sqrt2 in rho_rpi + // put (rho_r - rho_i)/2sqrt2 in rho_rmi + + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho) + xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho) + rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho) + rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho) + + // divide by sqrt(8), no shift needed ONE_OVER_SQRT_8 = Q1.16 + rho_rpi = _mm_mulhi_epi16(rho_rpi,ONE_OVER_SQRT_8); + rho_rmi = _mm_mulhi_epi16(rho_rmi,ONE_OVER_SQRT_8); +#elif defined(__arm__) + + +#endif + // Compute LLR for first bit of stream 0 + + // Compute real and imaginary parts of MF output for stream 0 +#if defined(__x86_64__) || defined(__i386__) + xmm0 = stream0_128i_in[i]; + xmm1 = stream0_128i_in[i+1]; + + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)] + y0i = _mm_unpackhi_epi64(xmm0,xmm1); + + y0r_over2 = _mm_srai_epi16(y0r,1); // divide by 2 + y0i_over2 = _mm_srai_epi16(y0i,1); // divide by 2 +#elif defined(__arm__) + + +#endif + // Compute real and imaginary parts of MF output for stream 1 +#if defined(__x86_64__) || defined(__i386__) + xmm0 = stream1_128i_in[i]; + xmm1 = stream1_128i_in[i+1]; + + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)] + y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)] + + y1r_over2 = _mm_srai_epi16(y1r,1); // divide by 2 + y1i_over2 = _mm_srai_epi16(y1i,1); // divide by 2 + + // Compute the terms for the LLR of first bit + + xmm0 = _mm_setzero_si128(); // ZERO + + // 1 term for numerator of LLR + xmm3 = _mm_subs_epi16(y1r_over2,rho_rpi); + A = _mm_abs_epi16(xmm3); // A = |y1r/2 - rho/sqrt(8)| + xmm2 = _mm_adds_epi16(A,y0i_over2); // = |y1r/2 - rho/sqrt(8)| + y0i/2 + xmm3 = _mm_subs_epi16(y1i_over2,rho_rmi); + B = _mm_abs_epi16(xmm3); // B = |y1i/2 - rho*/sqrt(8)| + logmax_num_re0 = _mm_adds_epi16(B,xmm2); // = |y1r/2 - rho/sqrt(8)|+|y1i/2 - rho*/sqrt(8)| + y0i/2 + + // 2 term for numerator of LLR + xmm3 = _mm_subs_epi16(y1r_over2,rho_rmi); + C = _mm_abs_epi16(xmm3); // C = |y1r/2 - rho*/4| + xmm2 = _mm_subs_epi16(C,y0i_over2); // = |y1r/2 - rho*/4| - y0i/2 + xmm3 = _mm_adds_epi16(y1i_over2,rho_rpi); + D = _mm_abs_epi16(xmm3); // D = |y1i/2 + rho/4| + xmm2 = _mm_adds_epi16(xmm2,D); // |y1r/2 - rho*/4| + |y1i/2 + rho/4| - y0i/2 + logmax_num_re0 = _mm_max_epi16(logmax_num_re0,xmm2); // max, numerator done + + // 1 term for denominator of LLR + xmm3 = _mm_adds_epi16(y1r_over2,rho_rmi); + E = _mm_abs_epi16(xmm3); // E = |y1r/2 + rho*/4| + xmm2 = _mm_adds_epi16(E,y0i_over2); // = |y1r/2 + rho*/4| + y0i/2 + xmm3 = _mm_subs_epi16(y1i_over2,rho_rpi); + F = _mm_abs_epi16(xmm3); // F = |y1i/2 - rho/4| + logmax_den_re0 = _mm_adds_epi16(F,xmm2); // = |y1r/2 + rho*/4| + |y1i/2 - rho/4| + y0i/2 + + // 2 term for denominator of LLR + xmm3 = _mm_adds_epi16(y1r_over2,rho_rpi); + G = _mm_abs_epi16(xmm3); // G = |y1r/2 + rho/4| + xmm2 = _mm_subs_epi16(G,y0i_over2); // = |y1r/2 + rho/4| - y0i/2 + xmm3 = _mm_adds_epi16(y1i_over2,rho_rmi); + H = _mm_abs_epi16(xmm3); // H = |y1i/2 + rho*/4| + xmm2 = _mm_adds_epi16(xmm2,H); // = |y1r/2 + rho/4| + |y1i/2 + rho*/4| - y0i/2 + logmax_den_re0 = _mm_max_epi16(logmax_den_re0,xmm2); // max, denominator done + + // Compute the terms for the LLR of first bit + + // 1 term for nominator of LLR + xmm2 = _mm_adds_epi16(A,y0r_over2); + logmax_num_im0 = _mm_adds_epi16(B,xmm2); // = |y1r/2 - rho/4| + |y1i/2 - rho*/4| + y0r/2 + + // 2 term for nominator of LLR + xmm2 = _mm_subs_epi16(E,y0r_over2); + xmm2 = _mm_adds_epi16(xmm2,F); // = |y1r/2 + rho*/4| + |y1i/2 - rho/4| - y0r/2 + + logmax_num_im0 = _mm_max_epi16(logmax_num_im0,xmm2); // max, nominator done + + // 1 term for denominator of LLR + xmm2 = _mm_adds_epi16(C,y0r_over2); + logmax_den_im0 = _mm_adds_epi16(D,xmm2); // = |y1r/2 - rho*/4| + |y1i/2 + rho/4| - y0r/2 + + xmm2 = _mm_subs_epi16(G,y0r_over2); + xmm2 = _mm_adds_epi16(xmm2,H); // = |y1r/2 + rho/4| + |y1i/2 + rho*/4| - y0r/2 + + logmax_den_im0 = _mm_max_epi16(logmax_den_im0,xmm2); // max, denominator done + + // LLR of first bit [L1(1), L1(2), L1(3), L1(4)] + y0r = _mm_adds_epi16(y0r,logmax_num_re0); + y0r = _mm_subs_epi16(y0r,logmax_den_re0); + + // LLR of second bit [L2(1), L2(2), L2(3), L2(4)] + y0i = _mm_adds_epi16(y0i,logmax_num_im0); + y0i = _mm_subs_epi16(y0i,logmax_den_im0); + + _mm_storeu_si128(&stream0_128i_out[i],_mm_unpacklo_epi16(y0r,y0i)); // = [L1(1), L2(1), L1(2), L2(2)] + + if (i<((length>>1) - 1)) // false if only 2 REs remain + _mm_storeu_si128(&stream0_128i_out[i+1],_mm_unpackhi_epi16(y0r,y0i)); + +#elif defined(__x86_64__) + +#endif + } + +#if defined(__x86_64__) || defined(__i386__) + _mm_empty(); + _m_empty(); +#endif +} + +int nr_dlsch_qpsk_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, + int32_t **rxdataF_comp, + int32_t **rxdataF_comp_i, + int32_t **dl_ch_mag_i, //|h_1|^2*(2/sqrt{10}) + int32_t **rho_i, + int16_t *dlsch_llr, + uint8_t symbol, + uint8_t first_symbol_flag, + uint16_t nb_rb, + uint16_t pbch_pss_sss_adjust, + int16_t **llr16p) +{ + + int16_t *rxF=(int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rxF_i=(int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag_i = (int16_t*)&dl_ch_mag_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rho=(int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *llr16; + int len; + uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol; + + if (first_symbol_flag == 1) { + llr16 = (int16_t*)dlsch_llr; + } else { + llr16 = (int16_t*)(*llr16p); + } + + AssertFatal(llr16!=NULL,"nr_dlsch_qpsk_qpsk_llr: llr is null, symbol %d\n",symbol); + + + if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) { + // if symbol has pilots + if (frame_parms->nb_antenna_ports_eNB!=1) + // in 2 antenna ports we have 8 REs per symbol per RB + len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3); + else + // for 1 antenna port we have 10 REs per symbol per RB + len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6); + } else { + // symbol has no pilots + len = (nb_rb*12) - pbch_pss_sss_adjust; + } + + qpsk_qam16((short *)rxF, + (short *)rxF_i, + (short *)ch_mag_i, + (short *)llr16, + (short *)rho, + len); + + llr16 += (len<<1); + *llr16p = (short *)llr16; + + return(0); +} + +/* +#if defined(__x86_64__) || defined(__i386__) +__m128i ONE_OVER_SQRT_2 __attribute__((aligned(16))); +__m128i ONE_OVER_SQRT_10 __attribute__((aligned(16))); +__m128i THREE_OVER_SQRT_10 __attribute__((aligned(16))); +__m128i ONE_OVER_SQRT_10_Q15 __attribute__((aligned(16))); +__m128i SQRT_10_OVER_FOUR __attribute__((aligned(16))); +__m128i ch_mag_int; +#endif +*/ +void qpsk_qam16(int16_t *stream0_in, + int16_t *stream1_in, + int16_t *ch_mag_i, + int16_t *stream0_out, + int16_t *rho01, + int32_t length + ) +{ + /* + This function computes the LLRs of stream 0 (s_0) in presence of the interfering stream 1 (s_1) assuming that both symbols are QPSK. It can be used for both MU-MIMO interference-aware receiver or for SU-MIMO receivers. + + Parameters: + stream0_in = Matched filter output y0' = (h0*g0)*y0 + stream1_in = Matched filter output y1' = (h0*g1)*y0 + stream0_out = LLRs + rho01 = Correlation between the two effective channels \rho_{10} = (h1*g1)*(h0*g0) + length = number of resource elements + */ + +#if defined(__x86_64__) || defined(__i386__) + __m128i *rho01_128i = (__m128i *)rho01; + __m128i *stream0_128i_in = (__m128i *)stream0_in; + __m128i *stream1_128i_in = (__m128i *)stream1_in; + __m128i *stream0_128i_out = (__m128i *)stream0_out; + __m128i *ch_mag_128i_i = (__m128i *)ch_mag_i; + __m128i ONE_OVER_SQRT_2 = _mm_set1_epi16(23170); // round(1/sqrt(2)*2^15) + __m128i ONE_OVER_SQRT_10_Q15 = _mm_set1_epi16(10362); // round(1/sqrt(10)*2^15) + __m128i THREE_OVER_SQRT_10 = _mm_set1_epi16(31086); // round(3/sqrt(10)*2^15) + __m128i SQRT_10_OVER_FOUR = _mm_set1_epi16(25905); // round(sqrt(10)/4*2^15) + __m128i ch_mag_int __attribute__((aligned(16))); +#elif defined(__arm__) + int16x8_t *rho01_128i = (int16x8_t *)rho01; + int16x8_t *stream0_128i_in = (int16x8_t *)stream0_in; + int16x8_t *stream1_128i_in = (int16x8_t *)stream1_in; + int16x8_t *stream0_128i_out = (int16x8_t *)stream0_out; + int16x8_t *ch_mag_128i_i = (int16x8_t *)ch_mag_i; + int16x8_t ONE_OVER_SQRT_2 = vdupq_n_s16(23170); // round(1/sqrt(2)*2^15) + int16x8_t ONE_OVER_SQRT_10_Q15 = vdupq_n_s16(10362); // round(1/sqrt(10)*2^15) + int16x8_t THREE_OVER_SQRT_10 = vdupq_n_s16(31086); // round(3/sqrt(10)*2^15) + int16x8_t SQRT_10_OVER_FOUR = vdupq_n_s16(25905); // round(sqrt(10)/4*2^15) + int16x8_t ch_mag_int __attribute__((aligned(16))); +#endif + +#ifdef DEBUG_LLR + print_shorts2("rho01_128i:\n",rho01_128i); +#endif + + int i; + + + for (i=0; i>2; i+=2) { + // in each iteration, we take 8 complex samples + +#if defined(__x86_64__) || defined(__i386__) + + xmm0 = rho01_128i[i]; // 4 symbols + xmm1 = rho01_128i[i+1]; + + // put (rho_r + rho_i)/2sqrt2 in rho_rpi + // put (rho_r - rho_i)/2sqrt2 in rho_rmi + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho) + xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho) + rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho) + rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho) + + // divide by sqrt(2) + rho_rpi = _mm_mulhi_epi16(rho_rpi, ONE_OVER_SQRT_2); + rho_rmi = _mm_mulhi_epi16(rho_rmi, ONE_OVER_SQRT_2); + rho_rpi = _mm_slli_epi16(rho_rpi,1); + rho_rmi = _mm_slli_epi16(rho_rmi,1); + + // Compute LLR for first bit of stream 0 + + // Compute real and imaginary parts of MF output for stream 0 + xmm0 = stream0_128i_in[i]; + xmm1 = stream0_128i_in[i+1]; + + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)] + y0i = _mm_unpackhi_epi64(xmm0,xmm1); + + // divide by sqrt(2) + y0r_over2 = _mm_mulhi_epi16(y0r, ONE_OVER_SQRT_2); + y0i_over2 = _mm_mulhi_epi16(y0i, ONE_OVER_SQRT_2); + y0r_over2 = _mm_slli_epi16(y0r,1); + y0i_over2 = _mm_slli_epi16(y0i,1); + + y0_p_1_1 = _mm_adds_epi16(y0r_over2, y0i_over2); + y0_m_1_1 = _mm_subs_epi16(y0r_over2, y0i_over2); + + // Compute real and imaginary parts of MF output for stream 1 + xmm0 = stream1_128i_in[i]; + xmm1 = stream1_128i_in[i+1]; + + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)] + y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)] + + xmm0 = _mm_setzero_si128(); // ZERO + + // compute psi + xmm3 = _mm_subs_epi16(y1r,rho_rpi); + psi_r_p1_p1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_subs_epi16(y1i,rho_rmi); + psi_i_p1_p1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_subs_epi16(y1r,rho_rmi); + psi_r_p1_m1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_adds_epi16(y1i,rho_rpi); + psi_i_p1_m1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_adds_epi16(y1r,rho_rmi); + psi_r_m1_p1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_subs_epi16(y1i,rho_rpi); + psi_i_m1_p1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_adds_epi16(y1r,rho_rpi); + psi_r_m1_m1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_adds_epi16(y1i,rho_rmi); + psi_i_m1_m1 = _mm_abs_epi16(xmm3); + + // Rearrange interfering channel magnitudes + xmm2 = ch_mag_128i_i[i]; + xmm3 = ch_mag_128i_i[i+1]; + + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + + ch_mag_int = _mm_unpacklo_epi64(xmm2,xmm3); + + // calculate optimal interference amplitudes + interference_abs_epi16(psi_r_p1_p1 , ch_mag_int, a_r_p1_p1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_p1 , ch_mag_int, a_i_p1_p1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_m1 , ch_mag_int, a_r_p1_m1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_m1 , ch_mag_int, a_i_p1_m1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_p1 , ch_mag_int, a_r_m1_p1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_p1 , ch_mag_int, a_i_m1_p1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_m1 , ch_mag_int, a_r_m1_m1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_m1 , ch_mag_int, a_i_m1_m1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + + // prodsum + prodsum_psi_a_epi16(psi_r_p1_p1, a_r_p1_p1, psi_i_p1_p1, a_i_p1_p1, psi_a_p1_p1); + prodsum_psi_a_epi16(psi_r_p1_m1, a_r_p1_m1, psi_i_p1_m1, a_i_p1_m1, psi_a_p1_m1); + prodsum_psi_a_epi16(psi_r_m1_p1, a_r_m1_p1, psi_i_m1_p1, a_i_m1_p1, psi_a_m1_p1); + prodsum_psi_a_epi16(psi_r_m1_m1, a_r_m1_m1, psi_i_m1_m1, a_i_m1_m1, psi_a_m1_m1); + + // squares + square_a_epi16(a_r_p1_p1, a_i_p1_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_p1); + square_a_epi16(a_r_p1_m1, a_i_p1_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_m1); + square_a_epi16(a_r_m1_p1, a_i_m1_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_p1); + square_a_epi16(a_r_m1_m1, a_i_m1_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_m1); + + // Computing Metrics + xmm0 = _mm_subs_epi16(psi_a_p1_p1, a_sq_p1_p1); + bit_met_p1_p1 = _mm_adds_epi16(xmm0, y0_p_1_1); + + xmm0 = _mm_subs_epi16(psi_a_p1_m1, a_sq_p1_m1); + bit_met_p1_m1 = _mm_adds_epi16(xmm0, y0_m_1_1); + + xmm0 = _mm_subs_epi16(psi_a_m1_p1, a_sq_m1_p1); + bit_met_m1_p1 = _mm_subs_epi16(xmm0, y0_m_1_1); + + xmm0 = _mm_subs_epi16(psi_a_m1_m1, a_sq_m1_m1); + bit_met_m1_m1 = _mm_subs_epi16(xmm0, y0_p_1_1); + + // MSB + logmax_num_re0 = _mm_max_epi16(bit_met_p1_p1,bit_met_p1_m1); // bit=0 + logmax_den_re0 = _mm_max_epi16(bit_met_m1_p1,bit_met_m1_m1); // bit=1 + + y0r = _mm_subs_epi16(logmax_num_re0,logmax_den_re0); + + // LSB + logmax_num_im0 = _mm_max_epi16(bit_met_p1_p1,bit_met_m1_p1); // bit=0 + logmax_den_im0 = _mm_max_epi16(bit_met_p1_m1,bit_met_m1_m1); // bit=1 + + y0i = _mm_subs_epi16(logmax_num_im0,logmax_den_im0); + + stream0_128i_out[i] = _mm_unpacklo_epi16(y0r,y0i); // = [L1(1), L2(1), L1(2), L2(2)] + + if (i<((length>>1) - 1)) // false if only 2 REs remain + stream0_128i_out[i+1] = _mm_unpackhi_epi16(y0r,y0i); + +#elif defined(__arm__) + +#endif + } + +#if defined(__x86_64__) || defined(__i386__) + _mm_empty(); + _m_empty(); +#endif +} + +int nr_dlsch_qpsk_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, + int32_t **rxdataF_comp, + int32_t **rxdataF_comp_i, + int32_t **dl_ch_mag_i, //|h_1|^2*(2/sqrt{10}) + int32_t **rho_i, + int16_t *dlsch_llr, + uint8_t symbol, + uint8_t first_symbol_flag, + uint16_t nb_rb, + uint16_t pbch_pss_sss_adjust, + int16_t **llr16p) +{ + + int16_t *rxF=(int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rxF_i=(int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag_i = (int16_t*)&dl_ch_mag_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rho=(int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *llr16; + int len; + uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol; + + + if (first_symbol_flag == 1) { + llr16 = (int16_t*)dlsch_llr; + } else { + llr16 = (int16_t*)(*llr16p); + } + + AssertFatal(llr16!=NULL,"nr_dlsch_qpsk_qam64_llr: llr is null, symbol %d\n",symbol); + + if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) { + // if symbol has pilots + if (frame_parms->nb_antenna_ports_eNB!=1) + // in 2 antenna ports we have 8 REs per symbol per RB + len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3); + else + // for 1 antenna port we have 10 REs per symbol per RB + len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6); + } else { + // symbol has no pilots + len = (nb_rb*12) - pbch_pss_sss_adjust; + } + + qpsk_qam64((short *)rxF, + (short *)rxF_i, + (short *)ch_mag_i, + (short *)llr16, + (short *)rho, + len); + + llr16 += (len<<1); + *llr16p = (short *)llr16; + + return(0); +} +/* +__m128i ONE_OVER_SQRT_2_42 __attribute__((aligned(16))); +__m128i THREE_OVER_SQRT_2_42 __attribute__((aligned(16))); +__m128i FIVE_OVER_SQRT_2_42 __attribute__((aligned(16))); +__m128i SEVEN_OVER_SQRT_2_42 __attribute__((aligned(16))); + +__m128i ch_mag_int_with_sigma2 __attribute__((aligned(16))); +__m128i two_ch_mag_int_with_sigma2 __attribute__((aligned(16))); +__m128i three_ch_mag_int_with_sigma2 __attribute__((aligned(16))); +__m128i SQRT_42_OVER_FOUR __attribute__((aligned(16))); +*/ +void qpsk_qam64(short *stream0_in, + short *stream1_in, + short *ch_mag_i, + short *stream0_out, + short *rho01, + int length + ) +{ + + /* + This function computes the LLRs of stream 0 (s_0) in presence of the interfering stream 1 (s_1) assuming that both symbols are QPSK. It can be used for both MU-MIMO interference-aware receiver or for SU-MIMO receivers. + + Parameters: + stream0_in = Matched filter output y0' = (h0*g0)*y0 + stream1_in = Matched filter output y1' = (h0*g1)*y0 + stream0_out = LLRs + rho01 = Correlation between the two effective channels \rho_{10} = (h1*g1)*(h0*g0) + length = number of resource elements + */ + +#if defined(__x86_64__) || defined(__i386__) + __m128i *rho01_128i = (__m128i *)rho01; + __m128i *stream0_128i_in = (__m128i *)stream0_in; + __m128i *stream1_128i_in = (__m128i *)stream1_in; + __m128i *stream0_128i_out = (__m128i *)stream0_out; + __m128i *ch_mag_128i_i = (__m128i *)ch_mag_i; + __m128i ONE_OVER_SQRT_2 = _mm_set1_epi16(23170); // round(1/sqrt(2)*2^15) + __m128i ONE_OVER_SQRT_2_42 = _mm_set1_epi16(3575); // round(1/sqrt(2*42)*2^15) + __m128i THREE_OVER_SQRT_2_42 = _mm_set1_epi16(10726); // round(3/sqrt(2*42)*2^15) + __m128i FIVE_OVER_SQRT_2_42 = _mm_set1_epi16(17876); // round(5/sqrt(2*42)*2^15) + __m128i SEVEN_OVER_SQRT_2_42 = _mm_set1_epi16(25027); // round(7/sqrt(2*42)*2^15) + __m128i SQRT_42_OVER_FOUR = _mm_set1_epi16(13272); // round(sqrt(42)/4*2^13), Q3.1 + __m128i ch_mag_int; + __m128i ch_mag_int_with_sigma2; + __m128i two_ch_mag_int_with_sigma2; + __m128i three_ch_mag_int_with_sigma2; +#elif defined(__arm__) + +#endif + +#ifdef DEBUG_LLR + print_shorts2("rho01_128i:\n",rho01_128i); +#endif + + int i; + + + for (i=0; i>2; i+=2) { + // in each iteration, we take 8 complex samples + +#if defined(__x86_64__) || defined(__i386__) + + xmm0 = rho01_128i[i]; // 4 symbols + xmm1 = rho01_128i[i+1]; + + // put (rho_r + rho_i)/sqrt2 in rho_rpi + // put (rho_r - rho_i)/sqrt2 in rho_rmi + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho) + xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho) + rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho) + rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho) + + // divide by sqrt(2) + rho_rpi = _mm_mulhi_epi16(rho_rpi, ONE_OVER_SQRT_2); + rho_rmi = _mm_mulhi_epi16(rho_rmi, ONE_OVER_SQRT_2); + rho_rpi = _mm_slli_epi16(rho_rpi,1); + rho_rmi = _mm_slli_epi16(rho_rmi,1); + + // Compute LLR for first bit of stream 0 + + // Compute real and imaginary parts of MF output for stream 0 + xmm0 = stream0_128i_in[i]; + xmm1 = stream0_128i_in[i+1]; + + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)] + y0i = _mm_unpackhi_epi64(xmm0,xmm1); + + // divide by sqrt(2) + y0r_over2 = _mm_mulhi_epi16(y0r, ONE_OVER_SQRT_2); + y0i_over2 = _mm_mulhi_epi16(y0i, ONE_OVER_SQRT_2); + y0r_over2 = _mm_slli_epi16(y0r,1); + y0i_over2 = _mm_slli_epi16(y0i,1); + + y0_p_1_1 = _mm_adds_epi16(y0r_over2, y0i_over2); + y0_m_1_1 = _mm_subs_epi16(y0r_over2, y0i_over2); + + // Compute real and imaginary parts of MF output for stream 1 + xmm0 = stream1_128i_in[i]; + xmm1 = stream1_128i_in[i+1]; + + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)] + y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)] + + xmm0 = _mm_setzero_si128(); // ZERO + + // compute psi + xmm3 = _mm_subs_epi16(y1r,rho_rpi); + psi_r_p1_p1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_subs_epi16(y1i,rho_rmi); + psi_i_p1_p1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_subs_epi16(y1r,rho_rmi); + psi_r_p1_m1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_adds_epi16(y1i,rho_rpi); + psi_i_p1_m1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_adds_epi16(y1r,rho_rmi); + psi_r_m1_p1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_subs_epi16(y1i,rho_rpi); + psi_i_m1_p1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_adds_epi16(y1r,rho_rpi); + psi_r_m1_m1 = _mm_abs_epi16(xmm3); + xmm3 = _mm_adds_epi16(y1i,rho_rmi); + psi_i_m1_m1 = _mm_abs_epi16(xmm3); + + // Rearrange interfering channel magnitudes + xmm2 = ch_mag_128i_i[i]; + xmm3 = ch_mag_128i_i[i+1]; + + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + + ch_mag_int = _mm_unpacklo_epi64(xmm2,xmm3); + ch_mag_int_with_sigma2 = _mm_srai_epi16(ch_mag_int, 1); // *2 + two_ch_mag_int_with_sigma2 = ch_mag_int; // *4 + three_ch_mag_int_with_sigma2 = _mm_adds_epi16(ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2); // *6 + + interference_abs_64qam_epi16(psi_r_p1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + + // prodsum + prodsum_psi_a_epi16(psi_r_p1_p1, a_r_p1_p1, psi_i_p1_p1, a_i_p1_p1, psi_a_p1_p1); + prodsum_psi_a_epi16(psi_r_p1_m1, a_r_p1_m1, psi_i_p1_m1, a_i_p1_m1, psi_a_p1_m1); + prodsum_psi_a_epi16(psi_r_m1_p1, a_r_m1_p1, psi_i_m1_p1, a_i_m1_p1, psi_a_m1_p1); + prodsum_psi_a_epi16(psi_r_m1_m1, a_r_m1_m1, psi_i_m1_m1, a_i_m1_m1, psi_a_m1_m1); + + // Multiply by sqrt(2) + psi_a_p1_p1 = _mm_mulhi_epi16(psi_a_p1_p1, ONE_OVER_SQRT_2); + psi_a_p1_p1 = _mm_slli_epi16(psi_a_p1_p1, 2); + psi_a_p1_m1 = _mm_mulhi_epi16(psi_a_p1_m1, ONE_OVER_SQRT_2); + psi_a_p1_m1 = _mm_slli_epi16(psi_a_p1_m1, 2); + psi_a_m1_p1 = _mm_mulhi_epi16(psi_a_m1_p1, ONE_OVER_SQRT_2); + psi_a_m1_p1 = _mm_slli_epi16(psi_a_m1_p1, 2); + psi_a_m1_m1 = _mm_mulhi_epi16(psi_a_m1_m1, ONE_OVER_SQRT_2); + psi_a_m1_m1 = _mm_slli_epi16(psi_a_m1_m1, 2); + + square_a_64qam_epi16(a_r_p1_p1, a_i_p1_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_p1); + square_a_64qam_epi16(a_r_p1_m1, a_i_p1_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_m1); + square_a_64qam_epi16(a_r_m1_p1, a_i_m1_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_p1); + square_a_64qam_epi16(a_r_m1_m1, a_i_m1_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_m1); + + // Computing Metrics + xmm0 = _mm_subs_epi16(psi_a_p1_p1, a_sq_p1_p1); + bit_met_p1_p1 = _mm_adds_epi16(xmm0, y0_p_1_1); + + xmm0 = _mm_subs_epi16(psi_a_p1_m1, a_sq_p1_m1); + bit_met_p1_m1 = _mm_adds_epi16(xmm0, y0_m_1_1); + + xmm0 = _mm_subs_epi16(psi_a_m1_p1, a_sq_m1_p1); + bit_met_m1_p1 = _mm_subs_epi16(xmm0, y0_m_1_1); + + xmm0 = _mm_subs_epi16(psi_a_m1_m1, a_sq_m1_m1); + bit_met_m1_m1 = _mm_subs_epi16(xmm0, y0_p_1_1); + + // MSB + logmax_num_re0 = _mm_max_epi16(bit_met_p1_p1,bit_met_p1_m1); // bit=0 + logmax_den_re0 = _mm_max_epi16(bit_met_m1_p1,bit_met_m1_m1); // bit=1 + + y0r = _mm_subs_epi16(logmax_num_re0,logmax_den_re0); + + // LSB + logmax_num_im0 = _mm_max_epi16(bit_met_p1_p1,bit_met_m1_p1); // bit=0 + logmax_den_im0 = _mm_max_epi16(bit_met_p1_m1,bit_met_m1_m1); // bit=1 + + y0i = _mm_subs_epi16(logmax_num_im0,logmax_den_im0); + + stream0_128i_out[i] = _mm_unpacklo_epi16(y0r,y0i); // = [L1(1), L2(1), L1(2), L2(2)] + + if (i<((length>>1) - 1)) // false if only 2 REs remain + stream0_128i_out[i+1] = _mm_unpackhi_epi16(y0r,y0i); + +#elif defined(__arm__) + +#endif + } + +#if defined(__x86_64__) || defined(__i386__) + _mm_empty(); + _m_empty(); +#endif +} + + +//---------------------------------------------------------------------------------------------- +// 16-QAM +//---------------------------------------------------------------------------------------------- + +/* +__m128i ONE_OVER_TWO_SQRT_10 __attribute__((aligned(16))); +__m128i NINE_OVER_TWO_SQRT_10 __attribute__((aligned(16))); + +__m128i y0r_over_sqrt10 __attribute__ ((aligned(16))); +__m128i y0i_over_sqrt10 __attribute__ ((aligned(16))); +__m128i y0r_three_over_sqrt10 __attribute__ ((aligned(16))); +__m128i y0i_three_over_sqrt10 __attribute__ ((aligned(16))); + +__m128i ch_mag_des __attribute__((aligned(16))); +__m128i ch_mag_over_10 __attribute__ ((aligned(16))); +__m128i ch_mag_over_2 __attribute__ ((aligned(16))); +__m128i ch_mag_9_over_10 __attribute__ ((aligned(16))); +*/ + +void qam16_qpsk(short *stream0_in, + short *stream1_in, + short *ch_mag, + short *stream0_out, + short *rho01, + int length + ) +{ + + /* + Author: Sebastian Wagner + Date: 2012-06-04 + + Input: + stream0_in: MF filter for 1st stream, i.e., y0=h0'*y + stream!_in: MF filter for 2nd stream, i.e., y1=h1'*y + ch_mag: 2*h0/sqrt(00), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + ch_mag_i: 2*h1/sqrt(00), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + rho01: Channel cross correlation, i.e., h1'*h0 + + Output: + stream0_out: output LLRs for 1st stream + */ + +#if defined(__x86_64__) || defined(__i386__) + __m128i *rho01_128i = (__m128i *)rho01; + __m128i *stream0_128i_in = (__m128i *)stream0_in; + __m128i *stream1_128i_in = (__m128i *)stream1_in; + __m128i *stream0_128i_out = (__m128i *)stream0_out; + __m128i *ch_mag_128i = (__m128i *)ch_mag; + __m128i ONE_OVER_SQRT_2 = _mm_set1_epi16(23170); // round(1/sqrt(2)*2^15) + __m128i ONE_OVER_SQRT_10 = _mm_set1_epi16(20724); // round(1/sqrt(10)*2^16) + __m128i THREE_OVER_SQRT_10 = _mm_set1_epi16(31086); // round(3/sqrt(10)*2^15) + __m128i SQRT_10_OVER_FOUR = _mm_set1_epi16(25905); // round(sqrt(10)/4*2^15) + __m128i ONE_OVER_TWO_SQRT_10 = _mm_set1_epi16(10362); // round(1/2/sqrt(10)*2^16) + __m128i NINE_OVER_TWO_SQRT_10 = _mm_set1_epi16(23315); // round(9/2/sqrt(10)*2^14) + __m128i y0r_over_sqrt10; + __m128i y0i_over_sqrt10; + __m128i y0r_three_over_sqrt10; + __m128i y0i_three_over_sqrt10; + + __m128i ch_mag_des; + __m128i ch_mag_over_10; + __m128i ch_mag_over_2; + __m128i ch_mag_9_over_10; +#elif defined(__arm__) + +#endif + + int i; + + + for (i=0; i>2; i+=2) { + // In one iteration, we deal with 8 REs + +#if defined(__x86_64__) || defined(__i386__) + // Get rho + xmm0 = rho01_128i[i]; + xmm1 = rho01_128i[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho) + xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho) + rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho) + rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho) + + // Compute the different rhos + rho_rpi_1_1 = _mm_mulhi_epi16(rho_rpi,ONE_OVER_SQRT_10); + rho_rmi_1_1 = _mm_mulhi_epi16(rho_rmi,ONE_OVER_SQRT_10); + rho_rpi_3_3 = _mm_mulhi_epi16(rho_rpi,THREE_OVER_SQRT_10); + rho_rmi_3_3 = _mm_mulhi_epi16(rho_rmi,THREE_OVER_SQRT_10); + rho_rpi_3_3 = _mm_slli_epi16(rho_rpi_3_3,1); + rho_rmi_3_3 = _mm_slli_epi16(rho_rmi_3_3,1); + + xmm4 = _mm_mulhi_epi16(xmm2,ONE_OVER_SQRT_10); // Re(rho) + xmm5 = _mm_mulhi_epi16(xmm3,THREE_OVER_SQRT_10); // Im(rho) + xmm5 = _mm_slli_epi16(xmm5,1); + + rho_rpi_1_3 = _mm_adds_epi16(xmm4,xmm5); + rho_rmi_1_3 = _mm_subs_epi16(xmm4,xmm5); + + xmm6 = _mm_mulhi_epi16(xmm2,THREE_OVER_SQRT_10); // Re(rho) + xmm7 = _mm_mulhi_epi16(xmm3,ONE_OVER_SQRT_10); // Im(rho) + xmm6 = _mm_slli_epi16(xmm6,1); + + rho_rpi_3_1 = _mm_adds_epi16(xmm6,xmm7); + rho_rmi_3_1 = _mm_subs_epi16(xmm6,xmm7); + + // Rearrange interfering MF output + xmm0 = stream1_128i_in[i]; + xmm1 = stream1_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)] + y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)] + + xmm0 = _mm_setzero_si128(); // ZERO + xmm2 = _mm_subs_epi16(rho_rpi_1_1,y1r); // = [Re(rho)+ Im(rho)]/sqrt(10) - y1r + psi_r_p1_p1 = _mm_abs_epi16(xmm2); // = |[Re(rho)+ Im(rho)]/sqrt(10) - y1r| + + xmm2= _mm_subs_epi16(rho_rmi_1_1,y1r); + psi_r_p1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_1_1,y1i); + psi_i_p1_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_1_3,y1r); + psi_r_p1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_1_3,y1r); + psi_r_p1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_1,y1i); + psi_i_p1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_1,y1r); + psi_r_p3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_1,y1r); + psi_r_p3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_1_3,y1i); + psi_i_p3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_3,y1r); + psi_r_p3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_3,y1r); + psi_r_p3_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_3,y1i); + psi_i_p3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_1_1,y1i); + psi_i_m1_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_1,y1i); + psi_i_m1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_1_3,y1i); + psi_i_m3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_3,y1i); + psi_i_m3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_1,y1i); + psi_i_p1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_1,y1i); + psi_i_p1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_3,y1i); + psi_i_p3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_3,y1i); + psi_i_p3_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_1,y1r); + psi_r_m1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_3,y1r); + psi_r_m1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_1,y1r); + psi_r_m3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_3,y1r); + psi_r_m3_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_1_1); + psi_r_m1_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_1_3); + psi_r_m1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_1_1); + psi_i_m1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_3_1); + psi_i_m1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_3_1); + psi_r_m3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_3_3); + psi_r_m3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_1_3); + psi_i_m3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_3_3); + psi_i_m3_m3 = _mm_abs_epi16(xmm2); + + // Rearrange desired MF output + xmm0 = stream0_128i_in[i]; + xmm1 = stream0_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)] + y0i = _mm_unpackhi_epi64(xmm0,xmm1); + + // Rearrange desired channel magnitudes + xmm2 = ch_mag_128i[i]; // = [|h|^2(1),|h|^2(1),|h|^2(2),|h|^2(2)]*(2/sqrt(10)) + xmm3 = ch_mag_128i[i+1]; // = [|h|^2(3),|h|^2(3),|h|^2(4),|h|^2(4)]*(2/sqrt(10)) + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + + ch_mag_des = _mm_unpacklo_epi64(xmm2,xmm3); // = [|h|^2(1),|h|^2(2),|h|^2(3),|h|^2(4)]*(2/sqrt(10)) + + // Scale MF output of desired signal + y0r_over_sqrt10 = _mm_mulhi_epi16(y0r,ONE_OVER_SQRT_10); + y0i_over_sqrt10 = _mm_mulhi_epi16(y0i,ONE_OVER_SQRT_10); + y0r_three_over_sqrt10 = _mm_mulhi_epi16(y0r,THREE_OVER_SQRT_10); + y0i_three_over_sqrt10 = _mm_mulhi_epi16(y0i,THREE_OVER_SQRT_10); + y0r_three_over_sqrt10 = _mm_slli_epi16(y0r_three_over_sqrt10,1); + y0i_three_over_sqrt10 = _mm_slli_epi16(y0i_three_over_sqrt10,1); + + // Compute necessary combination of required terms + y0_p_1_1 = _mm_adds_epi16(y0r_over_sqrt10,y0i_over_sqrt10); + y0_m_1_1 = _mm_subs_epi16(y0r_over_sqrt10,y0i_over_sqrt10); + + y0_p_1_3 = _mm_adds_epi16(y0r_over_sqrt10,y0i_three_over_sqrt10); + y0_m_1_3 = _mm_subs_epi16(y0r_over_sqrt10,y0i_three_over_sqrt10); + + y0_p_3_1 = _mm_adds_epi16(y0r_three_over_sqrt10,y0i_over_sqrt10); + y0_m_3_1 = _mm_subs_epi16(y0r_three_over_sqrt10,y0i_over_sqrt10); + + y0_p_3_3 = _mm_adds_epi16(y0r_three_over_sqrt10,y0i_three_over_sqrt10); + y0_m_3_3 = _mm_subs_epi16(y0r_three_over_sqrt10,y0i_three_over_sqrt10); + + // Add psi + psi_a_p1_p1 = _mm_adds_epi16(psi_r_p1_p1 ,psi_i_p1_p1); + psi_a_p1_p3 = _mm_adds_epi16(psi_r_p1_p3 ,psi_i_p1_p3); + psi_a_p3_p1 = _mm_adds_epi16(psi_r_p3_p1 ,psi_i_p3_p1); + psi_a_p3_p3 = _mm_adds_epi16(psi_r_p3_p3 ,psi_i_p3_p3); + psi_a_p1_m1 = _mm_adds_epi16(psi_r_p1_m1 ,psi_i_p1_m1); + psi_a_p1_m3 = _mm_adds_epi16(psi_r_p1_m3 ,psi_i_p1_m3); + psi_a_p3_m1 = _mm_adds_epi16(psi_r_p3_m1 ,psi_i_p3_m1); + psi_a_p3_m3 = _mm_adds_epi16(psi_r_p3_m3 ,psi_i_p3_m3); + psi_a_m1_p1 = _mm_adds_epi16(psi_r_m1_p1 ,psi_i_m1_p1); + psi_a_m1_p3 = _mm_adds_epi16(psi_r_m1_p3 ,psi_i_m1_p3); + psi_a_m3_p1 = _mm_adds_epi16(psi_r_m3_p1 ,psi_i_m3_p1); + psi_a_m3_p3 = _mm_adds_epi16(psi_r_m3_p3 ,psi_i_m3_p3); + psi_a_m1_m1 = _mm_adds_epi16(psi_r_m1_m1 ,psi_i_m1_m1); + psi_a_m1_m3 = _mm_adds_epi16(psi_r_m1_m3 ,psi_i_m1_m3); + psi_a_m3_m1 = _mm_adds_epi16(psi_r_m3_m1 ,psi_i_m3_m1); + psi_a_m3_m3 = _mm_adds_epi16(psi_r_m3_m3 ,psi_i_m3_m3); + + // scale by sqrt(2) + psi_a_p1_p1 = _mm_mulhi_epi16(psi_a_p1_p1,ONE_OVER_SQRT_2); + psi_a_p1_p1 = _mm_slli_epi16(psi_a_p1_p1,1); + psi_a_p1_p3 = _mm_mulhi_epi16(psi_a_p1_p3,ONE_OVER_SQRT_2); + psi_a_p1_p3 = _mm_slli_epi16(psi_a_p1_p3,1); + psi_a_p3_p1 = _mm_mulhi_epi16(psi_a_p3_p1,ONE_OVER_SQRT_2); + psi_a_p3_p1 = _mm_slli_epi16(psi_a_p3_p1,1); + psi_a_p3_p3 = _mm_mulhi_epi16(psi_a_p3_p3,ONE_OVER_SQRT_2); + psi_a_p3_p3 = _mm_slli_epi16(psi_a_p3_p3,1); + + psi_a_p1_m1 = _mm_mulhi_epi16(psi_a_p1_m1,ONE_OVER_SQRT_2); + psi_a_p1_m1 = _mm_slli_epi16(psi_a_p1_m1,1); + psi_a_p1_m3 = _mm_mulhi_epi16(psi_a_p1_m3,ONE_OVER_SQRT_2); + psi_a_p1_m3 = _mm_slli_epi16(psi_a_p1_m3,1); + psi_a_p3_m1 = _mm_mulhi_epi16(psi_a_p3_m1,ONE_OVER_SQRT_2); + psi_a_p3_m1 = _mm_slli_epi16(psi_a_p3_m1,1); + psi_a_p3_m3 = _mm_mulhi_epi16(psi_a_p3_m3,ONE_OVER_SQRT_2); + psi_a_p3_m3 = _mm_slli_epi16(psi_a_p3_m3,1); + + psi_a_m1_p1 = _mm_mulhi_epi16(psi_a_m1_p1,ONE_OVER_SQRT_2); + psi_a_m1_p1 = _mm_slli_epi16(psi_a_m1_p1,1); + psi_a_m1_p3 = _mm_mulhi_epi16(psi_a_m1_p3,ONE_OVER_SQRT_2); + psi_a_m1_p3 = _mm_slli_epi16(psi_a_m1_p3,1); + psi_a_m3_p1 = _mm_mulhi_epi16(psi_a_m3_p1,ONE_OVER_SQRT_2); + psi_a_m3_p1 = _mm_slli_epi16(psi_a_m3_p1,1); + psi_a_m3_p3 = _mm_mulhi_epi16(psi_a_m3_p3,ONE_OVER_SQRT_2); + psi_a_m3_p3 = _mm_slli_epi16(psi_a_m3_p3,1); + + psi_a_m1_m1 = _mm_mulhi_epi16(psi_a_m1_m1,ONE_OVER_SQRT_2); + psi_a_m1_m1 = _mm_slli_epi16(psi_a_m1_m1,1); + psi_a_m1_m3 = _mm_mulhi_epi16(psi_a_m1_m3,ONE_OVER_SQRT_2); + psi_a_m1_m3 = _mm_slli_epi16(psi_a_m1_m3,1); + psi_a_m3_m1 = _mm_mulhi_epi16(psi_a_m3_m1,ONE_OVER_SQRT_2); + psi_a_m3_m1 = _mm_slli_epi16(psi_a_m3_m1,1); + psi_a_m3_m3 = _mm_mulhi_epi16(psi_a_m3_m3,ONE_OVER_SQRT_2); + psi_a_m3_m3 = _mm_slli_epi16(psi_a_m3_m3,1); + + // Computing different multiples of channel norms + ch_mag_over_10=_mm_mulhi_epi16(ch_mag_des, ONE_OVER_TWO_SQRT_10); + ch_mag_over_2=_mm_mulhi_epi16(ch_mag_des, SQRT_10_OVER_FOUR); + ch_mag_over_2=_mm_slli_epi16(ch_mag_over_2, 1); + ch_mag_9_over_10=_mm_mulhi_epi16(ch_mag_des, NINE_OVER_TWO_SQRT_10); + ch_mag_9_over_10=_mm_slli_epi16(ch_mag_9_over_10, 2); + + // Computing Metrics + xmm1 = _mm_adds_epi16(psi_a_p1_p1, y0_p_1_1); + bit_met_p1_p1= _mm_subs_epi16(xmm1, ch_mag_over_10); + + xmm1 = _mm_adds_epi16(psi_a_p1_p3, y0_p_1_3); + bit_met_p1_p3= _mm_subs_epi16(xmm1, ch_mag_over_2); + + xmm1 = _mm_adds_epi16(psi_a_p1_m1, y0_m_1_1); + bit_met_p1_m1= _mm_subs_epi16(xmm1, ch_mag_over_10); + + xmm1 = _mm_adds_epi16(psi_a_p1_m3, y0_m_1_3); + bit_met_p1_m3= _mm_subs_epi16(xmm1, ch_mag_over_2); + + xmm1 = _mm_adds_epi16(psi_a_p3_p1, y0_p_3_1); + bit_met_p3_p1= _mm_subs_epi16(xmm1, ch_mag_over_2); + + xmm1 = _mm_adds_epi16(psi_a_p3_p3, y0_p_3_3); + bit_met_p3_p3= _mm_subs_epi16(xmm1, ch_mag_9_over_10); + + xmm1 = _mm_adds_epi16(psi_a_p3_m1, y0_m_3_1); + bit_met_p3_m1= _mm_subs_epi16(xmm1, ch_mag_over_2); + + xmm1 = _mm_adds_epi16(psi_a_p3_m3, y0_m_3_3); + bit_met_p3_m3= _mm_subs_epi16(xmm1, ch_mag_9_over_10); + + xmm1 = _mm_subs_epi16(psi_a_m1_p1, y0_m_1_1); + bit_met_m1_p1= _mm_subs_epi16(xmm1, ch_mag_over_10); + + xmm1 = _mm_subs_epi16(psi_a_m1_p3, y0_m_1_3); + bit_met_m1_p3= _mm_subs_epi16(xmm1, ch_mag_over_2); + + xmm1 = _mm_subs_epi16(psi_a_m1_m1, y0_p_1_1); + bit_met_m1_m1= _mm_subs_epi16(xmm1, ch_mag_over_10); + + xmm1 = _mm_subs_epi16(psi_a_m1_m3, y0_p_1_3); + bit_met_m1_m3= _mm_subs_epi16(xmm1, ch_mag_over_2); + + xmm1 = _mm_subs_epi16(psi_a_m3_p1, y0_m_3_1); + bit_met_m3_p1= _mm_subs_epi16(xmm1, ch_mag_over_2); + + xmm1 = _mm_subs_epi16(psi_a_m3_p3, y0_m_3_3); + bit_met_m3_p3= _mm_subs_epi16(xmm1, ch_mag_9_over_10); + + xmm1 = _mm_subs_epi16(psi_a_m3_m1, y0_p_3_1); + bit_met_m3_m1= _mm_subs_epi16(xmm1, ch_mag_over_2); + + xmm1 = _mm_subs_epi16(psi_a_m3_m3, y0_p_3_3); + bit_met_m3_m3= _mm_subs_epi16(xmm1, ch_mag_9_over_10); + + // LLR of the first bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_m1_p1,bit_met_m1_p3); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m1_m3); + xmm2 = _mm_max_epi16(bit_met_m3_p1,bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m3_m1,bit_met_m3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_re0= _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_p1_p1,bit_met_p1_p3); + xmm1 = _mm_max_epi16(bit_met_p1_m1,bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_p3_p1,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_m1,bit_met_p3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4,xmm5); + + // LLR of first bit [L1(1), L1(2), L1(3), L1(4), L1(5), L1(6), L1(7), L1(8)] + y0r = _mm_subs_epi16(logmax_den_re0,logmax_num_re0); + + // LLR of the second bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_p1_m1,bit_met_p3_m1); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_p1_m3,bit_met_p3_m3); + xmm3 = _mm_max_epi16(bit_met_m1_m3,bit_met_m3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_re1 = _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_p1_p1,bit_met_p3_p1); + xmm1 = _mm_max_epi16(bit_met_m1_p1,bit_met_m3_p1); + xmm2 = _mm_max_epi16(bit_met_p1_p3,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p3,bit_met_m3_p3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_re1 = _mm_max_epi16(xmm4,xmm5); + + // LLR of second bit [L2(1), L2(2), L2(3), L2(4)] + y1r = _mm_subs_epi16(logmax_den_re1,logmax_num_re1); + + // LLR of the third bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_m3_p1,bit_met_m3_p3); + xmm1 = _mm_max_epi16(bit_met_m3_m1,bit_met_m3_m3); + xmm2 = _mm_max_epi16(bit_met_p3_p1,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_m1,bit_met_p3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_im0 = _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_m1_p1,bit_met_m1_p3); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m1_m3); + xmm2 = _mm_max_epi16(bit_met_p1_p1,bit_met_p1_p3); + xmm3 = _mm_max_epi16(bit_met_p1_m1,bit_met_p1_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_im0 = _mm_max_epi16(xmm4,xmm5); + + // LLR of third bit [L3(1), L3(2), L3(3), L3(4)] + y0i = _mm_subs_epi16(logmax_den_im0,logmax_num_im0); + + // LLR of the fourth bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_p1_m3,bit_met_p3_m3); + xmm1 = _mm_max_epi16(bit_met_m1_m3,bit_met_m3_m3); + xmm2 = _mm_max_epi16(bit_met_p1_p3,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p3,bit_met_m3_p3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_im1 = _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_p1_m1,bit_met_p3_m1); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_p1_p1,bit_met_p3_p1); + xmm3 = _mm_max_epi16(bit_met_m1_p1,bit_met_m3_p1); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_im1 = _mm_max_epi16(xmm4,xmm5); + + // LLR of fourth bit [L4(1), L4(2), L4(3), L4(4)] + y1i = _mm_subs_epi16(logmax_den_im1,logmax_num_im1); + + // Pack LLRs in output + // [L1(1), L2(1), L1(2), L2(2), L1(3), L2(3), L1(4), L2(4)] + xmm0 = _mm_unpacklo_epi16(y0r,y1r); + // [L1(5), L2(5), L1(6), L2(6), L1(7), L2(7), L1(8), L2(8)] + xmm1 = _mm_unpackhi_epi16(y0r,y1r); + // [L3(1), L4(1), L3(2), L4(2), L3(3), L4(3), L3(4), L4(4)] + xmm2 = _mm_unpacklo_epi16(y0i,y1i); + // [L3(5), L4(5), L3(6), L4(6), L3(7), L4(7), L3(8), L4(8)] + xmm3 = _mm_unpackhi_epi16(y0i,y1i); + + stream0_128i_out[2*i+0] = _mm_unpacklo_epi32(xmm0,xmm2); // 8LLRs, 2REs + stream0_128i_out[2*i+1] = _mm_unpackhi_epi32(xmm0,xmm2); + stream0_128i_out[2*i+2] = _mm_unpacklo_epi32(xmm1,xmm3); + stream0_128i_out[2*i+3] = _mm_unpackhi_epi32(xmm1,xmm3); + +#elif defined(__arm__) + +#endif + } + +#if defined(__x86_64__) || defined(__i386__) + _mm_empty(); + _m_empty(); +#endif + +} + +int nr_dlsch_16qam_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, + int32_t **rxdataF_comp, + int32_t **rxdataF_comp_i, + int32_t **dl_ch_mag, //|h_0|^2*(2/sqrt{10}) + int32_t **rho_i, + int16_t *dlsch_llr, + uint8_t symbol, + uint8_t first_symbol_flag, + uint16_t nb_rb, + uint16_t pbch_pss_sss_adjust, + int16_t **llr16p) +{ + + int16_t *rxF = (int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rxF_i = (int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag = (int16_t*)&dl_ch_mag[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rho = (int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *llr16; + int len; + uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol; + + // first symbol has different structure due to more pilots + if (first_symbol_flag == 1) { + llr16 = (int16_t*)dlsch_llr; + } else { + llr16 = (int16_t*)(*llr16p); + } + + AssertFatal(llr16!=NULL,"nr_dlsch_16qam_qpsk_llr: llr is null, symbol %d\n",symbol); + + if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) { + // if symbol has pilots + if (frame_parms->nb_antenna_ports_eNB!=1) + // in 2 antenna ports we have 8 REs per symbol per RB + len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3); + else + // for 1 antenna port we have 10 REs per symbol per RB + len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6); + } else { + // symbol has no pilots + len = (nb_rb*12) - pbch_pss_sss_adjust; + } + + // printf("symbol %d: qam16_llr, len %d (llr16 %p)\n",symbol,len,llr16); + + qam16_qpsk((short *)rxF, + (short *)rxF_i, + (short *)ch_mag, + (short *)llr16, + (short *)rho, + len); + + llr16 += (len<<2); + *llr16p = (short *)llr16; + + return(0); +} + +void qam16_qam16(short *stream0_in, + short *stream1_in, + short *ch_mag, + short *ch_mag_i, + short *stream0_out, + short *rho01, + int length + ) +{ + + /* + Author: Sebastian Wagner + Date: 2012-06-04 + + Input: + stream0_in: MF filter for 1st stream, i.e., y0=h0'*y + stream!_in: MF filter for 2nd stream, i.e., y1=h1'*y + ch_mag: 2*h0/sqrt(00), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + ch_mag_i: 2*h1/sqrt(00), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + rho01: Channel cross correlation, i.e., h1'*h0 + + Output: + stream0_out: output LLRs for 1st stream + */ +#if defined(__x86_64__) || defined(__i386__) + __m128i *rho01_128i = (__m128i *)rho01; + __m128i *stream0_128i_in = (__m128i *)stream0_in; + __m128i *stream1_128i_in = (__m128i *)stream1_in; + __m128i *stream0_128i_out = (__m128i *)stream0_out; + __m128i *ch_mag_128i = (__m128i *)ch_mag; + __m128i *ch_mag_128i_i = (__m128i *)ch_mag_i; + + + + __m128i ONE_OVER_SQRT_10 = _mm_set1_epi16(20724); // round(1/sqrt(10)*2^16) + __m128i ONE_OVER_SQRT_10_Q15 = _mm_set1_epi16(10362); // round(1/sqrt(10)*2^15) + __m128i THREE_OVER_SQRT_10 = _mm_set1_epi16(31086); // round(3/sqrt(10)*2^15) + __m128i SQRT_10_OVER_FOUR = _mm_set1_epi16(25905); // round(sqrt(10)/4*2^15) + __m128i ONE_OVER_TWO_SQRT_10 = _mm_set1_epi16(10362); // round(1/2/sqrt(10)*2^16) + __m128i NINE_OVER_TWO_SQRT_10 = _mm_set1_epi16(23315); // round(9/2/sqrt(10)*2^14) + __m128i ch_mag_des,ch_mag_int; + __m128i y0r_over_sqrt10; + __m128i y0i_over_sqrt10; + __m128i y0r_three_over_sqrt10; + __m128i y0i_three_over_sqrt10; + __m128i ch_mag_over_10; + __m128i ch_mag_over_2; + __m128i ch_mag_9_over_10; +#elif defined(__arm__) + +#endif + + int i; + + for (i=0; i>2; i+=2) { + // In one iteration, we deal with 8 REs + +#if defined(__x86_64__) || defined(__i386__) + // Get rho + xmm0 = rho01_128i[i]; + xmm1 = rho01_128i[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho) + xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho) + rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho) + rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho) + + // Compute the different rhos + rho_rpi_1_1 = _mm_mulhi_epi16(rho_rpi,ONE_OVER_SQRT_10); + rho_rmi_1_1 = _mm_mulhi_epi16(rho_rmi,ONE_OVER_SQRT_10); + rho_rpi_3_3 = _mm_mulhi_epi16(rho_rpi,THREE_OVER_SQRT_10); + rho_rmi_3_3 = _mm_mulhi_epi16(rho_rmi,THREE_OVER_SQRT_10); + rho_rpi_3_3 = _mm_slli_epi16(rho_rpi_3_3,1); + rho_rmi_3_3 = _mm_slli_epi16(rho_rmi_3_3,1); + + xmm4 = _mm_mulhi_epi16(xmm2,ONE_OVER_SQRT_10); // Re(rho) + xmm5 = _mm_mulhi_epi16(xmm3,THREE_OVER_SQRT_10); // Im(rho) + xmm5 = _mm_slli_epi16(xmm5,1); + + rho_rpi_1_3 = _mm_adds_epi16(xmm4,xmm5); + rho_rmi_1_3 = _mm_subs_epi16(xmm4,xmm5); + + xmm6 = _mm_mulhi_epi16(xmm2,THREE_OVER_SQRT_10); // Re(rho) + xmm7 = _mm_mulhi_epi16(xmm3,ONE_OVER_SQRT_10); // Im(rho) + xmm6 = _mm_slli_epi16(xmm6,1); + + rho_rpi_3_1 = _mm_adds_epi16(xmm6,xmm7); + rho_rmi_3_1 = _mm_subs_epi16(xmm6,xmm7); + + // Rearrange interfering MF output + xmm0 = stream1_128i_in[i]; + xmm1 = stream1_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)] + y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)] + + xmm0 = _mm_setzero_si128(); // ZERO + xmm2 = _mm_subs_epi16(rho_rpi_1_1,y1r); // = [Re(rho)+ Im(rho)]/sqrt(10) - y1r + psi_r_p1_p1 = _mm_abs_epi16(xmm2); // = |[Re(rho)+ Im(rho)]/sqrt(10) - y1r| + + xmm2= _mm_subs_epi16(rho_rmi_1_1,y1r); + psi_r_p1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_1_1,y1i); + psi_i_p1_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_1_3,y1r); + psi_r_p1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_1_3,y1r); + psi_r_p1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_1,y1i); + psi_i_p1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_1,y1r); + psi_r_p3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_1,y1r); + psi_r_p3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_1_3,y1i); + psi_i_p3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_3,y1r); + psi_r_p3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_3,y1r); + psi_r_p3_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_3,y1i); + psi_i_p3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_1_1,y1i); + psi_i_m1_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_1,y1i); + psi_i_m1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_1_3,y1i); + psi_i_m3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_3,y1i); + psi_i_m3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_1,y1i); + psi_i_p1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_1,y1i); + psi_i_p1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_3,y1i); + psi_i_p3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_3,y1i); + psi_i_p3_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_1,y1r); + psi_r_m1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_3,y1r); + psi_r_m1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_1,y1r); + psi_r_m3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_3,y1r); + psi_r_m3_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_1_1); + psi_r_m1_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_1_3); + psi_r_m1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_1_1); + psi_i_m1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_3_1); + psi_i_m1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_3_1); + psi_r_m3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_3_3); + psi_r_m3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_1_3); + psi_i_m3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_3_3); + psi_i_m3_m3 = _mm_abs_epi16(xmm2); + + // Rearrange desired MF output + xmm0 = stream0_128i_in[i]; + xmm1 = stream0_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)] + y0i = _mm_unpackhi_epi64(xmm0,xmm1); + + // Rearrange desired channel magnitudes + xmm2 = ch_mag_128i[i]; // = [|h|^2(1),|h|^2(1),|h|^2(2),|h|^2(2)]*(2/sqrt(10)) + xmm3 = ch_mag_128i[i+1]; // = [|h|^2(3),|h|^2(3),|h|^2(4),|h|^2(4)]*(2/sqrt(10)) + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + + ch_mag_des = _mm_unpacklo_epi64(xmm2,xmm3); // = [|h|^2(1),|h|^2(2),|h|^2(3),|h|^2(4)]*(2/sqrt(10)) + + // Rearrange interfering channel magnitudes + xmm2 = ch_mag_128i_i[i]; + xmm3 = ch_mag_128i_i[i+1]; + + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + + ch_mag_int = _mm_unpacklo_epi64(xmm2,xmm3); + + // Scale MF output of desired signal + y0r_over_sqrt10 = _mm_mulhi_epi16(y0r,ONE_OVER_SQRT_10); + y0i_over_sqrt10 = _mm_mulhi_epi16(y0i,ONE_OVER_SQRT_10); + y0r_three_over_sqrt10 = _mm_mulhi_epi16(y0r,THREE_OVER_SQRT_10); + y0i_three_over_sqrt10 = _mm_mulhi_epi16(y0i,THREE_OVER_SQRT_10); + y0r_three_over_sqrt10 = _mm_slli_epi16(y0r_three_over_sqrt10,1); + y0i_three_over_sqrt10 = _mm_slli_epi16(y0i_three_over_sqrt10,1); + + // Compute necessary combination of required terms + y0_p_1_1 = _mm_adds_epi16(y0r_over_sqrt10,y0i_over_sqrt10); + y0_m_1_1 = _mm_subs_epi16(y0r_over_sqrt10,y0i_over_sqrt10); + + y0_p_1_3 = _mm_adds_epi16(y0r_over_sqrt10,y0i_three_over_sqrt10); + y0_m_1_3 = _mm_subs_epi16(y0r_over_sqrt10,y0i_three_over_sqrt10); + + y0_p_3_1 = _mm_adds_epi16(y0r_three_over_sqrt10,y0i_over_sqrt10); + y0_m_3_1 = _mm_subs_epi16(y0r_three_over_sqrt10,y0i_over_sqrt10); + + y0_p_3_3 = _mm_adds_epi16(y0r_three_over_sqrt10,y0i_three_over_sqrt10); + y0_m_3_3 = _mm_subs_epi16(y0r_three_over_sqrt10,y0i_three_over_sqrt10); + + // Compute optimal interfering symbol magnitude + interference_abs_epi16(psi_r_p1_p1 ,ch_mag_int,a_r_p1_p1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_p1 ,ch_mag_int,a_i_p1_p1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_p3 ,ch_mag_int,a_r_p1_p3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_p3 ,ch_mag_int,a_i_p1_p3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_m1 ,ch_mag_int,a_r_p1_m1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_m1 ,ch_mag_int,a_i_p1_m1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_m3 ,ch_mag_int,a_r_p1_m3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_m3 ,ch_mag_int,a_i_p1_m3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_p1 ,ch_mag_int,a_r_p3_p1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_p1 ,ch_mag_int,a_i_p3_p1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_p3 ,ch_mag_int,a_r_p3_p3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_p3 ,ch_mag_int,a_i_p3_p3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_m1 ,ch_mag_int,a_r_p3_m1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_m1 ,ch_mag_int,a_i_p3_m1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_m3 ,ch_mag_int,a_r_p3_m3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_m3 ,ch_mag_int,a_i_p3_m3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_p1 ,ch_mag_int,a_r_m1_p1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_p1 ,ch_mag_int,a_i_m1_p1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_p3 ,ch_mag_int,a_r_m1_p3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_p3 ,ch_mag_int,a_i_m1_p3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_m1 ,ch_mag_int,a_r_m1_m1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_m1 ,ch_mag_int,a_i_m1_m1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_m3 ,ch_mag_int,a_r_m1_m3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_m3 ,ch_mag_int,a_i_m1_m3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_p1 ,ch_mag_int,a_r_m3_p1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_p1 ,ch_mag_int,a_i_m3_p1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_p3 ,ch_mag_int,a_r_m3_p3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_p3 ,ch_mag_int,a_i_m3_p3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_m1 ,ch_mag_int,a_r_m3_m1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_m1 ,ch_mag_int,a_i_m3_m1,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_m3 ,ch_mag_int,a_r_m3_m3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_m3 ,ch_mag_int,a_i_m3_m3,ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + + // Calculation of groups of two terms in the bit metric involving product of psi and interference magnitude + prodsum_psi_a_epi16(psi_r_p1_p1,a_r_p1_p1,psi_i_p1_p1,a_i_p1_p1,psi_a_p1_p1); + prodsum_psi_a_epi16(psi_r_p1_p3,a_r_p1_p3,psi_i_p1_p3,a_i_p1_p3,psi_a_p1_p3); + prodsum_psi_a_epi16(psi_r_p3_p1,a_r_p3_p1,psi_i_p3_p1,a_i_p3_p1,psi_a_p3_p1); + prodsum_psi_a_epi16(psi_r_p3_p3,a_r_p3_p3,psi_i_p3_p3,a_i_p3_p3,psi_a_p3_p3); + prodsum_psi_a_epi16(psi_r_p1_m1,a_r_p1_m1,psi_i_p1_m1,a_i_p1_m1,psi_a_p1_m1); + prodsum_psi_a_epi16(psi_r_p1_m3,a_r_p1_m3,psi_i_p1_m3,a_i_p1_m3,psi_a_p1_m3); + prodsum_psi_a_epi16(psi_r_p3_m1,a_r_p3_m1,psi_i_p3_m1,a_i_p3_m1,psi_a_p3_m1); + prodsum_psi_a_epi16(psi_r_p3_m3,a_r_p3_m3,psi_i_p3_m3,a_i_p3_m3,psi_a_p3_m3); + prodsum_psi_a_epi16(psi_r_m1_p1,a_r_m1_p1,psi_i_m1_p1,a_i_m1_p1,psi_a_m1_p1); + prodsum_psi_a_epi16(psi_r_m1_p3,a_r_m1_p3,psi_i_m1_p3,a_i_m1_p3,psi_a_m1_p3); + prodsum_psi_a_epi16(psi_r_m3_p1,a_r_m3_p1,psi_i_m3_p1,a_i_m3_p1,psi_a_m3_p1); + prodsum_psi_a_epi16(psi_r_m3_p3,a_r_m3_p3,psi_i_m3_p3,a_i_m3_p3,psi_a_m3_p3); + prodsum_psi_a_epi16(psi_r_m1_m1,a_r_m1_m1,psi_i_m1_m1,a_i_m1_m1,psi_a_m1_m1); + prodsum_psi_a_epi16(psi_r_m1_m3,a_r_m1_m3,psi_i_m1_m3,a_i_m1_m3,psi_a_m1_m3); + prodsum_psi_a_epi16(psi_r_m3_m1,a_r_m3_m1,psi_i_m3_m1,a_i_m3_m1,psi_a_m3_m1); + prodsum_psi_a_epi16(psi_r_m3_m3,a_r_m3_m3,psi_i_m3_m3,a_i_m3_m3,psi_a_m3_m3); + + + // squared interference magnitude times int. ch. power + square_a_epi16(a_r_p1_p1,a_i_p1_p1,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_p1_p1); + square_a_epi16(a_r_p1_p3,a_i_p1_p3,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_p1_p3); + square_a_epi16(a_r_p3_p1,a_i_p3_p1,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_p3_p1); + square_a_epi16(a_r_p3_p3,a_i_p3_p3,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_p3_p3); + square_a_epi16(a_r_p1_m1,a_i_p1_m1,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_p1_m1); + square_a_epi16(a_r_p1_m3,a_i_p1_m3,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_p1_m3); + square_a_epi16(a_r_p3_m1,a_i_p3_m1,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_p3_m1); + square_a_epi16(a_r_p3_m3,a_i_p3_m3,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_p3_m3); + square_a_epi16(a_r_m1_p1,a_i_m1_p1,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_m1_p1); + square_a_epi16(a_r_m1_p3,a_i_m1_p3,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_m1_p3); + square_a_epi16(a_r_m3_p1,a_i_m3_p1,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_m3_p1); + square_a_epi16(a_r_m3_p3,a_i_m3_p3,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_m3_p3); + square_a_epi16(a_r_m1_m1,a_i_m1_m1,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_m1_m1); + square_a_epi16(a_r_m1_m3,a_i_m1_m3,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_m1_m3); + square_a_epi16(a_r_m3_m1,a_i_m3_m1,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_m3_m1); + square_a_epi16(a_r_m3_m3,a_i_m3_m3,ch_mag_int,SQRT_10_OVER_FOUR,a_sq_m3_m3); + + // Computing different multiples of channel norms + ch_mag_over_10=_mm_mulhi_epi16(ch_mag_des, ONE_OVER_TWO_SQRT_10); + ch_mag_over_2=_mm_mulhi_epi16(ch_mag_des, SQRT_10_OVER_FOUR); + ch_mag_over_2=_mm_slli_epi16(ch_mag_over_2, 1); + ch_mag_9_over_10=_mm_mulhi_epi16(ch_mag_des, NINE_OVER_TWO_SQRT_10); + ch_mag_9_over_10=_mm_slli_epi16(ch_mag_9_over_10, 2); + + // Computing Metrics + xmm0 = _mm_subs_epi16(psi_a_p1_p1,a_sq_p1_p1); + xmm1 = _mm_adds_epi16(xmm0,y0_p_1_1); + bit_met_p1_p1= _mm_subs_epi16(xmm1,ch_mag_over_10); + + xmm0 = _mm_subs_epi16(psi_a_p1_p3,a_sq_p1_p3); + xmm1 = _mm_adds_epi16(xmm0,y0_p_1_3); + bit_met_p1_p3= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_p1_m1,a_sq_p1_m1); + xmm1 = _mm_adds_epi16(xmm0,y0_m_1_1); + bit_met_p1_m1= _mm_subs_epi16(xmm1,ch_mag_over_10); + + xmm0 = _mm_subs_epi16(psi_a_p1_m3,a_sq_p1_m3); + xmm1 = _mm_adds_epi16(xmm0,y0_m_1_3); + bit_met_p1_m3= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_p3_p1,a_sq_p3_p1); + xmm1 = _mm_adds_epi16(xmm0,y0_p_3_1); + bit_met_p3_p1= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_p3_p3,a_sq_p3_p3); + xmm1 = _mm_adds_epi16(xmm0,y0_p_3_3); + bit_met_p3_p3= _mm_subs_epi16(xmm1,ch_mag_9_over_10); + + xmm0 = _mm_subs_epi16(psi_a_p3_m1,a_sq_p3_m1); + xmm1 = _mm_adds_epi16(xmm0,y0_m_3_1); + bit_met_p3_m1= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_p3_m3,a_sq_p3_m3); + xmm1 = _mm_adds_epi16(xmm0,y0_m_3_3); + bit_met_p3_m3= _mm_subs_epi16(xmm1,ch_mag_9_over_10); + + xmm0 = _mm_subs_epi16(psi_a_m1_p1,a_sq_m1_p1); + xmm1 = _mm_subs_epi16(xmm0,y0_m_1_1); + bit_met_m1_p1= _mm_subs_epi16(xmm1,ch_mag_over_10); + + xmm0 = _mm_subs_epi16(psi_a_m1_p3,a_sq_m1_p3); + xmm1 = _mm_subs_epi16(xmm0,y0_m_1_3); + bit_met_m1_p3= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_m1_m1,a_sq_m1_m1); + xmm1 = _mm_subs_epi16(xmm0,y0_p_1_1); + bit_met_m1_m1= _mm_subs_epi16(xmm1,ch_mag_over_10); + + xmm0 = _mm_subs_epi16(psi_a_m1_m3,a_sq_m1_m3); + xmm1 = _mm_subs_epi16(xmm0,y0_p_1_3); + bit_met_m1_m3= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_m3_p1,a_sq_m3_p1); + xmm1 = _mm_subs_epi16(xmm0,y0_m_3_1); + bit_met_m3_p1= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_m3_p3,a_sq_m3_p3); + xmm1 = _mm_subs_epi16(xmm0,y0_m_3_3); + bit_met_m3_p3= _mm_subs_epi16(xmm1,ch_mag_9_over_10); + + xmm0 = _mm_subs_epi16(psi_a_m3_m1,a_sq_m3_m1); + xmm1 = _mm_subs_epi16(xmm0,y0_p_3_1); + bit_met_m3_m1= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_m3_m3,a_sq_m3_m3); + xmm1 = _mm_subs_epi16(xmm0,y0_p_3_3); + bit_met_m3_m3= _mm_subs_epi16(xmm1,ch_mag_9_over_10); + + // LLR of the first bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_m1_p1,bit_met_m1_p3); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m1_m3); + xmm2 = _mm_max_epi16(bit_met_m3_p1,bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m3_m1,bit_met_m3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_re0= _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_p1_p1,bit_met_p1_p3); + xmm1 = _mm_max_epi16(bit_met_p1_m1,bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_p3_p1,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_m1,bit_met_p3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4,xmm5); + + // LLR of first bit [L1(1), L1(2), L1(3), L1(4), L1(5), L1(6), L1(7), L1(8)] + y0r = _mm_subs_epi16(logmax_den_re0,logmax_num_re0); + + // LLR of the second bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_p1_m1,bit_met_p3_m1); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_p1_m3,bit_met_p3_m3); + xmm3 = _mm_max_epi16(bit_met_m1_m3,bit_met_m3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_re1 = _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_p1_p1,bit_met_p3_p1); + xmm1 = _mm_max_epi16(bit_met_m1_p1,bit_met_m3_p1); + xmm2 = _mm_max_epi16(bit_met_p1_p3,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p3,bit_met_m3_p3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_re1 = _mm_max_epi16(xmm4,xmm5); + + // LLR of second bit [L2(1), L2(2), L2(3), L2(4)] + y1r = _mm_subs_epi16(logmax_den_re1,logmax_num_re1); + + // LLR of the third bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_m3_p1,bit_met_m3_p3); + xmm1 = _mm_max_epi16(bit_met_m3_m1,bit_met_m3_m3); + xmm2 = _mm_max_epi16(bit_met_p3_p1,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_m1,bit_met_p3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_im0 = _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_m1_p1,bit_met_m1_p3); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m1_m3); + xmm2 = _mm_max_epi16(bit_met_p1_p1,bit_met_p1_p3); + xmm3 = _mm_max_epi16(bit_met_p1_m1,bit_met_p1_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_im0 = _mm_max_epi16(xmm4,xmm5); + + // LLR of third bit [L3(1), L3(2), L3(3), L3(4)] + y0i = _mm_subs_epi16(logmax_den_im0,logmax_num_im0); + + // LLR of the fourth bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_p1_m3,bit_met_p3_m3); + xmm1 = _mm_max_epi16(bit_met_m1_m3,bit_met_m3_m3); + xmm2 = _mm_max_epi16(bit_met_p1_p3,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p3,bit_met_m3_p3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_im1 = _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_p1_m1,bit_met_p3_m1); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_p1_p1,bit_met_p3_p1); + xmm3 = _mm_max_epi16(bit_met_m1_p1,bit_met_m3_p1); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_im1 = _mm_max_epi16(xmm4,xmm5); + + // LLR of fourth bit [L4(1), L4(2), L4(3), L4(4)] + y1i = _mm_subs_epi16(logmax_den_im1,logmax_num_im1); + + // Pack LLRs in output + // [L1(1), L2(1), L1(2), L2(2), L1(3), L2(3), L1(4), L2(4)] + xmm0 = _mm_unpacklo_epi16(y0r,y1r); + // [L1(5), L2(5), L1(6), L2(6), L1(7), L2(7), L1(8), L2(8)] + xmm1 = _mm_unpackhi_epi16(y0r,y1r); + // [L3(1), L4(1), L3(2), L4(2), L3(3), L4(3), L3(4), L4(4)] + xmm2 = _mm_unpacklo_epi16(y0i,y1i); + // [L3(5), L4(5), L3(6), L4(6), L3(7), L4(7), L3(8), L4(8)] + xmm3 = _mm_unpackhi_epi16(y0i,y1i); + + stream0_128i_out[2*i+0] = _mm_unpacklo_epi32(xmm0,xmm2); // 8LLRs, 2REs + stream0_128i_out[2*i+1] = _mm_unpackhi_epi32(xmm0,xmm2); + stream0_128i_out[2*i+2] = _mm_unpacklo_epi32(xmm1,xmm3); + stream0_128i_out[2*i+3] = _mm_unpackhi_epi32(xmm1,xmm3); +#elif defined(__arm__) + +#endif + + } + +#if defined(__x86_64__) || defined(__i386__) + _mm_empty(); + _m_empty(); +#endif +} + +int nr_dlsch_16qam_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, + int32_t **rxdataF_comp, + int32_t **rxdataF_comp_i, + int32_t **dl_ch_mag, //|h_0|^2*(2/sqrt{10}) + int32_t **dl_ch_mag_i, //|h_1|^2*(2/sqrt{10}) + int32_t **rho_i, + int16_t *dlsch_llr, + uint8_t symbol, + uint8_t first_symbol_flag, + uint16_t nb_rb, + uint16_t pbch_pss_sss_adjust, + int16_t **llr16p) +{ + + int16_t *rxF = (int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rxF_i = (int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag = (int16_t*)&dl_ch_mag[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag_i = (int16_t*)&dl_ch_mag_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rho = (int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *llr16; + int len; + uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol; + + // first symbol has different structure due to more pilots + if (first_symbol_flag == 1) { + llr16 = (int16_t*)dlsch_llr; + } else { + llr16 = (int16_t*)(*llr16p); + } + + + AssertFatal(llr16!=NULL,"nr_dlsch_16qam_16qam_llr: llr is null, symbol %d\n",symbol); + + if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) { + // if symbol has pilots + if (frame_parms->nb_antenna_ports_eNB!=1) + // in 2 antenna ports we have 8 REs per symbol per RB + len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3); + else + // for 1 antenna port we have 10 REs per symbol per RB + len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6); + } else { + // symbol has no pilots + len = (nb_rb*12) - pbch_pss_sss_adjust; + } + + // printf("symbol %d: qam16_llr, len %d (llr16 %p)\n",symbol,len,llr16); + + qam16_qam16((short *)rxF, + (short *)rxF_i, + (short *)ch_mag, + (short *)ch_mag_i, + (short *)llr16, + (short *)rho, + len); + + llr16 += (len<<2); + *llr16p = (short *)llr16; + + return(0); +} + +void qam16_qam64(int16_t *stream0_in, + int16_t *stream1_in, + int16_t *ch_mag, + int16_t *ch_mag_i, + int16_t *stream0_out, + int16_t *rho01, + int32_t length + ) +{ + + /* + Author: Sebastian Wagner + Date: 2012-06-04 + + Input: + stream0_in: MF filter for 1st stream, i.e., y0=h0'*y + stream!_in: MF filter for 2nd stream, i.e., y1=h1'*y + ch_mag: 2*h0/sqrt(00), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + ch_mag_i: 2*h1/sqrt(00), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + rho01: Channel cross correlation, i.e., h1'*h0 + + Output: + stream0_out: output LLRs for 1st stream + */ + +#if defined(__x86_64__) || defined(__i386__) + __m128i *rho01_128i = (__m128i *)rho01; + __m128i *stream0_128i_in = (__m128i *)stream0_in; + __m128i *stream1_128i_in = (__m128i *)stream1_in; + __m128i *stream0_128i_out = (__m128i *)stream0_out; + __m128i *ch_mag_128i = (__m128i *)ch_mag; + __m128i *ch_mag_128i_i = (__m128i *)ch_mag_i; + + + __m128i ONE_OVER_SQRT_2 = _mm_set1_epi16(23170); // round(1/sqrt(2)*2^15) + __m128i ONE_OVER_SQRT_10 = _mm_set1_epi16(20724); // round(1/sqrt(10)*2^16) + __m128i THREE_OVER_SQRT_10 = _mm_set1_epi16(31086); // round(3/sqrt(10)*2^15) + __m128i SQRT_10_OVER_FOUR = _mm_set1_epi16(25905); // round(sqrt(10)/4*2^15) + __m128i ONE_OVER_TWO_SQRT_10 = _mm_set1_epi16(10362); // round(1/2/sqrt(10)*2^16) + __m128i NINE_OVER_TWO_SQRT_10 = _mm_set1_epi16(23315); // round(9/2/sqrt(10)*2^14) + __m128i ONE_OVER_SQRT_2_42 = _mm_set1_epi16(3575); // round(1/sqrt(2*42)*2^15) + __m128i THREE_OVER_SQRT_2_42 = _mm_set1_epi16(10726); // round(3/sqrt(2*42)*2^15) + __m128i FIVE_OVER_SQRT_2_42 = _mm_set1_epi16(17876); // round(5/sqrt(2*42)*2^15) + __m128i SEVEN_OVER_SQRT_2_42 = _mm_set1_epi16(25027); // round(7/sqrt(2*42)*2^15) + __m128i SQRT_42_OVER_FOUR = _mm_set1_epi16(13272); // round(sqrt(42)/4*2^13), Q3. + __m128i ch_mag_des,ch_mag_int; + __m128i y0r_over_sqrt10; + __m128i y0i_over_sqrt10; + __m128i y0r_three_over_sqrt10; + __m128i y0i_three_over_sqrt10; + __m128i ch_mag_over_10; + __m128i ch_mag_over_2; + __m128i ch_mag_9_over_10; + __m128i ch_mag_int_with_sigma2; + __m128i two_ch_mag_int_with_sigma2; + __m128i three_ch_mag_int_with_sigma2; + +#elif defined(__arm__) + +#endif + int i; + + for (i=0; i>2; i+=2) { + // In one iteration, we deal with 8 REs + +#if defined(__x86_64__) || defined(__i386__) + // Get rho + xmm0 = rho01_128i[i]; + xmm1 = rho01_128i[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho) + xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho) + rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho) + rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho) + + // Compute the different rhos + rho_rpi_1_1 = _mm_mulhi_epi16(rho_rpi,ONE_OVER_SQRT_10); + rho_rmi_1_1 = _mm_mulhi_epi16(rho_rmi,ONE_OVER_SQRT_10); + rho_rpi_3_3 = _mm_mulhi_epi16(rho_rpi,THREE_OVER_SQRT_10); + rho_rmi_3_3 = _mm_mulhi_epi16(rho_rmi,THREE_OVER_SQRT_10); + rho_rpi_3_3 = _mm_slli_epi16(rho_rpi_3_3,1); + rho_rmi_3_3 = _mm_slli_epi16(rho_rmi_3_3,1); + + xmm4 = _mm_mulhi_epi16(xmm2,ONE_OVER_SQRT_10); // Re(rho) + xmm5 = _mm_mulhi_epi16(xmm3,THREE_OVER_SQRT_10); // Im(rho) + xmm5 = _mm_slli_epi16(xmm5,1); + + rho_rpi_1_3 = _mm_adds_epi16(xmm4,xmm5); + rho_rmi_1_3 = _mm_subs_epi16(xmm4,xmm5); + + xmm6 = _mm_mulhi_epi16(xmm2,THREE_OVER_SQRT_10); // Re(rho) + xmm7 = _mm_mulhi_epi16(xmm3,ONE_OVER_SQRT_10); // Im(rho) + xmm6 = _mm_slli_epi16(xmm6,1); + + rho_rpi_3_1 = _mm_adds_epi16(xmm6,xmm7); + rho_rmi_3_1 = _mm_subs_epi16(xmm6,xmm7); + + // Rearrange interfering MF output + xmm0 = stream1_128i_in[i]; + xmm1 = stream1_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)] + y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)] + + xmm0 = _mm_setzero_si128(); // ZERO + xmm2 = _mm_subs_epi16(rho_rpi_1_1,y1r); // = [Re(rho)+ Im(rho)]/sqrt(10) - y1r + psi_r_p1_p1 = _mm_abs_epi16(xmm2); // = |[Re(rho)+ Im(rho)]/sqrt(10) - y1r| + + xmm2= _mm_subs_epi16(rho_rmi_1_1,y1r); + psi_r_p1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_1_1,y1i); + psi_i_p1_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_1_3,y1r); + psi_r_p1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_1_3,y1r); + psi_r_p1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_1,y1i); + psi_i_p1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_1,y1r); + psi_r_p3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_1,y1r); + psi_r_p3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_1_3,y1i); + psi_i_p3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_3,y1r); + psi_r_p3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_3,y1r); + psi_r_p3_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rmi_3_3,y1i); + psi_i_p3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_1_1,y1i); + psi_i_m1_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_1,y1i); + psi_i_m1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_1_3,y1i); + psi_i_m3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_subs_epi16(rho_rpi_3_3,y1i); + psi_i_m3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_1,y1i); + psi_i_p1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_1,y1i); + psi_i_p1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_3,y1i); + psi_i_p3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_3,y1i); + psi_i_p3_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_1,y1r); + psi_r_m1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_1_3,y1r); + psi_r_m1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_1,y1r); + psi_r_m3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(rho_rpi_3_3,y1r); + psi_r_m3_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_1_1); + psi_r_m1_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_1_3); + psi_r_m1_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_1_1); + psi_i_m1_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_3_1); + psi_i_m1_m3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_3_1); + psi_r_m3_p1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1r,rho_rmi_3_3); + psi_r_m3_p3 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_1_3); + psi_i_m3_m1 = _mm_abs_epi16(xmm2); + xmm2= _mm_adds_epi16(y1i,rho_rmi_3_3); + psi_i_m3_m3 = _mm_abs_epi16(xmm2); + + // Rearrange desired MF output + xmm0 = stream0_128i_in[i]; + xmm1 = stream0_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)] + y0i = _mm_unpackhi_epi64(xmm0,xmm1); + + // Rearrange desired channel magnitudes + xmm2 = ch_mag_128i[i]; // = [|h|^2(1),|h|^2(1),|h|^2(2),|h|^2(2)]*(2/sqrt(10)) + xmm3 = ch_mag_128i[i+1]; // = [|h|^2(3),|h|^2(3),|h|^2(4),|h|^2(4)]*(2/sqrt(10)) + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + + ch_mag_des = _mm_unpacklo_epi64(xmm2,xmm3); // = [|h|^2(1),|h|^2(2),|h|^2(3),|h|^2(4)]*(2/sqrt(10)) + + // Rearrange interfering channel magnitudes + xmm2 = ch_mag_128i_i[i]; + xmm3 = ch_mag_128i_i[i+1]; + + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + + ch_mag_int = _mm_unpacklo_epi64(xmm2,xmm3); + + // Scale MF output of desired signal + y0r_over_sqrt10 = _mm_mulhi_epi16(y0r,ONE_OVER_SQRT_10); + y0i_over_sqrt10 = _mm_mulhi_epi16(y0i,ONE_OVER_SQRT_10); + y0r_three_over_sqrt10 = _mm_mulhi_epi16(y0r,THREE_OVER_SQRT_10); + y0i_three_over_sqrt10 = _mm_mulhi_epi16(y0i,THREE_OVER_SQRT_10); + y0r_three_over_sqrt10 = _mm_slli_epi16(y0r_three_over_sqrt10,1); + y0i_three_over_sqrt10 = _mm_slli_epi16(y0i_three_over_sqrt10,1); + + // Compute necessary combination of required terms + y0_p_1_1 = _mm_adds_epi16(y0r_over_sqrt10,y0i_over_sqrt10); + y0_m_1_1 = _mm_subs_epi16(y0r_over_sqrt10,y0i_over_sqrt10); + + y0_p_1_3 = _mm_adds_epi16(y0r_over_sqrt10,y0i_three_over_sqrt10); + y0_m_1_3 = _mm_subs_epi16(y0r_over_sqrt10,y0i_three_over_sqrt10); + + y0_p_3_1 = _mm_adds_epi16(y0r_three_over_sqrt10,y0i_over_sqrt10); + y0_m_3_1 = _mm_subs_epi16(y0r_three_over_sqrt10,y0i_over_sqrt10); + + y0_p_3_3 = _mm_adds_epi16(y0r_three_over_sqrt10,y0i_three_over_sqrt10); + y0_m_3_3 = _mm_subs_epi16(y0r_three_over_sqrt10,y0i_three_over_sqrt10); + + // Compute optimal interfering symbol magnitude + ch_mag_int_with_sigma2 = _mm_srai_epi16(ch_mag_int, 1); // *2 + two_ch_mag_int_with_sigma2 = ch_mag_int; // *4 + three_ch_mag_int_with_sigma2 = _mm_adds_epi16(ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2); // *6 + + interference_abs_64qam_epi16(psi_r_p1_p1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_p1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_p1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_p1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_p3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_p3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_p3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_p3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_m1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_m1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_m1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_m1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_m3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_m3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_m3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_m3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_p1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_p1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_p1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_p1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_p3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_p3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_p3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_p3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_m1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_m1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_m1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_m1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_m3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_m3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_m3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_m3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_p1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_p1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_p1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_p1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_p3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_p3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_p3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_p3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_m1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_m1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_m1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_m1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_m3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_m3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_m3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_m3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_p1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_p1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_p1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_p1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_p3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_p3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_p3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_p3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_m1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_m1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_m1 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_m1,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_m3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_m3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_m3 ,ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_m3,ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42,FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + + // Calculation of groups of two terms in the bit metric involving product of psi and interference magnitude + prodsum_psi_a_epi16(psi_r_p1_p1,a_r_p1_p1,psi_i_p1_p1,a_i_p1_p1,psi_a_p1_p1); + prodsum_psi_a_epi16(psi_r_p1_p3,a_r_p1_p3,psi_i_p1_p3,a_i_p1_p3,psi_a_p1_p3); + prodsum_psi_a_epi16(psi_r_p3_p1,a_r_p3_p1,psi_i_p3_p1,a_i_p3_p1,psi_a_p3_p1); + prodsum_psi_a_epi16(psi_r_p3_p3,a_r_p3_p3,psi_i_p3_p3,a_i_p3_p3,psi_a_p3_p3); + prodsum_psi_a_epi16(psi_r_p1_m1,a_r_p1_m1,psi_i_p1_m1,a_i_p1_m1,psi_a_p1_m1); + prodsum_psi_a_epi16(psi_r_p1_m3,a_r_p1_m3,psi_i_p1_m3,a_i_p1_m3,psi_a_p1_m3); + prodsum_psi_a_epi16(psi_r_p3_m1,a_r_p3_m1,psi_i_p3_m1,a_i_p3_m1,psi_a_p3_m1); + prodsum_psi_a_epi16(psi_r_p3_m3,a_r_p3_m3,psi_i_p3_m3,a_i_p3_m3,psi_a_p3_m3); + prodsum_psi_a_epi16(psi_r_m1_p1,a_r_m1_p1,psi_i_m1_p1,a_i_m1_p1,psi_a_m1_p1); + prodsum_psi_a_epi16(psi_r_m1_p3,a_r_m1_p3,psi_i_m1_p3,a_i_m1_p3,psi_a_m1_p3); + prodsum_psi_a_epi16(psi_r_m3_p1,a_r_m3_p1,psi_i_m3_p1,a_i_m3_p1,psi_a_m3_p1); + prodsum_psi_a_epi16(psi_r_m3_p3,a_r_m3_p3,psi_i_m3_p3,a_i_m3_p3,psi_a_m3_p3); + prodsum_psi_a_epi16(psi_r_m1_m1,a_r_m1_m1,psi_i_m1_m1,a_i_m1_m1,psi_a_m1_m1); + prodsum_psi_a_epi16(psi_r_m1_m3,a_r_m1_m3,psi_i_m1_m3,a_i_m1_m3,psi_a_m1_m3); + prodsum_psi_a_epi16(psi_r_m3_m1,a_r_m3_m1,psi_i_m3_m1,a_i_m3_m1,psi_a_m3_m1); + prodsum_psi_a_epi16(psi_r_m3_m3,a_r_m3_m3,psi_i_m3_m3,a_i_m3_m3,psi_a_m3_m3); + + // Multiply by sqrt(2) + psi_a_p1_p1 = _mm_mulhi_epi16(psi_a_p1_p1, ONE_OVER_SQRT_2); + psi_a_p1_p1 = _mm_slli_epi16(psi_a_p1_p1, 2); + psi_a_p1_p3 = _mm_mulhi_epi16(psi_a_p1_p3, ONE_OVER_SQRT_2); + psi_a_p1_p3 = _mm_slli_epi16(psi_a_p1_p3, 2); + psi_a_p3_p1 = _mm_mulhi_epi16(psi_a_p3_p1, ONE_OVER_SQRT_2); + psi_a_p3_p1 = _mm_slli_epi16(psi_a_p3_p1, 2); + psi_a_p3_p3 = _mm_mulhi_epi16(psi_a_p3_p3, ONE_OVER_SQRT_2); + psi_a_p3_p3 = _mm_slli_epi16(psi_a_p3_p3, 2); + psi_a_p1_m1 = _mm_mulhi_epi16(psi_a_p1_m1, ONE_OVER_SQRT_2); + psi_a_p1_m1 = _mm_slli_epi16(psi_a_p1_m1, 2); + psi_a_p1_m3 = _mm_mulhi_epi16(psi_a_p1_m3, ONE_OVER_SQRT_2); + psi_a_p1_m3 = _mm_slli_epi16(psi_a_p1_m3, 2); + psi_a_p3_m1 = _mm_mulhi_epi16(psi_a_p3_m1, ONE_OVER_SQRT_2); + psi_a_p3_m1 = _mm_slli_epi16(psi_a_p3_m1, 2); + psi_a_p3_m3 = _mm_mulhi_epi16(psi_a_p3_m3, ONE_OVER_SQRT_2); + psi_a_p3_m3 = _mm_slli_epi16(psi_a_p3_m3, 2); + psi_a_m1_p1 = _mm_mulhi_epi16(psi_a_m1_p1, ONE_OVER_SQRT_2); + psi_a_m1_p1 = _mm_slli_epi16(psi_a_m1_p1, 2); + psi_a_m1_p3 = _mm_mulhi_epi16(psi_a_m1_p3, ONE_OVER_SQRT_2); + psi_a_m1_p3 = _mm_slli_epi16(psi_a_m1_p3, 2); + psi_a_m3_p1 = _mm_mulhi_epi16(psi_a_m3_p1, ONE_OVER_SQRT_2); + psi_a_m3_p1 = _mm_slli_epi16(psi_a_m3_p1, 2); + psi_a_m3_p3 = _mm_mulhi_epi16(psi_a_m3_p3, ONE_OVER_SQRT_2); + psi_a_m3_p3 = _mm_slli_epi16(psi_a_m3_p3, 2); + psi_a_m1_m1 = _mm_mulhi_epi16(psi_a_m1_m1, ONE_OVER_SQRT_2); + psi_a_m1_m1 = _mm_slli_epi16(psi_a_m1_m1, 2); + psi_a_m1_m3 = _mm_mulhi_epi16(psi_a_m1_m3, ONE_OVER_SQRT_2); + psi_a_m1_m3 = _mm_slli_epi16(psi_a_m1_m3, 2); + psi_a_m3_m1 = _mm_mulhi_epi16(psi_a_m3_m1, ONE_OVER_SQRT_2); + psi_a_m3_m1 = _mm_slli_epi16(psi_a_m3_m1, 2); + psi_a_m3_m3 = _mm_mulhi_epi16(psi_a_m3_m3, ONE_OVER_SQRT_2); + psi_a_m3_m3 = _mm_slli_epi16(psi_a_m3_m3, 2); + + // squared interference magnitude times int. ch. power + square_a_64qam_epi16(a_r_p1_p1,a_i_p1_p1,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_p1_p1); + square_a_64qam_epi16(a_r_p1_p3,a_i_p1_p3,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_p1_p3); + square_a_64qam_epi16(a_r_p3_p1,a_i_p3_p1,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_p3_p1); + square_a_64qam_epi16(a_r_p3_p3,a_i_p3_p3,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_p3_p3); + square_a_64qam_epi16(a_r_p1_m1,a_i_p1_m1,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_p1_m1); + square_a_64qam_epi16(a_r_p1_m3,a_i_p1_m3,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_p1_m3); + square_a_64qam_epi16(a_r_p3_m1,a_i_p3_m1,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_p3_m1); + square_a_64qam_epi16(a_r_p3_m3,a_i_p3_m3,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_p3_m3); + square_a_64qam_epi16(a_r_m1_p1,a_i_m1_p1,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_m1_p1); + square_a_64qam_epi16(a_r_m1_p3,a_i_m1_p3,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_m1_p3); + square_a_64qam_epi16(a_r_m3_p1,a_i_m3_p1,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_m3_p1); + square_a_64qam_epi16(a_r_m3_p3,a_i_m3_p3,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_m3_p3); + square_a_64qam_epi16(a_r_m1_m1,a_i_m1_m1,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_m1_m1); + square_a_64qam_epi16(a_r_m1_m3,a_i_m1_m3,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_m1_m3); + square_a_64qam_epi16(a_r_m3_m1,a_i_m3_m1,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_m3_m1); + square_a_64qam_epi16(a_r_m3_m3,a_i_m3_m3,ch_mag_int,SQRT_42_OVER_FOUR,a_sq_m3_m3); + + // Computing different multiples of channel norms + ch_mag_over_10=_mm_mulhi_epi16(ch_mag_des, ONE_OVER_TWO_SQRT_10); + ch_mag_over_2=_mm_mulhi_epi16(ch_mag_des, SQRT_10_OVER_FOUR); + ch_mag_over_2=_mm_slli_epi16(ch_mag_over_2, 1); + ch_mag_9_over_10=_mm_mulhi_epi16(ch_mag_des, NINE_OVER_TWO_SQRT_10); + ch_mag_9_over_10=_mm_slli_epi16(ch_mag_9_over_10, 2); + + // Computing Metrics + xmm0 = _mm_subs_epi16(psi_a_p1_p1,a_sq_p1_p1); + xmm1 = _mm_adds_epi16(xmm0,y0_p_1_1); + bit_met_p1_p1= _mm_subs_epi16(xmm1,ch_mag_over_10); + + xmm0 = _mm_subs_epi16(psi_a_p1_p3,a_sq_p1_p3); + xmm1 = _mm_adds_epi16(xmm0,y0_p_1_3); + bit_met_p1_p3= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_p1_m1,a_sq_p1_m1); + xmm1 = _mm_adds_epi16(xmm0,y0_m_1_1); + bit_met_p1_m1= _mm_subs_epi16(xmm1,ch_mag_over_10); + + xmm0 = _mm_subs_epi16(psi_a_p1_m3,a_sq_p1_m3); + xmm1 = _mm_adds_epi16(xmm0,y0_m_1_3); + bit_met_p1_m3= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_p3_p1,a_sq_p3_p1); + xmm1 = _mm_adds_epi16(xmm0,y0_p_3_1); + bit_met_p3_p1= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_p3_p3,a_sq_p3_p3); + xmm1 = _mm_adds_epi16(xmm0,y0_p_3_3); + bit_met_p3_p3= _mm_subs_epi16(xmm1,ch_mag_9_over_10); + + xmm0 = _mm_subs_epi16(psi_a_p3_m1,a_sq_p3_m1); + xmm1 = _mm_adds_epi16(xmm0,y0_m_3_1); + bit_met_p3_m1= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_p3_m3,a_sq_p3_m3); + xmm1 = _mm_adds_epi16(xmm0,y0_m_3_3); + bit_met_p3_m3= _mm_subs_epi16(xmm1,ch_mag_9_over_10); + + xmm0 = _mm_subs_epi16(psi_a_m1_p1,a_sq_m1_p1); + xmm1 = _mm_subs_epi16(xmm0,y0_m_1_1); + bit_met_m1_p1= _mm_subs_epi16(xmm1,ch_mag_over_10); + + xmm0 = _mm_subs_epi16(psi_a_m1_p3,a_sq_m1_p3); + xmm1 = _mm_subs_epi16(xmm0,y0_m_1_3); + bit_met_m1_p3= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_m1_m1,a_sq_m1_m1); + xmm1 = _mm_subs_epi16(xmm0,y0_p_1_1); + bit_met_m1_m1= _mm_subs_epi16(xmm1,ch_mag_over_10); + + xmm0 = _mm_subs_epi16(psi_a_m1_m3,a_sq_m1_m3); + xmm1 = _mm_subs_epi16(xmm0,y0_p_1_3); + bit_met_m1_m3= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_m3_p1,a_sq_m3_p1); + xmm1 = _mm_subs_epi16(xmm0,y0_m_3_1); + bit_met_m3_p1= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_m3_p3,a_sq_m3_p3); + xmm1 = _mm_subs_epi16(xmm0,y0_m_3_3); + bit_met_m3_p3= _mm_subs_epi16(xmm1,ch_mag_9_over_10); + + xmm0 = _mm_subs_epi16(psi_a_m3_m1,a_sq_m3_m1); + xmm1 = _mm_subs_epi16(xmm0,y0_p_3_1); + bit_met_m3_m1= _mm_subs_epi16(xmm1,ch_mag_over_2); + + xmm0 = _mm_subs_epi16(psi_a_m3_m3,a_sq_m3_m3); + xmm1 = _mm_subs_epi16(xmm0,y0_p_3_3); + bit_met_m3_m3= _mm_subs_epi16(xmm1,ch_mag_9_over_10); + + // LLR of the first bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_m1_p1,bit_met_m1_p3); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m1_m3); + xmm2 = _mm_max_epi16(bit_met_m3_p1,bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m3_m1,bit_met_m3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_re0= _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_p1_p1,bit_met_p1_p3); + xmm1 = _mm_max_epi16(bit_met_p1_m1,bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_p3_p1,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_m1,bit_met_p3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4,xmm5); + + // LLR of first bit [L1(1), L1(2), L1(3), L1(4), L1(5), L1(6), L1(7), L1(8)] + y0r = _mm_subs_epi16(logmax_den_re0,logmax_num_re0); + + // LLR of the second bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_p1_m1,bit_met_p3_m1); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_p1_m3,bit_met_p3_m3); + xmm3 = _mm_max_epi16(bit_met_m1_m3,bit_met_m3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_re1 = _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_p1_p1,bit_met_p3_p1); + xmm1 = _mm_max_epi16(bit_met_m1_p1,bit_met_m3_p1); + xmm2 = _mm_max_epi16(bit_met_p1_p3,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p3,bit_met_m3_p3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_re1 = _mm_max_epi16(xmm4,xmm5); + + // LLR of second bit [L2(1), L2(2), L2(3), L2(4)] + y1r = _mm_subs_epi16(logmax_den_re1,logmax_num_re1); + + // LLR of the third bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_m3_p1,bit_met_m3_p3); + xmm1 = _mm_max_epi16(bit_met_m3_m1,bit_met_m3_m3); + xmm2 = _mm_max_epi16(bit_met_p3_p1,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_m1,bit_met_p3_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_im0 = _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_m1_p1,bit_met_m1_p3); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m1_m3); + xmm2 = _mm_max_epi16(bit_met_p1_p1,bit_met_p1_p3); + xmm3 = _mm_max_epi16(bit_met_p1_m1,bit_met_p1_m3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_im0 = _mm_max_epi16(xmm4,xmm5); + + // LLR of third bit [L3(1), L3(2), L3(3), L3(4)] + y0i = _mm_subs_epi16(logmax_den_im0,logmax_num_im0); + + // LLR of the fourth bit + // Bit = 1 + xmm0 = _mm_max_epi16(bit_met_p1_m3,bit_met_p3_m3); + xmm1 = _mm_max_epi16(bit_met_m1_m3,bit_met_m3_m3); + xmm2 = _mm_max_epi16(bit_met_p1_p3,bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p3,bit_met_m3_p3); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_num_im1 = _mm_max_epi16(xmm4,xmm5); + + // Bit = 0 + xmm0 = _mm_max_epi16(bit_met_p1_m1,bit_met_p3_m1); + xmm1 = _mm_max_epi16(bit_met_m1_m1,bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_p1_p1,bit_met_p3_p1); + xmm3 = _mm_max_epi16(bit_met_m1_p1,bit_met_m3_p1); + xmm4 = _mm_max_epi16(xmm0,xmm1); + xmm5 = _mm_max_epi16(xmm2,xmm3); + logmax_den_im1 = _mm_max_epi16(xmm4,xmm5); + + // LLR of fourth bit [L4(1), L4(2), L4(3), L4(4)] + y1i = _mm_subs_epi16(logmax_den_im1,logmax_num_im1); + + // Pack LLRs in output + // [L1(1), L2(1), L1(2), L2(2), L1(3), L2(3), L1(4), L2(4)] + xmm0 = _mm_unpacklo_epi16(y0r,y1r); + // [L1(5), L2(5), L1(6), L2(6), L1(7), L2(7), L1(8), L2(8)] + xmm1 = _mm_unpackhi_epi16(y0r,y1r); + // [L3(1), L4(1), L3(2), L4(2), L3(3), L4(3), L3(4), L4(4)] + xmm2 = _mm_unpacklo_epi16(y0i,y1i); + // [L3(5), L4(5), L3(6), L4(6), L3(7), L4(7), L3(8), L4(8)] + xmm3 = _mm_unpackhi_epi16(y0i,y1i); + + stream0_128i_out[2*i+0] = _mm_unpacklo_epi32(xmm0,xmm2); // 8LLRs, 2REs + stream0_128i_out[2*i+1] = _mm_unpackhi_epi32(xmm0,xmm2); + stream0_128i_out[2*i+2] = _mm_unpacklo_epi32(xmm1,xmm3); + stream0_128i_out[2*i+3] = _mm_unpackhi_epi32(xmm1,xmm3); +#elif defined(__arm__) + +#endif + + } + +#if defined(__x86_64__) || defined(__i386__) + _mm_empty(); + _m_empty(); +#endif +} + +int nr_dlsch_16qam_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, + int32_t **rxdataF_comp, + int32_t **rxdataF_comp_i, + int32_t **dl_ch_mag, //|h_0|^2*(2/sqrt{10}) + int32_t **dl_ch_mag_i, //|h_1|^2*(2/sqrt{10}) + int32_t **rho_i, + int16_t *dlsch_llr, + uint8_t symbol, + uint8_t first_symbol_flag, + uint16_t nb_rb, + uint16_t pbch_pss_sss_adjust, + int16_t **llr16p) +{ + + int16_t *rxF = (int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rxF_i = (int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag = (int16_t*)&dl_ch_mag[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag_i = (int16_t*)&dl_ch_mag_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rho = (int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *llr16; + int len; + uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol; + + // first symbol has different structure due to more pilots + if (first_symbol_flag == 1) { + llr16 = (int16_t*)dlsch_llr; + } else { + llr16 = (int16_t*)(*llr16p); + } + + + AssertFatal(llr16!=NULL,"nr_dlsch_16qam_64qam_llr:llr is null, symbol %d\n",symbol); + + + if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) { + // if symbol has pilots + if (frame_parms->nb_antenna_ports_eNB!=1) + // in 2 antenna ports we have 8 REs per symbol per RB + len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3); + else + // for 1 antenna port we have 10 REs per symbol per RB + len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6); + } else { + // symbol has no pilots + len = (nb_rb*12) - pbch_pss_sss_adjust; + } + + // printf("symbol %d: qam16_llr, len %d (llr16 %p)\n",symbol,len,llr16); + + qam16_qam64((short *)rxF, + (short *)rxF_i, + (short *)ch_mag, + (short *)ch_mag_i, + (short *)llr16, + (short *)rho, + len); + + llr16 += (len<<2); + *llr16p = (short *)llr16; + + return(0); +} + +//---------------------------------------------------------------------------------------------- +// 64-QAM +//---------------------------------------------------------------------------------------------- + +/* +__m128i ONE_OVER_SQRT_42 __attribute__((aligned(16))); +__m128i THREE_OVER_SQRT_42 __attribute__((aligned(16))); +__m128i FIVE_OVER_SQRT_42 __attribute__((aligned(16))); +__m128i SEVEN_OVER_SQRT_42 __attribute__((aligned(16))); + +__m128i FORTYNINE_OVER_FOUR_SQRT_42 __attribute__((aligned(16))); +__m128i THIRTYSEVEN_OVER_FOUR_SQRT_42 __attribute__((aligned(16))); +__m128i TWENTYNINE_OVER_FOUR_SQRT_42 __attribute__((aligned(16))); +__m128i TWENTYFIVE_OVER_FOUR_SQRT_42 __attribute__((aligned(16))); +__m128i SEVENTEEN_OVER_FOUR_SQRT_42 __attribute__((aligned(16))); +__m128i NINE_OVER_FOUR_SQRT_42 __attribute__((aligned(16))); +__m128i THIRTEEN_OVER_FOUR_SQRT_42 __attribute__((aligned(16))); +__m128i FIVE_OVER_FOUR_SQRT_42 __attribute__((aligned(16))); +__m128i ONE_OVER_FOUR_SQRT_42 __attribute__((aligned(16))); + +__m128i y0r_one_over_sqrt_21 __attribute__((aligned(16))); +__m128i y0r_three_over_sqrt_21 __attribute__((aligned(16))); +__m128i y0r_five_over_sqrt_21 __attribute__((aligned(16))); +__m128i y0r_seven_over_sqrt_21 __attribute__((aligned(16))); +__m128i y0i_one_over_sqrt_21 __attribute__((aligned(16))); +__m128i y0i_three_over_sqrt_21 __attribute__((aligned(16))); +__m128i y0i_five_over_sqrt_21 __attribute__((aligned(16))); +__m128i y0i_seven_over_sqrt_21 __attribute__((aligned(16))); + +__m128i ch_mag_98_over_42_with_sigma2 __attribute__((aligned(16))); +__m128i ch_mag_74_over_42_with_sigma2 __attribute__((aligned(16))); +__m128i ch_mag_58_over_42_with_sigma2 __attribute__((aligned(16))); +__m128i ch_mag_50_over_42_with_sigma2 __attribute__((aligned(16))); +__m128i ch_mag_34_over_42_with_sigma2 __attribute__((aligned(16))); +__m128i ch_mag_18_over_42_with_sigma2 __attribute__((aligned(16))); +__m128i ch_mag_26_over_42_with_sigma2 __attribute__((aligned(16))); +__m128i ch_mag_10_over_42_with_sigma2 __attribute__((aligned(16))); +__m128i ch_mag_2_over_42_with_sigma2 __attribute__((aligned(16))); + +*/ + +void qam64_qpsk(int16_t *stream0_in, + int16_t *stream1_in, + int16_t *ch_mag, + int16_t *stream0_out, + int16_t *rho01, + int32_t length + ) +{ + + /* + Author: S. Wagner + Date: 31-07-12 + + Input: + stream0_in: MF filter for 1st stream, i.e., y0=h0'*y + stream1_in: MF filter for 2nd stream, i.e., y1=h1'*y + ch_mag: 4*h0/sqrt(42), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + ch_mag_i: 4*h1/sqrt(42), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + rho01: Channel cross correlation, i.e., h1'*h0 + + Output: + stream0_out: output LLRs for 1st stream + */ + +#if defined(__x86_64__) || defined(__i386__) + __m128i *rho01_128i = (__m128i *)rho01; + __m128i *stream0_128i_in = (__m128i *)stream0_in; + __m128i *stream1_128i_in = (__m128i *)stream1_in; + __m128i *ch_mag_128i = (__m128i *)ch_mag; + + + __m128i ONE_OVER_SQRT_42 = _mm_set1_epi16(10112); // round(1/sqrt(42)*2^16) + __m128i THREE_OVER_SQRT_42 = _mm_set1_epi16(30337); // round(3/sqrt(42)*2^16) + __m128i FIVE_OVER_SQRT_42 = _mm_set1_epi16(25281); // round(5/sqrt(42)*2^15) + __m128i SEVEN_OVER_SQRT_42 = _mm_set1_epi16(17697); // round(5/sqrt(42)*2^15) + __m128i ONE_OVER_SQRT_2 = _mm_set1_epi16(23170); // round(1/sqrt(2)*2^15) + __m128i FORTYNINE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(30969); // round(49/(4*sqrt(42))*2^14), Q2.14 + __m128i THIRTYSEVEN_OVER_FOUR_SQRT_42 = _mm_set1_epi16(23385); // round(37/(4*sqrt(42))*2^14), Q2.14 + __m128i TWENTYFIVE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(31601); // round(25/(4*sqrt(42))*2^15) + __m128i TWENTYNINE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(18329); // round(29/(4*sqrt(42))*2^15), Q2.14 + __m128i SEVENTEEN_OVER_FOUR_SQRT_42 = _mm_set1_epi16(21489); // round(17/(4*sqrt(42))*2^15) + __m128i NINE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(11376); // round(9/(4*sqrt(42))*2^15) + __m128i THIRTEEN_OVER_FOUR_SQRT_42 = _mm_set1_epi16(16433); // round(13/(4*sqrt(42))*2^15) + __m128i FIVE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(6320); // round(5/(4*sqrt(42))*2^15) + __m128i ONE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(1264); // round(1/(4*sqrt(42))*2^15) + + + __m128i ch_mag_des; + __m128i ch_mag_98_over_42_with_sigma2; + __m128i ch_mag_74_over_42_with_sigma2; + __m128i ch_mag_58_over_42_with_sigma2; + __m128i ch_mag_50_over_42_with_sigma2; + __m128i ch_mag_34_over_42_with_sigma2; + __m128i ch_mag_18_over_42_with_sigma2; + __m128i ch_mag_26_over_42_with_sigma2; + __m128i ch_mag_10_over_42_with_sigma2; + __m128i ch_mag_2_over_42_with_sigma2; + __m128i y0r_one_over_sqrt_21; + __m128i y0r_three_over_sqrt_21; + __m128i y0r_five_over_sqrt_21; + __m128i y0r_seven_over_sqrt_21; + __m128i y0i_one_over_sqrt_21; + __m128i y0i_three_over_sqrt_21; + __m128i y0i_five_over_sqrt_21; + __m128i y0i_seven_over_sqrt_21; +#elif defined(__arm__) + +#endif + + int i,j; + + for (i=0; i>2; i+=2) { + +#if defined(__x86_64) || defined(__i386__) + // Get rho + xmm0 = rho01_128i[i]; + xmm1 = rho01_128i[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho) + xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho) + rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho) + rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho) + + // Compute the different rhos + rho_rpi_1_1 = _mm_mulhi_epi16(rho_rpi, ONE_OVER_SQRT_42); + rho_rmi_1_1 = _mm_mulhi_epi16(rho_rmi, ONE_OVER_SQRT_42); + rho_rpi_3_3 = _mm_mulhi_epi16(rho_rpi, THREE_OVER_SQRT_42); + rho_rmi_3_3 = _mm_mulhi_epi16(rho_rmi, THREE_OVER_SQRT_42); + rho_rpi_5_5 = _mm_mulhi_epi16(rho_rpi, FIVE_OVER_SQRT_42); + rho_rmi_5_5 = _mm_mulhi_epi16(rho_rmi, FIVE_OVER_SQRT_42); + rho_rpi_7_7 = _mm_mulhi_epi16(rho_rpi, SEVEN_OVER_SQRT_42); + rho_rmi_7_7 = _mm_mulhi_epi16(rho_rmi, SEVEN_OVER_SQRT_42); + + rho_rpi_5_5 = _mm_slli_epi16(rho_rpi_5_5, 1); + rho_rmi_5_5 = _mm_slli_epi16(rho_rmi_5_5, 1); + rho_rpi_7_7 = _mm_slli_epi16(rho_rpi_7_7, 2); + rho_rmi_7_7 = _mm_slli_epi16(rho_rmi_7_7, 2); + + xmm4 = _mm_mulhi_epi16(xmm2, ONE_OVER_SQRT_42); + xmm5 = _mm_mulhi_epi16(xmm3, ONE_OVER_SQRT_42); + xmm6 = _mm_mulhi_epi16(xmm3, THREE_OVER_SQRT_42); + xmm7 = _mm_mulhi_epi16(xmm3, FIVE_OVER_SQRT_42); + xmm8 = _mm_mulhi_epi16(xmm3, SEVEN_OVER_SQRT_42); + xmm7 = _mm_slli_epi16(xmm7, 1); + xmm8 = _mm_slli_epi16(xmm8, 2); + + rho_rpi_1_3 = _mm_adds_epi16(xmm4, xmm6); + rho_rmi_1_3 = _mm_subs_epi16(xmm4, xmm6); + rho_rpi_1_5 = _mm_adds_epi16(xmm4, xmm7); + rho_rmi_1_5 = _mm_subs_epi16(xmm4, xmm7); + rho_rpi_1_7 = _mm_adds_epi16(xmm4, xmm8); + rho_rmi_1_7 = _mm_subs_epi16(xmm4, xmm8); + + xmm4 = _mm_mulhi_epi16(xmm2, THREE_OVER_SQRT_42); + rho_rpi_3_1 = _mm_adds_epi16(xmm4, xmm5); + rho_rmi_3_1 = _mm_subs_epi16(xmm4, xmm5); + rho_rpi_3_5 = _mm_adds_epi16(xmm4, xmm7); + rho_rmi_3_5 = _mm_subs_epi16(xmm4, xmm7); + rho_rpi_3_7 = _mm_adds_epi16(xmm4, xmm8); + rho_rmi_3_7 = _mm_subs_epi16(xmm4, xmm8); + + xmm4 = _mm_mulhi_epi16(xmm2, FIVE_OVER_SQRT_42); + xmm4 = _mm_slli_epi16(xmm4, 1); + rho_rpi_5_1 = _mm_adds_epi16(xmm4, xmm5); + rho_rmi_5_1 = _mm_subs_epi16(xmm4, xmm5); + rho_rpi_5_3 = _mm_adds_epi16(xmm4, xmm6); + rho_rmi_5_3 = _mm_subs_epi16(xmm4, xmm6); + rho_rpi_5_7 = _mm_adds_epi16(xmm4, xmm8); + rho_rmi_5_7 = _mm_subs_epi16(xmm4, xmm8); + + xmm4 = _mm_mulhi_epi16(xmm2, SEVEN_OVER_SQRT_42); + xmm4 = _mm_slli_epi16(xmm4, 2); + rho_rpi_7_1 = _mm_adds_epi16(xmm4, xmm5); + rho_rmi_7_1 = _mm_subs_epi16(xmm4, xmm5); + rho_rpi_7_3 = _mm_adds_epi16(xmm4, xmm6); + rho_rmi_7_3 = _mm_subs_epi16(xmm4, xmm6); + rho_rpi_7_5 = _mm_adds_epi16(xmm4, xmm7); + rho_rmi_7_5 = _mm_subs_epi16(xmm4, xmm7); + + // Rearrange interfering MF output + xmm0 = stream1_128i_in[i]; + xmm1 = stream1_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)] + y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)] + + // Psi_r calculation from rho_rpi or rho_rmi + xmm0 = _mm_setzero_si128(); // ZERO for abs_pi16 + xmm2 = _mm_subs_epi16(rho_rpi_7_7, y1r); + psi_r_p7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_5, y1r); + psi_r_p7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_3, y1r); + psi_r_p7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_1, y1r); + psi_r_p7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_1, y1r); + psi_r_p7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_3, y1r); + psi_r_p7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_5, y1r); + psi_r_p7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_7, y1r); + psi_r_p7_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_7, y1r); + psi_r_p5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_5, y1r); + psi_r_p5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_3, y1r); + psi_r_p5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_1, y1r); + psi_r_p5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_1, y1r); + psi_r_p5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_3, y1r); + psi_r_p5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_5, y1r); + psi_r_p5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_7, y1r); + psi_r_p5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_7, y1r); + psi_r_p3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_5, y1r); + psi_r_p3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_3, y1r); + psi_r_p3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_1, y1r); + psi_r_p3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_1, y1r); + psi_r_p3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_3, y1r); + psi_r_p3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_5, y1r); + psi_r_p3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_7, y1r); + psi_r_p3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_7, y1r); + psi_r_p1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_5, y1r); + psi_r_p1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_3, y1r); + psi_r_p1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_1, y1r); + psi_r_p1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_1, y1r); + psi_r_p1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_3, y1r); + psi_r_p1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_5, y1r); + psi_r_p1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_7, y1r); + psi_r_p1_m7 = _mm_abs_epi16(xmm2); + + xmm2 = _mm_adds_epi16(rho_rmi_1_7, y1r); + psi_r_m1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_5, y1r); + psi_r_m1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_3, y1r); + psi_r_m1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_1, y1r); + psi_r_m1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_1, y1r); + psi_r_m1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_3, y1r); + psi_r_m1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_5, y1r); + psi_r_m1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_7, y1r); + psi_r_m1_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_7, y1r); + psi_r_m3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_5, y1r); + psi_r_m3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_3, y1r); + psi_r_m3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_1, y1r); + psi_r_m3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_1, y1r); + psi_r_m3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_3, y1r); + psi_r_m3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_5, y1r); + psi_r_m3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_7, y1r); + psi_r_m3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_7, y1r); + psi_r_m5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_5, y1r); + psi_r_m5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_3, y1r); + psi_r_m5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_1, y1r); + psi_r_m5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_1, y1r); + psi_r_m5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_3, y1r); + psi_r_m5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_5, y1r); + psi_r_m5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_7, y1r); + psi_r_m5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_7, y1r); + psi_r_m7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_5, y1r); + psi_r_m7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_3, y1r); + psi_r_m7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_1, y1r); + psi_r_m7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_1, y1r); + psi_r_m7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_3, y1r); + psi_r_m7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_5, y1r); + psi_r_m7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_7, y1r); + psi_r_m7_m7 = _mm_abs_epi16(xmm2); + + // Psi_i calculation from rho_rpi or rho_rmi + xmm2 = _mm_subs_epi16(rho_rmi_7_7, y1i); + psi_i_p7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_7, y1i); + psi_i_p7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_7, y1i); + psi_i_p7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_7, y1i); + psi_i_p7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_7, y1i); + psi_i_p7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_7, y1i); + psi_i_p7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_7, y1i); + psi_i_p7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_7, y1i); + psi_i_p7_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_5, y1i); + psi_i_p5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_5, y1i); + psi_i_p5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_5, y1i); + psi_i_p5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_5, y1i); + psi_i_p5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_5, y1i); + psi_i_p5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_5, y1i); + psi_i_p5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_5, y1i); + psi_i_p5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_5, y1i); + psi_i_p5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_3, y1i); + psi_i_p3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_3, y1i); + psi_i_p3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_3, y1i); + psi_i_p3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_3, y1i); + psi_i_p3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_3, y1i); + psi_i_p3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_3, y1i); + psi_i_p3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_3, y1i); + psi_i_p3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_3, y1i); + psi_i_p3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_1, y1i); + psi_i_p1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_1, y1i); + psi_i_p1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_1, y1i); + psi_i_p1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_1, y1i); + psi_i_p1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_1, y1i); + psi_i_p1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_1, y1i); + psi_i_p1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_1, y1i); + psi_i_p1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_1, y1i); + psi_i_p1_m7 = _mm_abs_epi16(xmm2); + + xmm2 = _mm_subs_epi16(rho_rpi_7_1, y1i); + psi_i_m1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_1, y1i); + psi_i_m1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_1, y1i); + psi_i_m1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_1, y1i); + psi_i_m1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_1, y1i); + psi_i_m1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_1, y1i); + psi_i_m1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_1, y1i); + psi_i_m1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_1, y1i); + psi_i_m1_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_3, y1i); + psi_i_m3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_3, y1i); + psi_i_m3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_3, y1i); + psi_i_m3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_3, y1i); + psi_i_m3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_3, y1i); + psi_i_m3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_3, y1i); + psi_i_m3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_3, y1i); + psi_i_m3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_3, y1i); + psi_i_m3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_5, y1i); + psi_i_m5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_5, y1i); + psi_i_m5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_5, y1i); + psi_i_m5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_5, y1i); + psi_i_m5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_5, y1i); + psi_i_m5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_5, y1i); + psi_i_m5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_5, y1i); + psi_i_m5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_5, y1i); + psi_i_m5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_7, y1i); + psi_i_m7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_7, y1i); + psi_i_m7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_7, y1i); + psi_i_m7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_7, y1i); + psi_i_m7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_7, y1i); + psi_i_m7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_7, y1i); + psi_i_m7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_7, y1i); + psi_i_m7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_7, y1i); + psi_i_m7_m7 = _mm_abs_epi16(xmm2); + + + // Rearrange desired MF output + xmm0 = stream0_128i_in[i]; + xmm1 = stream0_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)] + y0i = _mm_unpackhi_epi64(xmm0,xmm1); + + // Rearrange desired channel magnitudes + xmm2 = ch_mag_128i[i]; // = [|h|^2(1),|h|^2(1),|h|^2(2),|h|^2(2)]*(2/sqrt(10)) + xmm3 = ch_mag_128i[i+1]; // = [|h|^2(3),|h|^2(3),|h|^2(4),|h|^2(4)]*(2/sqrt(10)) + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + ch_mag_des = _mm_unpacklo_epi64(xmm2,xmm3); + + y0r_one_over_sqrt_21 = _mm_mulhi_epi16(y0r, ONE_OVER_SQRT_42); + y0r_three_over_sqrt_21 = _mm_mulhi_epi16(y0r, THREE_OVER_SQRT_42); + y0r_five_over_sqrt_21 = _mm_mulhi_epi16(y0r, FIVE_OVER_SQRT_42); + y0r_five_over_sqrt_21 = _mm_slli_epi16(y0r_five_over_sqrt_21, 1); + y0r_seven_over_sqrt_21 = _mm_mulhi_epi16(y0r, SEVEN_OVER_SQRT_42); + y0r_seven_over_sqrt_21 = _mm_slli_epi16(y0r_seven_over_sqrt_21, 2); // Q2.14 + + y0i_one_over_sqrt_21 = _mm_mulhi_epi16(y0i, ONE_OVER_SQRT_42); + y0i_three_over_sqrt_21 = _mm_mulhi_epi16(y0i, THREE_OVER_SQRT_42); + y0i_five_over_sqrt_21 = _mm_mulhi_epi16(y0i, FIVE_OVER_SQRT_42); + y0i_five_over_sqrt_21 = _mm_slli_epi16(y0i_five_over_sqrt_21, 1); + y0i_seven_over_sqrt_21 = _mm_mulhi_epi16(y0i, SEVEN_OVER_SQRT_42); + y0i_seven_over_sqrt_21 = _mm_slli_epi16(y0i_seven_over_sqrt_21, 2); // Q2.14 + + y0_p_7_1 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_7_3 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_7_5 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_7_7 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_p_5_1 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_5_3 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_5_5 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_5_7 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_p_3_1 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_3_3 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_3_5 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_3_7 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_p_1_1 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_1_3 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_1_5 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_1_7 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_seven_over_sqrt_21); + + y0_m_1_1 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_1_3 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_1_5 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_1_7 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_m_3_1 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_3_3 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_3_5 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_3_7 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_m_5_1 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_5_3 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_5_5 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_5_7 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_m_7_1 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_7_3 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_7_5 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_7_7 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_seven_over_sqrt_21); + + // divide by sqrt(2) + psi_r_p7_p7 = _mm_mulhi_epi16(psi_r_p7_p7, ONE_OVER_SQRT_2); + psi_r_p7_p7 = _mm_slli_epi16(psi_r_p7_p7, 1); + psi_r_p7_p5 = _mm_mulhi_epi16(psi_r_p7_p5, ONE_OVER_SQRT_2); + psi_r_p7_p5 = _mm_slli_epi16(psi_r_p7_p5, 1); + psi_r_p7_p3 = _mm_mulhi_epi16(psi_r_p7_p3, ONE_OVER_SQRT_2); + psi_r_p7_p3 = _mm_slli_epi16(psi_r_p7_p3, 1); + psi_r_p7_p1 = _mm_mulhi_epi16(psi_r_p7_p1, ONE_OVER_SQRT_2); + psi_r_p7_p1 = _mm_slli_epi16(psi_r_p7_p1, 1); + psi_r_p7_m1 = _mm_mulhi_epi16(psi_r_p7_m1, ONE_OVER_SQRT_2); + psi_r_p7_m1 = _mm_slli_epi16(psi_r_p7_m1, 1); + psi_r_p7_m3 = _mm_mulhi_epi16(psi_r_p7_m3, ONE_OVER_SQRT_2); + psi_r_p7_m3 = _mm_slli_epi16(psi_r_p7_m3, 1); + psi_r_p7_m5 = _mm_mulhi_epi16(psi_r_p7_m5, ONE_OVER_SQRT_2); + psi_r_p7_m5 = _mm_slli_epi16(psi_r_p7_m5, 1); + psi_r_p7_m7 = _mm_mulhi_epi16(psi_r_p7_m7, ONE_OVER_SQRT_2); + psi_r_p7_m7 = _mm_slli_epi16(psi_r_p7_m7, 1); + psi_r_p5_p7 = _mm_mulhi_epi16(psi_r_p5_p7, ONE_OVER_SQRT_2); + psi_r_p5_p7 = _mm_slli_epi16(psi_r_p5_p7, 1); + psi_r_p5_p5 = _mm_mulhi_epi16(psi_r_p5_p5, ONE_OVER_SQRT_2); + psi_r_p5_p5 = _mm_slli_epi16(psi_r_p5_p5, 1); + psi_r_p5_p3 = _mm_mulhi_epi16(psi_r_p5_p3, ONE_OVER_SQRT_2); + psi_r_p5_p3 = _mm_slli_epi16(psi_r_p5_p3, 1); + psi_r_p5_p1 = _mm_mulhi_epi16(psi_r_p5_p1, ONE_OVER_SQRT_2); + psi_r_p5_p1 = _mm_slli_epi16(psi_r_p5_p1, 1); + psi_r_p5_m1 = _mm_mulhi_epi16(psi_r_p5_m1, ONE_OVER_SQRT_2); + psi_r_p5_m1 = _mm_slli_epi16(psi_r_p5_m1, 1); + psi_r_p5_m3 = _mm_mulhi_epi16(psi_r_p5_m3, ONE_OVER_SQRT_2); + psi_r_p5_m3 = _mm_slli_epi16(psi_r_p5_m3, 1); + psi_r_p5_m5 = _mm_mulhi_epi16(psi_r_p5_m5, ONE_OVER_SQRT_2); + psi_r_p5_m5 = _mm_slli_epi16(psi_r_p5_m5, 1); + psi_r_p5_m7 = _mm_mulhi_epi16(psi_r_p5_m7, ONE_OVER_SQRT_2); + psi_r_p5_m7 = _mm_slli_epi16(psi_r_p5_m7, 1); + psi_r_p3_p7 = _mm_mulhi_epi16(psi_r_p3_p7, ONE_OVER_SQRT_2); + psi_r_p3_p7 = _mm_slli_epi16(psi_r_p3_p7, 1); + psi_r_p3_p5 = _mm_mulhi_epi16(psi_r_p3_p5, ONE_OVER_SQRT_2); + psi_r_p3_p5 = _mm_slli_epi16(psi_r_p3_p5, 1); + psi_r_p3_p3 = _mm_mulhi_epi16(psi_r_p3_p3, ONE_OVER_SQRT_2); + psi_r_p3_p3 = _mm_slli_epi16(psi_r_p3_p3, 1); + psi_r_p3_p1 = _mm_mulhi_epi16(psi_r_p3_p1, ONE_OVER_SQRT_2); + psi_r_p3_p1 = _mm_slli_epi16(psi_r_p3_p1, 1); + psi_r_p3_m1 = _mm_mulhi_epi16(psi_r_p3_m1, ONE_OVER_SQRT_2); + psi_r_p3_m1 = _mm_slli_epi16(psi_r_p3_m1, 1); + psi_r_p3_m3 = _mm_mulhi_epi16(psi_r_p3_m3, ONE_OVER_SQRT_2); + psi_r_p3_m3 = _mm_slli_epi16(psi_r_p3_m3, 1); + psi_r_p3_m5 = _mm_mulhi_epi16(psi_r_p3_m5, ONE_OVER_SQRT_2); + psi_r_p3_m5 = _mm_slli_epi16(psi_r_p3_m5, 1); + psi_r_p3_m7 = _mm_mulhi_epi16(psi_r_p3_m7, ONE_OVER_SQRT_2); + psi_r_p3_m7 = _mm_slli_epi16(psi_r_p3_m7, 1); + psi_r_p1_p7 = _mm_mulhi_epi16(psi_r_p1_p7, ONE_OVER_SQRT_2); + psi_r_p1_p7 = _mm_slli_epi16(psi_r_p1_p7, 1); + psi_r_p1_p5 = _mm_mulhi_epi16(psi_r_p1_p5, ONE_OVER_SQRT_2); + psi_r_p1_p5 = _mm_slli_epi16(psi_r_p1_p5, 1); + psi_r_p1_p3 = _mm_mulhi_epi16(psi_r_p1_p3, ONE_OVER_SQRT_2); + psi_r_p1_p3 = _mm_slli_epi16(psi_r_p1_p3, 1); + psi_r_p1_p1 = _mm_mulhi_epi16(psi_r_p1_p1, ONE_OVER_SQRT_2); + psi_r_p1_p1 = _mm_slli_epi16(psi_r_p1_p1, 1); + psi_r_p1_m1 = _mm_mulhi_epi16(psi_r_p1_m1, ONE_OVER_SQRT_2); + psi_r_p1_m1 = _mm_slli_epi16(psi_r_p1_m1, 1); + psi_r_p1_m3 = _mm_mulhi_epi16(psi_r_p1_m3, ONE_OVER_SQRT_2); + psi_r_p1_m3 = _mm_slli_epi16(psi_r_p1_m3, 1); + psi_r_p1_m5 = _mm_mulhi_epi16(psi_r_p1_m5, ONE_OVER_SQRT_2); + psi_r_p1_m5 = _mm_slli_epi16(psi_r_p1_m5, 1); + psi_r_p1_m7 = _mm_mulhi_epi16(psi_r_p1_m7, ONE_OVER_SQRT_2); + psi_r_p1_m7 = _mm_slli_epi16(psi_r_p1_m7, 1); + psi_r_m1_p7 = _mm_mulhi_epi16(psi_r_m1_p7, ONE_OVER_SQRT_2); + psi_r_m1_p7 = _mm_slli_epi16(psi_r_m1_p7, 1); + psi_r_m1_p5 = _mm_mulhi_epi16(psi_r_m1_p5, ONE_OVER_SQRT_2); + psi_r_m1_p5 = _mm_slli_epi16(psi_r_m1_p5, 1); + psi_r_m1_p3 = _mm_mulhi_epi16(psi_r_m1_p3, ONE_OVER_SQRT_2); + psi_r_m1_p3 = _mm_slli_epi16(psi_r_m1_p3, 1); + psi_r_m1_p1 = _mm_mulhi_epi16(psi_r_m1_p1, ONE_OVER_SQRT_2); + psi_r_m1_p1 = _mm_slli_epi16(psi_r_m1_p1, 1); + psi_r_m1_m1 = _mm_mulhi_epi16(psi_r_m1_m1, ONE_OVER_SQRT_2); + psi_r_m1_m1 = _mm_slli_epi16(psi_r_m1_m1, 1); + psi_r_m1_m3 = _mm_mulhi_epi16(psi_r_m1_m3, ONE_OVER_SQRT_2); + psi_r_m1_m3 = _mm_slli_epi16(psi_r_m1_m3, 1); + psi_r_m1_m5 = _mm_mulhi_epi16(psi_r_m1_m5, ONE_OVER_SQRT_2); + psi_r_m1_m5 = _mm_slli_epi16(psi_r_m1_m5, 1); + psi_r_m1_m7 = _mm_mulhi_epi16(psi_r_m1_m7, ONE_OVER_SQRT_2); + psi_r_m1_m7 = _mm_slli_epi16(psi_r_m1_m7, 1); + psi_r_m3_p7 = _mm_mulhi_epi16(psi_r_m3_p7, ONE_OVER_SQRT_2); + psi_r_m3_p7 = _mm_slli_epi16(psi_r_m3_p7, 1); + psi_r_m3_p5 = _mm_mulhi_epi16(psi_r_m3_p5, ONE_OVER_SQRT_2); + psi_r_m3_p5 = _mm_slli_epi16(psi_r_m3_p5, 1); + psi_r_m3_p3 = _mm_mulhi_epi16(psi_r_m3_p3, ONE_OVER_SQRT_2); + psi_r_m3_p3 = _mm_slli_epi16(psi_r_m3_p3, 1); + psi_r_m3_p1 = _mm_mulhi_epi16(psi_r_m3_p1, ONE_OVER_SQRT_2); + psi_r_m3_p1 = _mm_slli_epi16(psi_r_m3_p1, 1); + psi_r_m3_m1 = _mm_mulhi_epi16(psi_r_m3_m1, ONE_OVER_SQRT_2); + psi_r_m3_m1 = _mm_slli_epi16(psi_r_m3_m1, 1); + psi_r_m3_m3 = _mm_mulhi_epi16(psi_r_m3_m3, ONE_OVER_SQRT_2); + psi_r_m3_m3 = _mm_slli_epi16(psi_r_m3_m3, 1); + psi_r_m3_m5 = _mm_mulhi_epi16(psi_r_m3_m5, ONE_OVER_SQRT_2); + psi_r_m3_m5 = _mm_slli_epi16(psi_r_m3_m5, 1); + psi_r_m3_m7 = _mm_mulhi_epi16(psi_r_m3_m7, ONE_OVER_SQRT_2); + psi_r_m3_m7 = _mm_slli_epi16(psi_r_m3_m7, 1); + psi_r_m5_p7 = _mm_mulhi_epi16(psi_r_m5_p7, ONE_OVER_SQRT_2); + psi_r_m5_p7 = _mm_slli_epi16(psi_r_m5_p7, 1); + psi_r_m5_p5 = _mm_mulhi_epi16(psi_r_m5_p5, ONE_OVER_SQRT_2); + psi_r_m5_p5 = _mm_slli_epi16(psi_r_m5_p5, 1); + psi_r_m5_p3 = _mm_mulhi_epi16(psi_r_m5_p3, ONE_OVER_SQRT_2); + psi_r_m5_p3 = _mm_slli_epi16(psi_r_m5_p3, 1); + psi_r_m5_p1 = _mm_mulhi_epi16(psi_r_m5_p1, ONE_OVER_SQRT_2); + psi_r_m5_p1 = _mm_slli_epi16(psi_r_m5_p1, 1); + psi_r_m5_m1 = _mm_mulhi_epi16(psi_r_m5_m1, ONE_OVER_SQRT_2); + psi_r_m5_m1 = _mm_slli_epi16(psi_r_m5_m1, 1); + psi_r_m5_m3 = _mm_mulhi_epi16(psi_r_m5_m3, ONE_OVER_SQRT_2); + psi_r_m5_m3 = _mm_slli_epi16(psi_r_m5_m3, 1); + psi_r_m5_m5 = _mm_mulhi_epi16(psi_r_m5_m5, ONE_OVER_SQRT_2); + psi_r_m5_m5 = _mm_slli_epi16(psi_r_m5_m5, 1); + psi_r_m5_m7 = _mm_mulhi_epi16(psi_r_m5_m7, ONE_OVER_SQRT_2); + psi_r_m5_m7 = _mm_slli_epi16(psi_r_m5_m7, 1); + psi_r_m7_p7 = _mm_mulhi_epi16(psi_r_m7_p7, ONE_OVER_SQRT_2); + psi_r_m7_p7 = _mm_slli_epi16(psi_r_m7_p7, 1); + psi_r_m7_p5 = _mm_mulhi_epi16(psi_r_m7_p5, ONE_OVER_SQRT_2); + psi_r_m7_p5 = _mm_slli_epi16(psi_r_m7_p5, 1); + psi_r_m7_p3 = _mm_mulhi_epi16(psi_r_m7_p3, ONE_OVER_SQRT_2); + psi_r_m7_p3 = _mm_slli_epi16(psi_r_m7_p3, 1); + psi_r_m7_p1 = _mm_mulhi_epi16(psi_r_m7_p1, ONE_OVER_SQRT_2); + psi_r_m7_p1 = _mm_slli_epi16(psi_r_m7_p1, 1); + psi_r_m7_m1 = _mm_mulhi_epi16(psi_r_m7_m1, ONE_OVER_SQRT_2); + psi_r_m7_m1 = _mm_slli_epi16(psi_r_m7_m1, 1); + psi_r_m7_m3 = _mm_mulhi_epi16(psi_r_m7_m3, ONE_OVER_SQRT_2); + psi_r_m7_m3 = _mm_slli_epi16(psi_r_m7_m3, 1); + psi_r_m7_m5 = _mm_mulhi_epi16(psi_r_m7_m5, ONE_OVER_SQRT_2); + psi_r_m7_m5 = _mm_slli_epi16(psi_r_m7_m5, 1); + psi_r_m7_m7 = _mm_mulhi_epi16(psi_r_m7_m7, ONE_OVER_SQRT_2); + psi_r_m7_m7 = _mm_slli_epi16(psi_r_m7_m7, 1); + + psi_i_p7_p7 = _mm_mulhi_epi16(psi_i_p7_p7, ONE_OVER_SQRT_2); + psi_i_p7_p7 = _mm_slli_epi16(psi_i_p7_p7, 1); + psi_i_p7_p5 = _mm_mulhi_epi16(psi_i_p7_p5, ONE_OVER_SQRT_2); + psi_i_p7_p5 = _mm_slli_epi16(psi_i_p7_p5, 1); + psi_i_p7_p3 = _mm_mulhi_epi16(psi_i_p7_p3, ONE_OVER_SQRT_2); + psi_i_p7_p3 = _mm_slli_epi16(psi_i_p7_p3, 1); + psi_i_p7_p1 = _mm_mulhi_epi16(psi_i_p7_p1, ONE_OVER_SQRT_2); + psi_i_p7_p1 = _mm_slli_epi16(psi_i_p7_p1, 1); + psi_i_p7_m1 = _mm_mulhi_epi16(psi_i_p7_m1, ONE_OVER_SQRT_2); + psi_i_p7_m1 = _mm_slli_epi16(psi_i_p7_m1, 1); + psi_i_p7_m3 = _mm_mulhi_epi16(psi_i_p7_m3, ONE_OVER_SQRT_2); + psi_i_p7_m3 = _mm_slli_epi16(psi_i_p7_m3, 1); + psi_i_p7_m5 = _mm_mulhi_epi16(psi_i_p7_m5, ONE_OVER_SQRT_2); + psi_i_p7_m5 = _mm_slli_epi16(psi_i_p7_m5, 1); + psi_i_p7_m7 = _mm_mulhi_epi16(psi_i_p7_m7, ONE_OVER_SQRT_2); + psi_i_p7_m7 = _mm_slli_epi16(psi_i_p7_m7, 1); + psi_i_p5_p7 = _mm_mulhi_epi16(psi_i_p5_p7, ONE_OVER_SQRT_2); + psi_i_p5_p7 = _mm_slli_epi16(psi_i_p5_p7, 1); + psi_i_p5_p5 = _mm_mulhi_epi16(psi_i_p5_p5, ONE_OVER_SQRT_2); + psi_i_p5_p5 = _mm_slli_epi16(psi_i_p5_p5, 1); + psi_i_p5_p3 = _mm_mulhi_epi16(psi_i_p5_p3, ONE_OVER_SQRT_2); + psi_i_p5_p3 = _mm_slli_epi16(psi_i_p5_p3, 1); + psi_i_p5_p1 = _mm_mulhi_epi16(psi_i_p5_p1, ONE_OVER_SQRT_2); + psi_i_p5_p1 = _mm_slli_epi16(psi_i_p5_p1, 1); + psi_i_p5_m1 = _mm_mulhi_epi16(psi_i_p5_m1, ONE_OVER_SQRT_2); + psi_i_p5_m1 = _mm_slli_epi16(psi_i_p5_m1, 1); + psi_i_p5_m3 = _mm_mulhi_epi16(psi_i_p5_m3, ONE_OVER_SQRT_2); + psi_i_p5_m3 = _mm_slli_epi16(psi_i_p5_m3, 1); + psi_i_p5_m5 = _mm_mulhi_epi16(psi_i_p5_m5, ONE_OVER_SQRT_2); + psi_i_p5_m5 = _mm_slli_epi16(psi_i_p5_m5, 1); + psi_i_p5_m7 = _mm_mulhi_epi16(psi_i_p5_m7, ONE_OVER_SQRT_2); + psi_i_p5_m7 = _mm_slli_epi16(psi_i_p5_m7, 1); + psi_i_p3_p7 = _mm_mulhi_epi16(psi_i_p3_p7, ONE_OVER_SQRT_2); + psi_i_p3_p7 = _mm_slli_epi16(psi_i_p3_p7, 1); + psi_i_p3_p5 = _mm_mulhi_epi16(psi_i_p3_p5, ONE_OVER_SQRT_2); + psi_i_p3_p5 = _mm_slli_epi16(psi_i_p3_p5, 1); + psi_i_p3_p3 = _mm_mulhi_epi16(psi_i_p3_p3, ONE_OVER_SQRT_2); + psi_i_p3_p3 = _mm_slli_epi16(psi_i_p3_p3, 1); + psi_i_p3_p1 = _mm_mulhi_epi16(psi_i_p3_p1, ONE_OVER_SQRT_2); + psi_i_p3_p1 = _mm_slli_epi16(psi_i_p3_p1, 1); + psi_i_p3_m1 = _mm_mulhi_epi16(psi_i_p3_m1, ONE_OVER_SQRT_2); + psi_i_p3_m1 = _mm_slli_epi16(psi_i_p3_m1, 1); + psi_i_p3_m3 = _mm_mulhi_epi16(psi_i_p3_m3, ONE_OVER_SQRT_2); + psi_i_p3_m3 = _mm_slli_epi16(psi_i_p3_m3, 1); + psi_i_p3_m5 = _mm_mulhi_epi16(psi_i_p3_m5, ONE_OVER_SQRT_2); + psi_i_p3_m5 = _mm_slli_epi16(psi_i_p3_m5, 1); + psi_i_p3_m7 = _mm_mulhi_epi16(psi_i_p3_m7, ONE_OVER_SQRT_2); + psi_i_p3_m7 = _mm_slli_epi16(psi_i_p3_m7, 1); + psi_i_p1_p7 = _mm_mulhi_epi16(psi_i_p1_p7, ONE_OVER_SQRT_2); + psi_i_p1_p7 = _mm_slli_epi16(psi_i_p1_p7, 1); + psi_i_p1_p5 = _mm_mulhi_epi16(psi_i_p1_p5, ONE_OVER_SQRT_2); + psi_i_p1_p5 = _mm_slli_epi16(psi_i_p1_p5, 1); + psi_i_p1_p3 = _mm_mulhi_epi16(psi_i_p1_p3, ONE_OVER_SQRT_2); + psi_i_p1_p3 = _mm_slli_epi16(psi_i_p1_p3, 1); + psi_i_p1_p1 = _mm_mulhi_epi16(psi_i_p1_p1, ONE_OVER_SQRT_2); + psi_i_p1_p1 = _mm_slli_epi16(psi_i_p1_p1, 1); + psi_i_p1_m1 = _mm_mulhi_epi16(psi_i_p1_m1, ONE_OVER_SQRT_2); + psi_i_p1_m1 = _mm_slli_epi16(psi_i_p1_m1, 1); + psi_i_p1_m3 = _mm_mulhi_epi16(psi_i_p1_m3, ONE_OVER_SQRT_2); + psi_i_p1_m3 = _mm_slli_epi16(psi_i_p1_m3, 1); + psi_i_p1_m5 = _mm_mulhi_epi16(psi_i_p1_m5, ONE_OVER_SQRT_2); + psi_i_p1_m5 = _mm_slli_epi16(psi_i_p1_m5, 1); + psi_i_p1_m7 = _mm_mulhi_epi16(psi_i_p1_m7, ONE_OVER_SQRT_2); + psi_i_p1_m7 = _mm_slli_epi16(psi_i_p1_m7, 1); + psi_i_m1_p7 = _mm_mulhi_epi16(psi_i_m1_p7, ONE_OVER_SQRT_2); + psi_i_m1_p7 = _mm_slli_epi16(psi_i_m1_p7, 1); + psi_i_m1_p5 = _mm_mulhi_epi16(psi_i_m1_p5, ONE_OVER_SQRT_2); + psi_i_m1_p5 = _mm_slli_epi16(psi_i_m1_p5, 1); + psi_i_m1_p3 = _mm_mulhi_epi16(psi_i_m1_p3, ONE_OVER_SQRT_2); + psi_i_m1_p3 = _mm_slli_epi16(psi_i_m1_p3, 1); + psi_i_m1_p1 = _mm_mulhi_epi16(psi_i_m1_p1, ONE_OVER_SQRT_2); + psi_i_m1_p1 = _mm_slli_epi16(psi_i_m1_p1, 1); + psi_i_m1_m1 = _mm_mulhi_epi16(psi_i_m1_m1, ONE_OVER_SQRT_2); + psi_i_m1_m1 = _mm_slli_epi16(psi_i_m1_m1, 1); + psi_i_m1_m3 = _mm_mulhi_epi16(psi_i_m1_m3, ONE_OVER_SQRT_2); + psi_i_m1_m3 = _mm_slli_epi16(psi_i_m1_m3, 1); + psi_i_m1_m5 = _mm_mulhi_epi16(psi_i_m1_m5, ONE_OVER_SQRT_2); + psi_i_m1_m5 = _mm_slli_epi16(psi_i_m1_m5, 1); + psi_i_m1_m7 = _mm_mulhi_epi16(psi_i_m1_m7, ONE_OVER_SQRT_2); + psi_i_m1_m7 = _mm_slli_epi16(psi_i_m1_m7, 1); + psi_i_m3_p7 = _mm_mulhi_epi16(psi_i_m3_p7, ONE_OVER_SQRT_2); + psi_i_m3_p7 = _mm_slli_epi16(psi_i_m3_p7, 1); + psi_i_m3_p5 = _mm_mulhi_epi16(psi_i_m3_p5, ONE_OVER_SQRT_2); + psi_i_m3_p5 = _mm_slli_epi16(psi_i_m3_p5, 1); + psi_i_m3_p3 = _mm_mulhi_epi16(psi_i_m3_p3, ONE_OVER_SQRT_2); + psi_i_m3_p3 = _mm_slli_epi16(psi_i_m3_p3, 1); + psi_i_m3_p1 = _mm_mulhi_epi16(psi_i_m3_p1, ONE_OVER_SQRT_2); + psi_i_m3_p1 = _mm_slli_epi16(psi_i_m3_p1, 1); + psi_i_m3_m1 = _mm_mulhi_epi16(psi_i_m3_m1, ONE_OVER_SQRT_2); + psi_i_m3_m1 = _mm_slli_epi16(psi_i_m3_m1, 1); + psi_i_m3_m3 = _mm_mulhi_epi16(psi_i_m3_m3, ONE_OVER_SQRT_2); + psi_i_m3_m3 = _mm_slli_epi16(psi_i_m3_m3, 1); + psi_i_m3_m5 = _mm_mulhi_epi16(psi_i_m3_m5, ONE_OVER_SQRT_2); + psi_i_m3_m5 = _mm_slli_epi16(psi_i_m3_m5, 1); + psi_i_m3_m7 = _mm_mulhi_epi16(psi_i_m3_m7, ONE_OVER_SQRT_2); + psi_i_m3_m7 = _mm_slli_epi16(psi_i_m3_m7, 1); + psi_i_m5_p7 = _mm_mulhi_epi16(psi_i_m5_p7, ONE_OVER_SQRT_2); + psi_i_m5_p7 = _mm_slli_epi16(psi_i_m5_p7, 1); + psi_i_m5_p5 = _mm_mulhi_epi16(psi_i_m5_p5, ONE_OVER_SQRT_2); + psi_i_m5_p5 = _mm_slli_epi16(psi_i_m5_p5, 1); + psi_i_m5_p3 = _mm_mulhi_epi16(psi_i_m5_p3, ONE_OVER_SQRT_2); + psi_i_m5_p3 = _mm_slli_epi16(psi_i_m5_p3, 1); + psi_i_m5_p1 = _mm_mulhi_epi16(psi_i_m5_p1, ONE_OVER_SQRT_2); + psi_i_m5_p1 = _mm_slli_epi16(psi_i_m5_p1, 1); + psi_i_m5_m1 = _mm_mulhi_epi16(psi_i_m5_m1, ONE_OVER_SQRT_2); + psi_i_m5_m1 = _mm_slli_epi16(psi_i_m5_m1, 1); + psi_i_m5_m3 = _mm_mulhi_epi16(psi_i_m5_m3, ONE_OVER_SQRT_2); + psi_i_m5_m3 = _mm_slli_epi16(psi_i_m5_m3, 1); + psi_i_m5_m5 = _mm_mulhi_epi16(psi_i_m5_m5, ONE_OVER_SQRT_2); + psi_i_m5_m5 = _mm_slli_epi16(psi_i_m5_m5, 1); + psi_i_m5_m7 = _mm_mulhi_epi16(psi_i_m5_m7, ONE_OVER_SQRT_2); + psi_i_m5_m7 = _mm_slli_epi16(psi_i_m5_m7, 1); + psi_i_m7_p7 = _mm_mulhi_epi16(psi_i_m7_p7, ONE_OVER_SQRT_2); + psi_i_m7_p7 = _mm_slli_epi16(psi_i_m7_p7, 1); + psi_i_m7_p5 = _mm_mulhi_epi16(psi_i_m7_p5, ONE_OVER_SQRT_2); + psi_i_m7_p5 = _mm_slli_epi16(psi_i_m7_p5, 1); + psi_i_m7_p3 = _mm_mulhi_epi16(psi_i_m7_p3, ONE_OVER_SQRT_2); + psi_i_m7_p3 = _mm_slli_epi16(psi_i_m7_p3, 1); + psi_i_m7_p1 = _mm_mulhi_epi16(psi_i_m7_p1, ONE_OVER_SQRT_2); + psi_i_m7_p1 = _mm_slli_epi16(psi_i_m7_p1, 1); + psi_i_m7_m1 = _mm_mulhi_epi16(psi_i_m7_m1, ONE_OVER_SQRT_2); + psi_i_m7_m1 = _mm_slli_epi16(psi_i_m7_m1, 1); + psi_i_m7_m3 = _mm_mulhi_epi16(psi_i_m7_m3, ONE_OVER_SQRT_2); + psi_i_m7_m3 = _mm_slli_epi16(psi_i_m7_m3, 1); + psi_i_m7_m5 = _mm_mulhi_epi16(psi_i_m7_m5, ONE_OVER_SQRT_2); + psi_i_m7_m5 = _mm_slli_epi16(psi_i_m7_m5, 1); + psi_i_m7_m7 = _mm_mulhi_epi16(psi_i_m7_m7, ONE_OVER_SQRT_2); + psi_i_m7_m7 = _mm_slli_epi16(psi_i_m7_m7, 1); + + psi_a_p7_p7 = _mm_adds_epi16(psi_r_p7_p7, psi_i_p7_p7); + psi_a_p7_p5 = _mm_adds_epi16(psi_r_p7_p5, psi_i_p7_p5); + psi_a_p7_p3 = _mm_adds_epi16(psi_r_p7_p3, psi_i_p7_p3); + psi_a_p7_p1 = _mm_adds_epi16(psi_r_p7_p1, psi_i_p7_p1); + psi_a_p7_m1 = _mm_adds_epi16(psi_r_p7_m1, psi_i_p7_m1); + psi_a_p7_m3 = _mm_adds_epi16(psi_r_p7_m3, psi_i_p7_m3); + psi_a_p7_m5 = _mm_adds_epi16(psi_r_p7_m5, psi_i_p7_m5); + psi_a_p7_m7 = _mm_adds_epi16(psi_r_p7_m7, psi_i_p7_m7); + psi_a_p5_p7 = _mm_adds_epi16(psi_r_p5_p7, psi_i_p5_p7); + psi_a_p5_p5 = _mm_adds_epi16(psi_r_p5_p5, psi_i_p5_p5); + psi_a_p5_p3 = _mm_adds_epi16(psi_r_p5_p3, psi_i_p5_p3); + psi_a_p5_p1 = _mm_adds_epi16(psi_r_p5_p1, psi_i_p5_p1); + psi_a_p5_m1 = _mm_adds_epi16(psi_r_p5_m1, psi_i_p5_m1); + psi_a_p5_m3 = _mm_adds_epi16(psi_r_p5_m3, psi_i_p5_m3); + psi_a_p5_m5 = _mm_adds_epi16(psi_r_p5_m5, psi_i_p5_m5); + psi_a_p5_m7 = _mm_adds_epi16(psi_r_p5_m7, psi_i_p5_m7); + psi_a_p3_p7 = _mm_adds_epi16(psi_r_p3_p7, psi_i_p3_p7); + psi_a_p3_p5 = _mm_adds_epi16(psi_r_p3_p5, psi_i_p3_p5); + psi_a_p3_p3 = _mm_adds_epi16(psi_r_p3_p3, psi_i_p3_p3); + psi_a_p3_p1 = _mm_adds_epi16(psi_r_p3_p1, psi_i_p3_p1); + psi_a_p3_m1 = _mm_adds_epi16(psi_r_p3_m1, psi_i_p3_m1); + psi_a_p3_m3 = _mm_adds_epi16(psi_r_p3_m3, psi_i_p3_m3); + psi_a_p3_m5 = _mm_adds_epi16(psi_r_p3_m5, psi_i_p3_m5); + psi_a_p3_m7 = _mm_adds_epi16(psi_r_p3_m7, psi_i_p3_m7); + psi_a_p1_p7 = _mm_adds_epi16(psi_r_p1_p7, psi_i_p1_p7); + psi_a_p1_p5 = _mm_adds_epi16(psi_r_p1_p5, psi_i_p1_p5); + psi_a_p1_p3 = _mm_adds_epi16(psi_r_p1_p3, psi_i_p1_p3); + psi_a_p1_p1 = _mm_adds_epi16(psi_r_p1_p1, psi_i_p1_p1); + psi_a_p1_m1 = _mm_adds_epi16(psi_r_p1_m1, psi_i_p1_m1); + psi_a_p1_m3 = _mm_adds_epi16(psi_r_p1_m3, psi_i_p1_m3); + psi_a_p1_m5 = _mm_adds_epi16(psi_r_p1_m5, psi_i_p1_m5); + psi_a_p1_m7 = _mm_adds_epi16(psi_r_p1_m7, psi_i_p1_m7); + psi_a_m1_p7 = _mm_adds_epi16(psi_r_m1_p7, psi_i_m1_p7); + psi_a_m1_p5 = _mm_adds_epi16(psi_r_m1_p5, psi_i_m1_p5); + psi_a_m1_p3 = _mm_adds_epi16(psi_r_m1_p3, psi_i_m1_p3); + psi_a_m1_p1 = _mm_adds_epi16(psi_r_m1_p1, psi_i_m1_p1); + psi_a_m1_m1 = _mm_adds_epi16(psi_r_m1_m1, psi_i_m1_m1); + psi_a_m1_m3 = _mm_adds_epi16(psi_r_m1_m3, psi_i_m1_m3); + psi_a_m1_m5 = _mm_adds_epi16(psi_r_m1_m5, psi_i_m1_m5); + psi_a_m1_m7 = _mm_adds_epi16(psi_r_m1_m7, psi_i_m1_m7); + psi_a_m3_p7 = _mm_adds_epi16(psi_r_m3_p7, psi_i_m3_p7); + psi_a_m3_p5 = _mm_adds_epi16(psi_r_m3_p5, psi_i_m3_p5); + psi_a_m3_p3 = _mm_adds_epi16(psi_r_m3_p3, psi_i_m3_p3); + psi_a_m3_p1 = _mm_adds_epi16(psi_r_m3_p1, psi_i_m3_p1); + psi_a_m3_m1 = _mm_adds_epi16(psi_r_m3_m1, psi_i_m3_m1); + psi_a_m3_m3 = _mm_adds_epi16(psi_r_m3_m3, psi_i_m3_m3); + psi_a_m3_m5 = _mm_adds_epi16(psi_r_m3_m5, psi_i_m3_m5); + psi_a_m3_m7 = _mm_adds_epi16(psi_r_m3_m7, psi_i_m3_m7); + psi_a_m5_p7 = _mm_adds_epi16(psi_r_m5_p7, psi_i_m5_p7); + psi_a_m5_p5 = _mm_adds_epi16(psi_r_m5_p5, psi_i_m5_p5); + psi_a_m5_p3 = _mm_adds_epi16(psi_r_m5_p3, psi_i_m5_p3); + psi_a_m5_p1 = _mm_adds_epi16(psi_r_m5_p1, psi_i_m5_p1); + psi_a_m5_m1 = _mm_adds_epi16(psi_r_m5_m1, psi_i_m5_m1); + psi_a_m5_m3 = _mm_adds_epi16(psi_r_m5_m3, psi_i_m5_m3); + psi_a_m5_m5 = _mm_adds_epi16(psi_r_m5_m5, psi_i_m5_m5); + psi_a_m5_m7 = _mm_adds_epi16(psi_r_m5_m7, psi_i_m5_m7); + psi_a_m7_p7 = _mm_adds_epi16(psi_r_m7_p7, psi_i_m7_p7); + psi_a_m7_p5 = _mm_adds_epi16(psi_r_m7_p5, psi_i_m7_p5); + psi_a_m7_p3 = _mm_adds_epi16(psi_r_m7_p3, psi_i_m7_p3); + psi_a_m7_p1 = _mm_adds_epi16(psi_r_m7_p1, psi_i_m7_p1); + psi_a_m7_m1 = _mm_adds_epi16(psi_r_m7_m1, psi_i_m7_m1); + psi_a_m7_m3 = _mm_adds_epi16(psi_r_m7_m3, psi_i_m7_m3); + psi_a_m7_m5 = _mm_adds_epi16(psi_r_m7_m5, psi_i_m7_m5); + psi_a_m7_m7 = _mm_adds_epi16(psi_r_m7_m7, psi_i_m7_m7); + + // Computing different multiples of ||h0||^2 + // x=1, y=1 + ch_mag_2_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,ONE_OVER_FOUR_SQRT_42); + ch_mag_2_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_2_over_42_with_sigma2,1); + // x=1, y=3 + ch_mag_10_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,FIVE_OVER_FOUR_SQRT_42); + ch_mag_10_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_10_over_42_with_sigma2,1); + // x=1, x=5 + ch_mag_26_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,THIRTEEN_OVER_FOUR_SQRT_42); + ch_mag_26_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_26_over_42_with_sigma2,1); + // x=1, y=7 + ch_mag_50_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,TWENTYFIVE_OVER_FOUR_SQRT_42); + ch_mag_50_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_50_over_42_with_sigma2,1); + // x=3, y=3 + ch_mag_18_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,NINE_OVER_FOUR_SQRT_42); + ch_mag_18_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_18_over_42_with_sigma2,1); + // x=3, y=5 + ch_mag_34_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,SEVENTEEN_OVER_FOUR_SQRT_42); + ch_mag_34_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_34_over_42_with_sigma2,1); + // x=3, y=7 + ch_mag_58_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,TWENTYNINE_OVER_FOUR_SQRT_42); + ch_mag_58_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_58_over_42_with_sigma2,2); + // x=5, y=5 + ch_mag_50_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,TWENTYFIVE_OVER_FOUR_SQRT_42); + ch_mag_50_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_50_over_42_with_sigma2,1); + // x=5, y=7 + ch_mag_74_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,THIRTYSEVEN_OVER_FOUR_SQRT_42); + ch_mag_74_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_74_over_42_with_sigma2,2); + // x=7, y=7 + ch_mag_98_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,FORTYNINE_OVER_FOUR_SQRT_42); + ch_mag_98_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_98_over_42_with_sigma2,2); + + // Computing Metrics + xmm1 = _mm_adds_epi16(psi_a_p7_p7, y0_p_7_7); + bit_met_p7_p7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p7_p5, y0_p_7_5); + bit_met_p7_p5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p7_p3, y0_p_7_3); + bit_met_p7_p3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p7_p1, y0_p_7_1); + bit_met_p7_p1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p7_m1, y0_m_7_1); + bit_met_p7_m1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p7_m3, y0_m_7_3); + bit_met_p7_m3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p7_m5, y0_m_7_5); + bit_met_p7_m5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p7_m7, y0_m_7_7); + bit_met_p7_m7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p5_p7, y0_p_5_7); + bit_met_p5_p7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p5_p5, y0_p_5_5); + bit_met_p5_p5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p5_p3, y0_p_5_3); + bit_met_p5_p3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p5_p1, y0_p_5_1); + bit_met_p5_p1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p5_m1, y0_m_5_1); + bit_met_p5_m1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p5_m3, y0_m_5_3); + bit_met_p5_m3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p5_m5, y0_m_5_5); + bit_met_p5_m5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p5_m7, y0_m_5_7); + bit_met_p5_m7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p3_p7, y0_p_3_7); + bit_met_p3_p7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p3_p5, y0_p_3_5); + bit_met_p3_p5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p3_p3, y0_p_3_3); + bit_met_p3_p3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p3_p1, y0_p_3_1); + bit_met_p3_p1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p3_m1, y0_m_3_1); + bit_met_p3_m1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p3_m3, y0_m_3_3); + bit_met_p3_m3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p3_m5, y0_m_3_5); + bit_met_p3_m5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p3_m7, y0_m_3_7); + bit_met_p3_m7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p1_p7, y0_p_1_7); + bit_met_p1_p7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p1_p5, y0_p_1_5); + bit_met_p1_p5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p1_p3, y0_p_1_3); + bit_met_p1_p3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p1_p1, y0_p_1_1); + bit_met_p1_p1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p1_m1, y0_m_1_1); + bit_met_p1_m1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p1_m3, y0_m_1_3); + bit_met_p1_m3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p1_m5, y0_m_1_5); + bit_met_p1_m5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm1 = _mm_adds_epi16(psi_a_p1_m7, y0_m_1_7); + bit_met_p1_m7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + + xmm1 = _mm_subs_epi16(psi_a_m1_p7, y0_m_1_7); + bit_met_m1_p7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m1_p5, y0_m_1_5); + bit_met_m1_p5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m1_p3, y0_m_1_3); + bit_met_m1_p3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m1_p1, y0_m_1_1); + bit_met_m1_p1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m1_m1, y0_p_1_1); + bit_met_m1_m1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m1_m3, y0_p_1_3); + bit_met_m1_m3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m1_m5, y0_p_1_5); + bit_met_m1_m5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m1_m7, y0_p_1_7); + bit_met_m1_m7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m3_p7, y0_m_3_7); + bit_met_m3_p7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m3_p5, y0_m_3_5); + bit_met_m3_p5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m3_p3, y0_m_3_3); + bit_met_m3_p3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m3_p1, y0_m_3_1); + bit_met_m3_p1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m3_m1, y0_p_3_1); + bit_met_m3_m1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m3_m3, y0_p_3_3); + bit_met_m3_m3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m3_m5, y0_p_3_5); + bit_met_m3_m5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m3_m7, y0_p_3_7); + bit_met_m3_m7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m5_p7, y0_m_5_7); + bit_met_m5_p7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m5_p5, y0_m_5_5); + bit_met_m5_p5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m5_p3, y0_m_5_3); + bit_met_m5_p3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m5_p1, y0_m_5_1); + bit_met_m5_p1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m5_m1, y0_p_5_1); + bit_met_m5_m1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m5_m3, y0_p_5_3); + bit_met_m5_m3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m5_m5, y0_p_5_5); + bit_met_m5_m5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m5_m7, y0_p_5_7); + bit_met_m5_m7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m7_p7, y0_m_7_7); + bit_met_m7_p7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m7_p5, y0_m_7_5); + bit_met_m7_p5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m7_p3, y0_m_7_3); + bit_met_m7_p3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m7_p1, y0_m_7_1); + bit_met_m7_p1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m7_m1, y0_p_7_1); + bit_met_m7_m1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m7_m3, y0_p_7_3); + bit_met_m7_m3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m7_m5, y0_p_7_5); + bit_met_m7_m5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm1 = _mm_subs_epi16(psi_a_m7_m7, y0_p_7_7); + bit_met_m7_m7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + + // Detection for 1st bit (LTE mapping) + // bit = 1 + xmm0 = _mm_max_epi16(bit_met_m7_p7, bit_met_m7_p5); + xmm1 = _mm_max_epi16(bit_met_m7_p3, bit_met_m7_p1); + xmm2 = _mm_max_epi16(bit_met_m7_m1, bit_met_m7_m3); + xmm3 = _mm_max_epi16(bit_met_m7_m5, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m5_p7, bit_met_m5_p5); + xmm1 = _mm_max_epi16(bit_met_m5_p3, bit_met_m5_p1); + xmm2 = _mm_max_epi16(bit_met_m5_m1, bit_met_m5_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m5_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_m3_p7, bit_met_m3_p5); + xmm1 = _mm_max_epi16(bit_met_m3_p3, bit_met_m3_p1); + xmm2 = _mm_max_epi16(bit_met_m3_m1, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m3_m5, bit_met_m3_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_m1_p7, bit_met_m1_p5); + xmm1 = _mm_max_epi16(bit_met_m1_p3, bit_met_m1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m1_m3); + xmm3 = _mm_max_epi16(bit_met_m1_m5, bit_met_m1_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + // bit = 0 + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p7_p5); + xmm1 = _mm_max_epi16(bit_met_p7_p3, bit_met_p7_p1); + xmm2 = _mm_max_epi16(bit_met_p7_m1, bit_met_p7_m3); + xmm3 = _mm_max_epi16(bit_met_p7_m5, bit_met_p7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p5_p7, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p5_p3, bit_met_p5_p1); + xmm2 = _mm_max_epi16(bit_met_p5_m1, bit_met_p5_m3); + xmm3 = _mm_max_epi16(bit_met_p5_m5, bit_met_p5_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p3_p7, bit_met_p3_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p3_p1); + xmm2 = _mm_max_epi16(bit_met_p3_m1, bit_met_p3_m3); + xmm3 = _mm_max_epi16(bit_met_p3_m5, bit_met_p3_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p1_p7, bit_met_p1_p5); + xmm1 = _mm_max_epi16(bit_met_p1_p3, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_p1_m1, bit_met_p1_m3); + xmm3 = _mm_max_epi16(bit_met_p1_m5, bit_met_p1_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y0r = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 2nd bit (LTE mapping) + // bit = 1 + xmm0 = _mm_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = _mm_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = _mm_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = _mm_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = _mm_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = _mm_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = _mm_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + // bit = 0 + xmm0 = _mm_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = _mm_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = _mm_max_epi16(bit_met_m5_p1, bit_met_m7_p1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = _mm_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = _mm_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m7_p5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = _mm_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = _mm_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = _mm_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y1r = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 3rd bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_m7_m7, bit_met_m7_m5); + xmm1 = _mm_max_epi16(bit_met_m7_m3, bit_met_m7_m1); + xmm2 = _mm_max_epi16(bit_met_m7_p1, bit_met_m7_p3); + xmm3 = _mm_max_epi16(bit_met_m7_p5, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m5_m7, bit_met_m5_m5); + xmm1 = _mm_max_epi16(bit_met_m5_m3, bit_met_m5_m1); + xmm2 = _mm_max_epi16(bit_met_m5_p1, bit_met_m5_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p5_m7, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p5_m3, bit_met_p5_m1); + xmm2 = _mm_max_epi16(bit_met_p5_p1, bit_met_p5_p3); + xmm3 = _mm_max_epi16(bit_met_p5_p5, bit_met_p5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p7_m5); + xmm1 = _mm_max_epi16(bit_met_p7_m3, bit_met_p7_m1); + xmm2 = _mm_max_epi16(bit_met_p7_p1, bit_met_p7_p3); + xmm3 = _mm_max_epi16(bit_met_p7_p5, bit_met_p7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_m3_m7, bit_met_m3_m5); + xmm1 = _mm_max_epi16(bit_met_m3_m3, bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_m3_p1, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m3_p5, bit_met_m3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m1_m7, bit_met_m1_m5); + xmm1 = _mm_max_epi16(bit_met_m1_m3, bit_met_m1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m1_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p5, bit_met_m1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p1_m7, bit_met_p1_m5); + xmm1 = _mm_max_epi16(bit_met_p1_m3, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_p1_p1, bit_met_p1_p3); + xmm3 = _mm_max_epi16(bit_met_p1_p5, bit_met_p1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p3_m7, bit_met_p3_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p3_m1); + xmm2 = _mm_max_epi16(bit_met_p3_p1, bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_p5, bit_met_p3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y2r = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 4th bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = _mm_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = _mm_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = _mm_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = _mm_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m5_p5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = _mm_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = _mm_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = _mm_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = _mm_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = _mm_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = _mm_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = _mm_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = _mm_max_epi16(bit_met_m5_p1, bit_met_m7_p1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = _mm_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y0i = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + + // Detection for 5th bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_m7_m7, bit_met_m7_m5); + xmm1 = _mm_max_epi16(bit_met_m7_m3, bit_met_m7_m1); + xmm2 = _mm_max_epi16(bit_met_m7_p1, bit_met_m7_p3); + xmm3 = _mm_max_epi16(bit_met_m7_p5, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m1_m7, bit_met_m1_m5); + xmm1 = _mm_max_epi16(bit_met_m1_m3, bit_met_m1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m1_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p5, bit_met_m1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p1_m7, bit_met_p1_m5); + xmm1 = _mm_max_epi16(bit_met_p1_m3, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_p1_p1, bit_met_p1_p3); + xmm3 = _mm_max_epi16(bit_met_p1_p5, bit_met_p1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p7_m5); + xmm1 = _mm_max_epi16(bit_met_p7_m3, bit_met_p7_m1); + xmm2 = _mm_max_epi16(bit_met_p7_p1, bit_met_p7_p3); + xmm3 = _mm_max_epi16(bit_met_p7_p5, bit_met_p7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_m5_m7, bit_met_m5_m5); + xmm1 = _mm_max_epi16(bit_met_m5_m3, bit_met_m5_m1); + xmm2 = _mm_max_epi16(bit_met_m5_p1, bit_met_m5_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m3_m7, bit_met_m3_m5); + xmm1 = _mm_max_epi16(bit_met_m3_m3, bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_m3_p1, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m3_p5, bit_met_m3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p3_m7, bit_met_p3_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p3_m1); + xmm2 = _mm_max_epi16(bit_met_p3_p1, bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_p5, bit_met_p3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p5_m7, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p5_m3, bit_met_p5_m1); + xmm2 = _mm_max_epi16(bit_met_p5_p1, bit_met_p5_p3); + xmm3 = _mm_max_epi16(bit_met_p5_p5, bit_met_p5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y1i = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 6th bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = _mm_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = _mm_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = _mm_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = _mm_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = _mm_max_epi16(bit_met_m5_p1, bit_met_m5_p1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = _mm_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = _mm_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = _mm_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = _mm_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = _mm_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = _mm_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = _mm_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = _mm_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m7_p5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y2i = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + + // map to output stream, difficult to do in SIMD since we have 6 16bit LLRs + // RE 1 + j = 24*i; + stream0_out[j + 0] = ((short *)&y0r)[0]; + stream0_out[j + 1] = ((short *)&y1r)[0]; + stream0_out[j + 2] = ((short *)&y2r)[0]; + stream0_out[j + 3] = ((short *)&y0i)[0]; + stream0_out[j + 4] = ((short *)&y1i)[0]; + stream0_out[j + 5] = ((short *)&y2i)[0]; + // RE 2 + stream0_out[j + 6] = ((short *)&y0r)[1]; + stream0_out[j + 7] = ((short *)&y1r)[1]; + stream0_out[j + 8] = ((short *)&y2r)[1]; + stream0_out[j + 9] = ((short *)&y0i)[1]; + stream0_out[j + 10] = ((short *)&y1i)[1]; + stream0_out[j + 11] = ((short *)&y2i)[1]; + // RE 3 + stream0_out[j + 12] = ((short *)&y0r)[2]; + stream0_out[j + 13] = ((short *)&y1r)[2]; + stream0_out[j + 14] = ((short *)&y2r)[2]; + stream0_out[j + 15] = ((short *)&y0i)[2]; + stream0_out[j + 16] = ((short *)&y1i)[2]; + stream0_out[j + 17] = ((short *)&y2i)[2]; + // RE 4 + stream0_out[j + 18] = ((short *)&y0r)[3]; + stream0_out[j + 19] = ((short *)&y1r)[3]; + stream0_out[j + 20] = ((short *)&y2r)[3]; + stream0_out[j + 21] = ((short *)&y0i)[3]; + stream0_out[j + 22] = ((short *)&y1i)[3]; + stream0_out[j + 23] = ((short *)&y2i)[3]; + // RE 5 + stream0_out[j + 24] = ((short *)&y0r)[4]; + stream0_out[j + 25] = ((short *)&y1r)[4]; + stream0_out[j + 26] = ((short *)&y2r)[4]; + stream0_out[j + 27] = ((short *)&y0i)[4]; + stream0_out[j + 28] = ((short *)&y1i)[4]; + stream0_out[j + 29] = ((short *)&y2i)[4]; + // RE 6 + stream0_out[j + 30] = ((short *)&y0r)[5]; + stream0_out[j + 31] = ((short *)&y1r)[5]; + stream0_out[j + 32] = ((short *)&y2r)[5]; + stream0_out[j + 33] = ((short *)&y0i)[5]; + stream0_out[j + 34] = ((short *)&y1i)[5]; + stream0_out[j + 35] = ((short *)&y2i)[5]; + // RE 7 + stream0_out[j + 36] = ((short *)&y0r)[6]; + stream0_out[j + 37] = ((short *)&y1r)[6]; + stream0_out[j + 38] = ((short *)&y2r)[6]; + stream0_out[j + 39] = ((short *)&y0i)[6]; + stream0_out[j + 40] = ((short *)&y1i)[6]; + stream0_out[j + 41] = ((short *)&y2i)[6]; + // RE 8 + stream0_out[j + 42] = ((short *)&y0r)[7]; + stream0_out[j + 43] = ((short *)&y1r)[7]; + stream0_out[j + 44] = ((short *)&y2r)[7]; + stream0_out[j + 45] = ((short *)&y0i)[7]; + stream0_out[j + 46] = ((short *)&y1i)[7]; + stream0_out[j + 47] = ((short *)&y2i)[7]; +#elif defined(__arm__) + +#endif + } + +#if defined(__x86_64__) || defined(__i386__) + _mm_empty(); + _m_empty(); +#endif +} + + +int nr_dlsch_64qam_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, + int32_t **rxdataF_comp, + int32_t **rxdataF_comp_i, + int32_t **dl_ch_mag, + int32_t **rho_i, + int16_t *dlsch_llr, + uint8_t symbol, + uint8_t first_symbol_flag, + uint16_t nb_rb, + uint16_t pbch_pss_sss_adjust, + int16_t **llr16p) +{ + + int16_t *rxF = (int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rxF_i = (int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag = (int16_t*)&dl_ch_mag[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rho = (int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *llr16; + int len; + uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol; + + //first symbol has different structure due to more pilots + if (first_symbol_flag == 1) { + llr16 = (int16_t*)dlsch_llr; + } else { + llr16 = (int16_t*)(*llr16p); + } + + AssertFatal(llr16!=NULL,"nr_dlsch_16qam_64qam_llr:llr is null, symbol %d\n",symbol); + + if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) { + // if symbol has pilots + if (frame_parms->nb_antenna_ports_eNB!=1) + // in 2 antenna ports we have 8 REs per symbol per RB + len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3); + else + // for 1 antenna port we have 10 REs per symbol per RB + len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6); + } else { + // symbol has no pilots + len = (nb_rb*12) - pbch_pss_sss_adjust; + } + + qam64_qpsk((short *)rxF, + (short *)rxF_i, + (short *)ch_mag, + (short *)llr16, + (short *)rho, + len); + + llr16 += (6*len); + *llr16p = (short *)llr16; + return(0); +} + + + +void qam64_qam16(short *stream0_in, + short *stream1_in, + short *ch_mag, + short *ch_mag_i, + short *stream0_out, + short *rho01, + int length + ) +{ + + /* + Author: S. Wagner + Date: 31-07-12 + + Input: + stream0_in: MF filter for 1st stream, i.e., y0=h0'*y + stream1_in: MF filter for 2nd stream, i.e., y1=h1'*y + ch_mag: 4*h0/sqrt(42), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + ch_mag_i: 4*h1/sqrt(42), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + rho01: Channel cross correlation, i.e., h1'*h0 + + Output: + stream0_out: output LLRs for 1st stream + */ + +#if defined(__x86_64__) || defined(__i386__) + + __m128i *rho01_128i = (__m128i *)rho01; + __m128i *stream0_128i_in = (__m128i *)stream0_in; + __m128i *stream1_128i_in = (__m128i *)stream1_in; + __m128i *ch_mag_128i = (__m128i *)ch_mag; + __m128i *ch_mag_128i_i = (__m128i *)ch_mag_i; + + __m128i ONE_OVER_SQRT_42 = _mm_set1_epi16(10112); // round(1/sqrt(42)*2^16) + __m128i THREE_OVER_SQRT_42 = _mm_set1_epi16(30337); // round(3/sqrt(42)*2^16) + __m128i FIVE_OVER_SQRT_42 = _mm_set1_epi16(25281); // round(5/sqrt(42)*2^15) + __m128i SEVEN_OVER_SQRT_42 = _mm_set1_epi16(17697); // round(5/sqrt(42)*2^15) + __m128i FORTYNINE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(30969); // round(49/(4*sqrt(42))*2^14), Q2.14 + __m128i THIRTYSEVEN_OVER_FOUR_SQRT_42 = _mm_set1_epi16(23385); // round(37/(4*sqrt(42))*2^14), Q2.14 + __m128i TWENTYFIVE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(31601); // round(25/(4*sqrt(42))*2^15) + __m128i TWENTYNINE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(18329); // round(29/(4*sqrt(42))*2^15), Q2.14 + __m128i SEVENTEEN_OVER_FOUR_SQRT_42 = _mm_set1_epi16(21489); // round(17/(4*sqrt(42))*2^15) + __m128i NINE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(11376); // round(9/(4*sqrt(42))*2^15) + __m128i THIRTEEN_OVER_FOUR_SQRT_42 = _mm_set1_epi16(16433); // round(13/(4*sqrt(42))*2^15) + __m128i FIVE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(6320); // round(5/(4*sqrt(42))*2^15) + __m128i ONE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(1264); // round(1/(4*sqrt(42))*2^15) + __m128i ONE_OVER_SQRT_10_Q15 = _mm_set1_epi16(10362); // round(1/sqrt(10)*2^15) + __m128i THREE_OVER_SQRT_10 = _mm_set1_epi16(31086); // round(3/sqrt(10)*2^15) + __m128i SQRT_10_OVER_FOUR = _mm_set1_epi16(25905); // round(sqrt(10)/4*2^15) + + + __m128i ch_mag_int; + __m128i ch_mag_des; + __m128i ch_mag_98_over_42_with_sigma2; + __m128i ch_mag_74_over_42_with_sigma2; + __m128i ch_mag_58_over_42_with_sigma2; + __m128i ch_mag_50_over_42_with_sigma2; + __m128i ch_mag_34_over_42_with_sigma2; + __m128i ch_mag_18_over_42_with_sigma2; + __m128i ch_mag_26_over_42_with_sigma2; + __m128i ch_mag_10_over_42_with_sigma2; + __m128i ch_mag_2_over_42_with_sigma2; + __m128i y0r_one_over_sqrt_21; + __m128i y0r_three_over_sqrt_21; + __m128i y0r_five_over_sqrt_21; + __m128i y0r_seven_over_sqrt_21; + __m128i y0i_one_over_sqrt_21; + __m128i y0i_three_over_sqrt_21; + __m128i y0i_five_over_sqrt_21; + __m128i y0i_seven_over_sqrt_21; + +#elif defined(__arm__) + +#endif + int i,j; + + + + for (i=0; i>2; i+=2) { + +#if defined(__x86_64__) || defined(__i386__) + // Get rho + xmm0 = rho01_128i[i]; + xmm1 = rho01_128i[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho) + xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho) + rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho) + rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho) + + // Compute the different rhos + rho_rpi_1_1 = _mm_mulhi_epi16(rho_rpi, ONE_OVER_SQRT_42); + rho_rmi_1_1 = _mm_mulhi_epi16(rho_rmi, ONE_OVER_SQRT_42); + rho_rpi_3_3 = _mm_mulhi_epi16(rho_rpi, THREE_OVER_SQRT_42); + rho_rmi_3_3 = _mm_mulhi_epi16(rho_rmi, THREE_OVER_SQRT_42); + rho_rpi_5_5 = _mm_mulhi_epi16(rho_rpi, FIVE_OVER_SQRT_42); + rho_rmi_5_5 = _mm_mulhi_epi16(rho_rmi, FIVE_OVER_SQRT_42); + rho_rpi_7_7 = _mm_mulhi_epi16(rho_rpi, SEVEN_OVER_SQRT_42); + rho_rmi_7_7 = _mm_mulhi_epi16(rho_rmi, SEVEN_OVER_SQRT_42); + + rho_rpi_5_5 = _mm_slli_epi16(rho_rpi_5_5, 1); + rho_rmi_5_5 = _mm_slli_epi16(rho_rmi_5_5, 1); + rho_rpi_7_7 = _mm_slli_epi16(rho_rpi_7_7, 2); + rho_rmi_7_7 = _mm_slli_epi16(rho_rmi_7_7, 2); + + xmm4 = _mm_mulhi_epi16(xmm2, ONE_OVER_SQRT_42); + xmm5 = _mm_mulhi_epi16(xmm3, ONE_OVER_SQRT_42); + xmm6 = _mm_mulhi_epi16(xmm3, THREE_OVER_SQRT_42); + xmm7 = _mm_mulhi_epi16(xmm3, FIVE_OVER_SQRT_42); + xmm8 = _mm_mulhi_epi16(xmm3, SEVEN_OVER_SQRT_42); + xmm7 = _mm_slli_epi16(xmm7, 1); + xmm8 = _mm_slli_epi16(xmm8, 2); + + rho_rpi_1_3 = _mm_adds_epi16(xmm4, xmm6); + rho_rmi_1_3 = _mm_subs_epi16(xmm4, xmm6); + rho_rpi_1_5 = _mm_adds_epi16(xmm4, xmm7); + rho_rmi_1_5 = _mm_subs_epi16(xmm4, xmm7); + rho_rpi_1_7 = _mm_adds_epi16(xmm4, xmm8); + rho_rmi_1_7 = _mm_subs_epi16(xmm4, xmm8); + + xmm4 = _mm_mulhi_epi16(xmm2, THREE_OVER_SQRT_42); + rho_rpi_3_1 = _mm_adds_epi16(xmm4, xmm5); + rho_rmi_3_1 = _mm_subs_epi16(xmm4, xmm5); + rho_rpi_3_5 = _mm_adds_epi16(xmm4, xmm7); + rho_rmi_3_5 = _mm_subs_epi16(xmm4, xmm7); + rho_rpi_3_7 = _mm_adds_epi16(xmm4, xmm8); + rho_rmi_3_7 = _mm_subs_epi16(xmm4, xmm8); + + xmm4 = _mm_mulhi_epi16(xmm2, FIVE_OVER_SQRT_42); + xmm4 = _mm_slli_epi16(xmm4, 1); + rho_rpi_5_1 = _mm_adds_epi16(xmm4, xmm5); + rho_rmi_5_1 = _mm_subs_epi16(xmm4, xmm5); + rho_rpi_5_3 = _mm_adds_epi16(xmm4, xmm6); + rho_rmi_5_3 = _mm_subs_epi16(xmm4, xmm6); + rho_rpi_5_7 = _mm_adds_epi16(xmm4, xmm8); + rho_rmi_5_7 = _mm_subs_epi16(xmm4, xmm8); + + xmm4 = _mm_mulhi_epi16(xmm2, SEVEN_OVER_SQRT_42); + xmm4 = _mm_slli_epi16(xmm4, 2); + rho_rpi_7_1 = _mm_adds_epi16(xmm4, xmm5); + rho_rmi_7_1 = _mm_subs_epi16(xmm4, xmm5); + rho_rpi_7_3 = _mm_adds_epi16(xmm4, xmm6); + rho_rmi_7_3 = _mm_subs_epi16(xmm4, xmm6); + rho_rpi_7_5 = _mm_adds_epi16(xmm4, xmm7); + rho_rmi_7_5 = _mm_subs_epi16(xmm4, xmm7); + + // Rearrange interfering MF output + xmm0 = stream1_128i_in[i]; + xmm1 = stream1_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)] + y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)] + + // Psi_r calculation from rho_rpi or rho_rmi + xmm0 = _mm_setzero_si128(); // ZERO for abs_pi16 + xmm2 = _mm_subs_epi16(rho_rpi_7_7, y1r); + psi_r_p7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_5, y1r); + psi_r_p7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_3, y1r); + psi_r_p7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_1, y1r); + psi_r_p7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_1, y1r); + psi_r_p7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_3, y1r); + psi_r_p7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_5, y1r); + psi_r_p7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_7, y1r); + psi_r_p7_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_7, y1r); + psi_r_p5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_5, y1r); + psi_r_p5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_3, y1r); + psi_r_p5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_1, y1r); + psi_r_p5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_1, y1r); + psi_r_p5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_3, y1r); + psi_r_p5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_5, y1r); + psi_r_p5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_7, y1r); + psi_r_p5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_7, y1r); + psi_r_p3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_5, y1r); + psi_r_p3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_3, y1r); + psi_r_p3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_1, y1r); + psi_r_p3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_1, y1r); + psi_r_p3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_3, y1r); + psi_r_p3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_5, y1r); + psi_r_p3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_7, y1r); + psi_r_p3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_7, y1r); + psi_r_p1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_5, y1r); + psi_r_p1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_3, y1r); + psi_r_p1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_1, y1r); + psi_r_p1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_1, y1r); + psi_r_p1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_3, y1r); + psi_r_p1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_5, y1r); + psi_r_p1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_7, y1r); + psi_r_p1_m7 = _mm_abs_epi16(xmm2); + + xmm2 = _mm_adds_epi16(rho_rmi_1_7, y1r); + psi_r_m1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_5, y1r); + psi_r_m1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_3, y1r); + psi_r_m1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_1, y1r); + psi_r_m1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_1, y1r); + psi_r_m1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_3, y1r); + psi_r_m1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_5, y1r); + psi_r_m1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_7, y1r); + psi_r_m1_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_7, y1r); + psi_r_m3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_5, y1r); + psi_r_m3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_3, y1r); + psi_r_m3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_1, y1r); + psi_r_m3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_1, y1r); + psi_r_m3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_3, y1r); + psi_r_m3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_5, y1r); + psi_r_m3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_7, y1r); + psi_r_m3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_7, y1r); + psi_r_m5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_5, y1r); + psi_r_m5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_3, y1r); + psi_r_m5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_1, y1r); + psi_r_m5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_1, y1r); + psi_r_m5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_3, y1r); + psi_r_m5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_5, y1r); + psi_r_m5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_7, y1r); + psi_r_m5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_7, y1r); + psi_r_m7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_5, y1r); + psi_r_m7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_3, y1r); + psi_r_m7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_1, y1r); + psi_r_m7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_1, y1r); + psi_r_m7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_3, y1r); + psi_r_m7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_5, y1r); + psi_r_m7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_7, y1r); + psi_r_m7_m7 = _mm_abs_epi16(xmm2); + + // Psi_i calculation from rho_rpi or rho_rmi + xmm2 = _mm_subs_epi16(rho_rmi_7_7, y1i); + psi_i_p7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_7, y1i); + psi_i_p7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_7, y1i); + psi_i_p7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_7, y1i); + psi_i_p7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_7, y1i); + psi_i_p7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_7, y1i); + psi_i_p7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_7, y1i); + psi_i_p7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_7, y1i); + psi_i_p7_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_5, y1i); + psi_i_p5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_5, y1i); + psi_i_p5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_5, y1i); + psi_i_p5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_5, y1i); + psi_i_p5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_5, y1i); + psi_i_p5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_5, y1i); + psi_i_p5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_5, y1i); + psi_i_p5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_5, y1i); + psi_i_p5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_3, y1i); + psi_i_p3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_3, y1i); + psi_i_p3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_3, y1i); + psi_i_p3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_3, y1i); + psi_i_p3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_3, y1i); + psi_i_p3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_3, y1i); + psi_i_p3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_3, y1i); + psi_i_p3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_3, y1i); + psi_i_p3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_1, y1i); + psi_i_p1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_1, y1i); + psi_i_p1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_1, y1i); + psi_i_p1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_1, y1i); + psi_i_p1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_1, y1i); + psi_i_p1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_1, y1i); + psi_i_p1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_1, y1i); + psi_i_p1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_1, y1i); + psi_i_p1_m7 = _mm_abs_epi16(xmm2); + + xmm2 = _mm_subs_epi16(rho_rpi_7_1, y1i); + psi_i_m1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_1, y1i); + psi_i_m1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_1, y1i); + psi_i_m1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_1, y1i); + psi_i_m1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_1, y1i); + psi_i_m1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_1, y1i); + psi_i_m1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_1, y1i); + psi_i_m1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_1, y1i); + psi_i_m1_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_3, y1i); + psi_i_m3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_3, y1i); + psi_i_m3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_3, y1i); + psi_i_m3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_3, y1i); + psi_i_m3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_3, y1i); + psi_i_m3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_3, y1i); + psi_i_m3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_3, y1i); + psi_i_m3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_3, y1i); + psi_i_m3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_5, y1i); + psi_i_m5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_5, y1i); + psi_i_m5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_5, y1i); + psi_i_m5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_5, y1i); + psi_i_m5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_5, y1i); + psi_i_m5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_5, y1i); + psi_i_m5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_5, y1i); + psi_i_m5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_5, y1i); + psi_i_m5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_7, y1i); + psi_i_m7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_7, y1i); + psi_i_m7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_7, y1i); + psi_i_m7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_7, y1i); + psi_i_m7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_7, y1i); + psi_i_m7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_7, y1i); + psi_i_m7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_7, y1i); + psi_i_m7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_7, y1i); + psi_i_m7_m7 = _mm_abs_epi16(xmm2); + + + // Rearrange desired MF output + xmm0 = stream0_128i_in[i]; + xmm1 = stream0_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)] + y0i = _mm_unpackhi_epi64(xmm0,xmm1); + + // Rearrange desired channel magnitudes + xmm2 = ch_mag_128i[i]; // = [|h|^2(1),|h|^2(1),|h|^2(2),|h|^2(2)]*(2/sqrt(10)) + xmm3 = ch_mag_128i[i+1]; // = [|h|^2(3),|h|^2(3),|h|^2(4),|h|^2(4)]*(2/sqrt(10)) + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + ch_mag_des = _mm_unpacklo_epi64(xmm2,xmm3); + + // Rearrange interfering channel magnitudes + xmm2 = ch_mag_128i_i[i]; + xmm3 = ch_mag_128i_i[i+1]; + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + ch_mag_int = _mm_unpacklo_epi64(xmm2,xmm3); + + y0r_one_over_sqrt_21 = _mm_mulhi_epi16(y0r, ONE_OVER_SQRT_42); + y0r_three_over_sqrt_21 = _mm_mulhi_epi16(y0r, THREE_OVER_SQRT_42); + y0r_five_over_sqrt_21 = _mm_mulhi_epi16(y0r, FIVE_OVER_SQRT_42); + y0r_five_over_sqrt_21 = _mm_slli_epi16(y0r_five_over_sqrt_21, 1); + y0r_seven_over_sqrt_21 = _mm_mulhi_epi16(y0r, SEVEN_OVER_SQRT_42); + y0r_seven_over_sqrt_21 = _mm_slli_epi16(y0r_seven_over_sqrt_21, 2); // Q2.14 + + y0i_one_over_sqrt_21 = _mm_mulhi_epi16(y0i, ONE_OVER_SQRT_42); + y0i_three_over_sqrt_21 = _mm_mulhi_epi16(y0i, THREE_OVER_SQRT_42); + y0i_five_over_sqrt_21 = _mm_mulhi_epi16(y0i, FIVE_OVER_SQRT_42); + y0i_five_over_sqrt_21 = _mm_slli_epi16(y0i_five_over_sqrt_21, 1); + y0i_seven_over_sqrt_21 = _mm_mulhi_epi16(y0i, SEVEN_OVER_SQRT_42); + y0i_seven_over_sqrt_21 = _mm_slli_epi16(y0i_seven_over_sqrt_21, 2); // Q2.14 + + y0_p_7_1 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_7_3 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_7_5 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_7_7 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_p_5_1 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_5_3 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_5_5 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_5_7 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_p_3_1 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_3_3 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_3_5 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_3_7 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_p_1_1 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_1_3 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_1_5 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_1_7 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_seven_over_sqrt_21); + + y0_m_1_1 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_1_3 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_1_5 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_1_7 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_m_3_1 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_3_3 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_3_5 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_3_7 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_m_5_1 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_5_3 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_5_5 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_5_7 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_m_7_1 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_7_3 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_7_5 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_7_7 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_seven_over_sqrt_21); + + interference_abs_epi16(psi_r_p7_p7, ch_mag_int, a_r_p7_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p7_p5, ch_mag_int, a_r_p7_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p7_p3, ch_mag_int, a_r_p7_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p7_p1, ch_mag_int, a_r_p7_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p7_m1, ch_mag_int, a_r_p7_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p7_m3, ch_mag_int, a_r_p7_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p7_m5, ch_mag_int, a_r_p7_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p7_m7, ch_mag_int, a_r_p7_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p5_p7, ch_mag_int, a_r_p5_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p5_p5, ch_mag_int, a_r_p5_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p5_p3, ch_mag_int, a_r_p5_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p5_p1, ch_mag_int, a_r_p5_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p5_m1, ch_mag_int, a_r_p5_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p5_m3, ch_mag_int, a_r_p5_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p5_m5, ch_mag_int, a_r_p5_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p5_m7, ch_mag_int, a_r_p5_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_p7, ch_mag_int, a_r_p3_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_p5, ch_mag_int, a_r_p3_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_p3, ch_mag_int, a_r_p3_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_p1, ch_mag_int, a_r_p3_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_m1, ch_mag_int, a_r_p3_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_m3, ch_mag_int, a_r_p3_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_m5, ch_mag_int, a_r_p3_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p3_m7, ch_mag_int, a_r_p3_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_p7, ch_mag_int, a_r_p1_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_p5, ch_mag_int, a_r_p1_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_p3, ch_mag_int, a_r_p1_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_p1, ch_mag_int, a_r_p1_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_m1, ch_mag_int, a_r_p1_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_m3, ch_mag_int, a_r_p1_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_m5, ch_mag_int, a_r_p1_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_p1_m7, ch_mag_int, a_r_p1_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_p7, ch_mag_int, a_r_m1_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_p5, ch_mag_int, a_r_m1_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_p3, ch_mag_int, a_r_m1_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_p1, ch_mag_int, a_r_m1_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_m1, ch_mag_int, a_r_m1_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_m3, ch_mag_int, a_r_m1_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_m5, ch_mag_int, a_r_m1_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m1_m7, ch_mag_int, a_r_m1_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_p7, ch_mag_int, a_r_m3_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_p5, ch_mag_int, a_r_m3_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_p3, ch_mag_int, a_r_m3_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_p1, ch_mag_int, a_r_m3_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_m1, ch_mag_int, a_r_m3_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_m3, ch_mag_int, a_r_m3_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_m5, ch_mag_int, a_r_m3_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m3_m7, ch_mag_int, a_r_m3_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m5_p7, ch_mag_int, a_r_m5_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m5_p5, ch_mag_int, a_r_m5_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m5_p3, ch_mag_int, a_r_m5_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m5_p1, ch_mag_int, a_r_m5_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m5_m1, ch_mag_int, a_r_m5_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m5_m3, ch_mag_int, a_r_m5_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m5_m5, ch_mag_int, a_r_m5_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m5_m7, ch_mag_int, a_r_m5_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m7_p7, ch_mag_int, a_r_m7_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m7_p5, ch_mag_int, a_r_m7_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m7_p3, ch_mag_int, a_r_m7_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m7_p1, ch_mag_int, a_r_m7_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m7_m1, ch_mag_int, a_r_m7_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m7_m3, ch_mag_int, a_r_m7_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m7_m5, ch_mag_int, a_r_m7_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_r_m7_m7, ch_mag_int, a_r_m7_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + + interference_abs_epi16(psi_i_p7_p7, ch_mag_int, a_i_p7_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p7_p5, ch_mag_int, a_i_p7_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p7_p3, ch_mag_int, a_i_p7_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p7_p1, ch_mag_int, a_i_p7_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p7_m1, ch_mag_int, a_i_p7_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p7_m3, ch_mag_int, a_i_p7_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p7_m5, ch_mag_int, a_i_p7_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p7_m7, ch_mag_int, a_i_p7_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p5_p7, ch_mag_int, a_i_p5_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p5_p5, ch_mag_int, a_i_p5_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p5_p3, ch_mag_int, a_i_p5_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p5_p1, ch_mag_int, a_i_p5_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p5_m1, ch_mag_int, a_i_p5_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p5_m3, ch_mag_int, a_i_p5_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p5_m5, ch_mag_int, a_i_p5_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p5_m7, ch_mag_int, a_i_p5_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_p7, ch_mag_int, a_i_p3_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_p5, ch_mag_int, a_i_p3_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_p3, ch_mag_int, a_i_p3_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_p1, ch_mag_int, a_i_p3_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_m1, ch_mag_int, a_i_p3_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_m3, ch_mag_int, a_i_p3_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_m5, ch_mag_int, a_i_p3_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p3_m7, ch_mag_int, a_i_p3_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_p7, ch_mag_int, a_i_p1_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_p5, ch_mag_int, a_i_p1_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_p3, ch_mag_int, a_i_p1_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_p1, ch_mag_int, a_i_p1_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_m1, ch_mag_int, a_i_p1_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_m3, ch_mag_int, a_i_p1_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_m5, ch_mag_int, a_i_p1_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_p1_m7, ch_mag_int, a_i_p1_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_p7, ch_mag_int, a_i_m1_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_p5, ch_mag_int, a_i_m1_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_p3, ch_mag_int, a_i_m1_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_p1, ch_mag_int, a_i_m1_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_m1, ch_mag_int, a_i_m1_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_m3, ch_mag_int, a_i_m1_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_m5, ch_mag_int, a_i_m1_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m1_m7, ch_mag_int, a_i_m1_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_p7, ch_mag_int, a_i_m3_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_p5, ch_mag_int, a_i_m3_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_p3, ch_mag_int, a_i_m3_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_p1, ch_mag_int, a_i_m3_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_m1, ch_mag_int, a_i_m3_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_m3, ch_mag_int, a_i_m3_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_m5, ch_mag_int, a_i_m3_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m3_m7, ch_mag_int, a_i_m3_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m5_p7, ch_mag_int, a_i_m5_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m5_p5, ch_mag_int, a_i_m5_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m5_p3, ch_mag_int, a_i_m5_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m5_p1, ch_mag_int, a_i_m5_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m5_m1, ch_mag_int, a_i_m5_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m5_m3, ch_mag_int, a_i_m5_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m5_m5, ch_mag_int, a_i_m5_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m5_m7, ch_mag_int, a_i_m5_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m7_p7, ch_mag_int, a_i_m7_p7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m7_p5, ch_mag_int, a_i_m7_p5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m7_p3, ch_mag_int, a_i_m7_p3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m7_p1, ch_mag_int, a_i_m7_p1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m7_m1, ch_mag_int, a_i_m7_m1, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m7_m3, ch_mag_int, a_i_m7_m3, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m7_m5, ch_mag_int, a_i_m7_m5, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + interference_abs_epi16(psi_i_m7_m7, ch_mag_int, a_i_m7_m7, ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10); + + // Calculation of a group of two terms in the bit metric involving product of psi and interference + prodsum_psi_a_epi16(psi_r_p7_p7, a_r_p7_p7, psi_i_p7_p7, a_i_p7_p7, psi_a_p7_p7); + prodsum_psi_a_epi16(psi_r_p7_p5, a_r_p7_p5, psi_i_p7_p5, a_i_p7_p5, psi_a_p7_p5); + prodsum_psi_a_epi16(psi_r_p7_p3, a_r_p7_p3, psi_i_p7_p3, a_i_p7_p3, psi_a_p7_p3); + prodsum_psi_a_epi16(psi_r_p7_p1, a_r_p7_p1, psi_i_p7_p1, a_i_p7_p1, psi_a_p7_p1); + prodsum_psi_a_epi16(psi_r_p7_m1, a_r_p7_m1, psi_i_p7_m1, a_i_p7_m1, psi_a_p7_m1); + prodsum_psi_a_epi16(psi_r_p7_m3, a_r_p7_m3, psi_i_p7_m3, a_i_p7_m3, psi_a_p7_m3); + prodsum_psi_a_epi16(psi_r_p7_m5, a_r_p7_m5, psi_i_p7_m5, a_i_p7_m5, psi_a_p7_m5); + prodsum_psi_a_epi16(psi_r_p7_m7, a_r_p7_m7, psi_i_p7_m7, a_i_p7_m7, psi_a_p7_m7); + prodsum_psi_a_epi16(psi_r_p5_p7, a_r_p5_p7, psi_i_p5_p7, a_i_p5_p7, psi_a_p5_p7); + prodsum_psi_a_epi16(psi_r_p5_p5, a_r_p5_p5, psi_i_p5_p5, a_i_p5_p5, psi_a_p5_p5); + prodsum_psi_a_epi16(psi_r_p5_p3, a_r_p5_p3, psi_i_p5_p3, a_i_p5_p3, psi_a_p5_p3); + prodsum_psi_a_epi16(psi_r_p5_p1, a_r_p5_p1, psi_i_p5_p1, a_i_p5_p1, psi_a_p5_p1); + prodsum_psi_a_epi16(psi_r_p5_m1, a_r_p5_m1, psi_i_p5_m1, a_i_p5_m1, psi_a_p5_m1); + prodsum_psi_a_epi16(psi_r_p5_m3, a_r_p5_m3, psi_i_p5_m3, a_i_p5_m3, psi_a_p5_m3); + prodsum_psi_a_epi16(psi_r_p5_m5, a_r_p5_m5, psi_i_p5_m5, a_i_p5_m5, psi_a_p5_m5); + prodsum_psi_a_epi16(psi_r_p5_m7, a_r_p5_m7, psi_i_p5_m7, a_i_p5_m7, psi_a_p5_m7); + prodsum_psi_a_epi16(psi_r_p3_p7, a_r_p3_p7, psi_i_p3_p7, a_i_p3_p7, psi_a_p3_p7); + prodsum_psi_a_epi16(psi_r_p3_p5, a_r_p3_p5, psi_i_p3_p5, a_i_p3_p5, psi_a_p3_p5); + prodsum_psi_a_epi16(psi_r_p3_p3, a_r_p3_p3, psi_i_p3_p3, a_i_p3_p3, psi_a_p3_p3); + prodsum_psi_a_epi16(psi_r_p3_p1, a_r_p3_p1, psi_i_p3_p1, a_i_p3_p1, psi_a_p3_p1); + prodsum_psi_a_epi16(psi_r_p3_m1, a_r_p3_m1, psi_i_p3_m1, a_i_p3_m1, psi_a_p3_m1); + prodsum_psi_a_epi16(psi_r_p3_m3, a_r_p3_m3, psi_i_p3_m3, a_i_p3_m3, psi_a_p3_m3); + prodsum_psi_a_epi16(psi_r_p3_m5, a_r_p3_m5, psi_i_p3_m5, a_i_p3_m5, psi_a_p3_m5); + prodsum_psi_a_epi16(psi_r_p3_m7, a_r_p3_m7, psi_i_p3_m7, a_i_p3_m7, psi_a_p3_m7); + prodsum_psi_a_epi16(psi_r_p1_p7, a_r_p1_p7, psi_i_p1_p7, a_i_p1_p7, psi_a_p1_p7); + prodsum_psi_a_epi16(psi_r_p1_p5, a_r_p1_p5, psi_i_p1_p5, a_i_p1_p5, psi_a_p1_p5); + prodsum_psi_a_epi16(psi_r_p1_p3, a_r_p1_p3, psi_i_p1_p3, a_i_p1_p3, psi_a_p1_p3); + prodsum_psi_a_epi16(psi_r_p1_p1, a_r_p1_p1, psi_i_p1_p1, a_i_p1_p1, psi_a_p1_p1); + prodsum_psi_a_epi16(psi_r_p1_m1, a_r_p1_m1, psi_i_p1_m1, a_i_p1_m1, psi_a_p1_m1); + prodsum_psi_a_epi16(psi_r_p1_m3, a_r_p1_m3, psi_i_p1_m3, a_i_p1_m3, psi_a_p1_m3); + prodsum_psi_a_epi16(psi_r_p1_m5, a_r_p1_m5, psi_i_p1_m5, a_i_p1_m5, psi_a_p1_m5); + prodsum_psi_a_epi16(psi_r_p1_m7, a_r_p1_m7, psi_i_p1_m7, a_i_p1_m7, psi_a_p1_m7); + prodsum_psi_a_epi16(psi_r_m1_p7, a_r_m1_p7, psi_i_m1_p7, a_i_m1_p7, psi_a_m1_p7); + prodsum_psi_a_epi16(psi_r_m1_p5, a_r_m1_p5, psi_i_m1_p5, a_i_m1_p5, psi_a_m1_p5); + prodsum_psi_a_epi16(psi_r_m1_p3, a_r_m1_p3, psi_i_m1_p3, a_i_m1_p3, psi_a_m1_p3); + prodsum_psi_a_epi16(psi_r_m1_p1, a_r_m1_p1, psi_i_m1_p1, a_i_m1_p1, psi_a_m1_p1); + prodsum_psi_a_epi16(psi_r_m1_m1, a_r_m1_m1, psi_i_m1_m1, a_i_m1_m1, psi_a_m1_m1); + prodsum_psi_a_epi16(psi_r_m1_m3, a_r_m1_m3, psi_i_m1_m3, a_i_m1_m3, psi_a_m1_m3); + prodsum_psi_a_epi16(psi_r_m1_m5, a_r_m1_m5, psi_i_m1_m5, a_i_m1_m5, psi_a_m1_m5); + prodsum_psi_a_epi16(psi_r_m1_m7, a_r_m1_m7, psi_i_m1_m7, a_i_m1_m7, psi_a_m1_m7); + prodsum_psi_a_epi16(psi_r_m3_p7, a_r_m3_p7, psi_i_m3_p7, a_i_m3_p7, psi_a_m3_p7); + prodsum_psi_a_epi16(psi_r_m3_p5, a_r_m3_p5, psi_i_m3_p5, a_i_m3_p5, psi_a_m3_p5); + prodsum_psi_a_epi16(psi_r_m3_p3, a_r_m3_p3, psi_i_m3_p3, a_i_m3_p3, psi_a_m3_p3); + prodsum_psi_a_epi16(psi_r_m3_p1, a_r_m3_p1, psi_i_m3_p1, a_i_m3_p1, psi_a_m3_p1); + prodsum_psi_a_epi16(psi_r_m3_m1, a_r_m3_m1, psi_i_m3_m1, a_i_m3_m1, psi_a_m3_m1); + prodsum_psi_a_epi16(psi_r_m3_m3, a_r_m3_m3, psi_i_m3_m3, a_i_m3_m3, psi_a_m3_m3); + prodsum_psi_a_epi16(psi_r_m3_m5, a_r_m3_m5, psi_i_m3_m5, a_i_m3_m5, psi_a_m3_m5); + prodsum_psi_a_epi16(psi_r_m3_m7, a_r_m3_m7, psi_i_m3_m7, a_i_m3_m7, psi_a_m3_m7); + prodsum_psi_a_epi16(psi_r_m5_p7, a_r_m5_p7, psi_i_m5_p7, a_i_m5_p7, psi_a_m5_p7); + prodsum_psi_a_epi16(psi_r_m5_p5, a_r_m5_p5, psi_i_m5_p5, a_i_m5_p5, psi_a_m5_p5); + prodsum_psi_a_epi16(psi_r_m5_p3, a_r_m5_p3, psi_i_m5_p3, a_i_m5_p3, psi_a_m5_p3); + prodsum_psi_a_epi16(psi_r_m5_p1, a_r_m5_p1, psi_i_m5_p1, a_i_m5_p1, psi_a_m5_p1); + prodsum_psi_a_epi16(psi_r_m5_m1, a_r_m5_m1, psi_i_m5_m1, a_i_m5_m1, psi_a_m5_m1); + prodsum_psi_a_epi16(psi_r_m5_m3, a_r_m5_m3, psi_i_m5_m3, a_i_m5_m3, psi_a_m5_m3); + prodsum_psi_a_epi16(psi_r_m5_m5, a_r_m5_m5, psi_i_m5_m5, a_i_m5_m5, psi_a_m5_m5); + prodsum_psi_a_epi16(psi_r_m5_m7, a_r_m5_m7, psi_i_m5_m7, a_i_m5_m7, psi_a_m5_m7); + prodsum_psi_a_epi16(psi_r_m7_p7, a_r_m7_p7, psi_i_m7_p7, a_i_m7_p7, psi_a_m7_p7); + prodsum_psi_a_epi16(psi_r_m7_p5, a_r_m7_p5, psi_i_m7_p5, a_i_m7_p5, psi_a_m7_p5); + prodsum_psi_a_epi16(psi_r_m7_p3, a_r_m7_p3, psi_i_m7_p3, a_i_m7_p3, psi_a_m7_p3); + prodsum_psi_a_epi16(psi_r_m7_p1, a_r_m7_p1, psi_i_m7_p1, a_i_m7_p1, psi_a_m7_p1); + prodsum_psi_a_epi16(psi_r_m7_m1, a_r_m7_m1, psi_i_m7_m1, a_i_m7_m1, psi_a_m7_m1); + prodsum_psi_a_epi16(psi_r_m7_m3, a_r_m7_m3, psi_i_m7_m3, a_i_m7_m3, psi_a_m7_m3); + prodsum_psi_a_epi16(psi_r_m7_m5, a_r_m7_m5, psi_i_m7_m5, a_i_m7_m5, psi_a_m7_m5); + prodsum_psi_a_epi16(psi_r_m7_m7, a_r_m7_m7, psi_i_m7_m7, a_i_m7_m7, psi_a_m7_m7); + + // Calculation of a group of two terms in the bit metric involving squares of interference + square_a_epi16(a_r_p7_p7, a_i_p7_p7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p7_p7); + square_a_epi16(a_r_p7_p5, a_i_p7_p5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p7_p5); + square_a_epi16(a_r_p7_p3, a_i_p7_p3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p7_p3); + square_a_epi16(a_r_p7_p1, a_i_p7_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p7_p1); + square_a_epi16(a_r_p7_m1, a_i_p7_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p7_m1); + square_a_epi16(a_r_p7_m3, a_i_p7_m3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p7_m3); + square_a_epi16(a_r_p7_m5, a_i_p7_m5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p7_m5); + square_a_epi16(a_r_p7_m7, a_i_p7_m7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p7_m7); + square_a_epi16(a_r_p5_p7, a_i_p5_p7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p5_p7); + square_a_epi16(a_r_p5_p5, a_i_p5_p5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p5_p5); + square_a_epi16(a_r_p5_p3, a_i_p5_p3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p5_p3); + square_a_epi16(a_r_p5_p1, a_i_p5_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p5_p1); + square_a_epi16(a_r_p5_m1, a_i_p5_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p5_m1); + square_a_epi16(a_r_p5_m3, a_i_p5_m3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p5_m3); + square_a_epi16(a_r_p5_m5, a_i_p5_m5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p5_m5); + square_a_epi16(a_r_p5_m7, a_i_p5_m7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p5_m7); + square_a_epi16(a_r_p3_p7, a_i_p3_p7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p3_p7); + square_a_epi16(a_r_p3_p5, a_i_p3_p5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p3_p5); + square_a_epi16(a_r_p3_p3, a_i_p3_p3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p3_p3); + square_a_epi16(a_r_p3_p1, a_i_p3_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p3_p1); + square_a_epi16(a_r_p3_m1, a_i_p3_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p3_m1); + square_a_epi16(a_r_p3_m3, a_i_p3_m3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p3_m3); + square_a_epi16(a_r_p3_m5, a_i_p3_m5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p3_m5); + square_a_epi16(a_r_p3_m7, a_i_p3_m7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p3_m7); + square_a_epi16(a_r_p1_p7, a_i_p1_p7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_p7); + square_a_epi16(a_r_p1_p5, a_i_p1_p5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_p5); + square_a_epi16(a_r_p1_p3, a_i_p1_p3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_p3); + square_a_epi16(a_r_p1_p1, a_i_p1_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_p1); + square_a_epi16(a_r_p1_m1, a_i_p1_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_m1); + square_a_epi16(a_r_p1_m3, a_i_p1_m3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_m3); + square_a_epi16(a_r_p1_m5, a_i_p1_m5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_m5); + square_a_epi16(a_r_p1_m7, a_i_p1_m7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_m7); + square_a_epi16(a_r_m1_p7, a_i_m1_p7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_p7); + square_a_epi16(a_r_m1_p5, a_i_m1_p5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_p5); + square_a_epi16(a_r_m1_p3, a_i_m1_p3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_p3); + square_a_epi16(a_r_m1_p1, a_i_m1_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_p1); + square_a_epi16(a_r_m1_m1, a_i_m1_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_m1); + square_a_epi16(a_r_m1_m3, a_i_m1_m3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_m3); + square_a_epi16(a_r_m1_m5, a_i_m1_m5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_m5); + square_a_epi16(a_r_m1_m7, a_i_m1_m7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_m7); + square_a_epi16(a_r_m3_p7, a_i_m3_p7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m3_p7); + square_a_epi16(a_r_m3_p5, a_i_m3_p5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m3_p5); + square_a_epi16(a_r_m3_p3, a_i_m3_p3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m3_p3); + square_a_epi16(a_r_m3_p1, a_i_m3_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m3_p1); + square_a_epi16(a_r_m3_m1, a_i_m3_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m3_m1); + square_a_epi16(a_r_m3_m3, a_i_m3_m3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m3_m3); + square_a_epi16(a_r_m3_m5, a_i_m3_m5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m3_m5); + square_a_epi16(a_r_m3_m7, a_i_m3_m7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m3_m7); + square_a_epi16(a_r_m5_p7, a_i_m5_p7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m5_p7); + square_a_epi16(a_r_m5_p5, a_i_m5_p5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m5_p5); + square_a_epi16(a_r_m5_p3, a_i_m5_p3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m5_p3); + square_a_epi16(a_r_m5_p1, a_i_m5_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m5_p1); + square_a_epi16(a_r_m5_m1, a_i_m5_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m5_m1); + square_a_epi16(a_r_m5_m3, a_i_m5_m3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m5_m3); + square_a_epi16(a_r_m5_m5, a_i_m5_m5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m5_m5); + square_a_epi16(a_r_m5_m7, a_i_m5_m7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m5_m7); + square_a_epi16(a_r_m7_p7, a_i_m7_p7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m7_p7); + square_a_epi16(a_r_m7_p5, a_i_m7_p5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m7_p5); + square_a_epi16(a_r_m7_p3, a_i_m7_p3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m7_p3); + square_a_epi16(a_r_m7_p1, a_i_m7_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m7_p1); + square_a_epi16(a_r_m7_m1, a_i_m7_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m7_m1); + square_a_epi16(a_r_m7_m3, a_i_m7_m3, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m7_m3); + square_a_epi16(a_r_m7_m5, a_i_m7_m5, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m7_m5); + square_a_epi16(a_r_m7_m7, a_i_m7_m7, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m7_m7); + + // Computing different multiples of ||h0||^2 + // x=1, y=1 + ch_mag_2_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,ONE_OVER_FOUR_SQRT_42); + ch_mag_2_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_2_over_42_with_sigma2,1); + // x=1, y=3 + ch_mag_10_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,FIVE_OVER_FOUR_SQRT_42); + ch_mag_10_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_10_over_42_with_sigma2,1); + // x=1, x=5 + ch_mag_26_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,THIRTEEN_OVER_FOUR_SQRT_42); + ch_mag_26_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_26_over_42_with_sigma2,1); + // x=1, y=7 + ch_mag_50_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,TWENTYFIVE_OVER_FOUR_SQRT_42); + ch_mag_50_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_50_over_42_with_sigma2,1); + // x=3, y=3 + ch_mag_18_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,NINE_OVER_FOUR_SQRT_42); + ch_mag_18_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_18_over_42_with_sigma2,1); + // x=3, y=5 + ch_mag_34_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,SEVENTEEN_OVER_FOUR_SQRT_42); + ch_mag_34_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_34_over_42_with_sigma2,1); + // x=3, y=7 + ch_mag_58_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,TWENTYNINE_OVER_FOUR_SQRT_42); + ch_mag_58_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_58_over_42_with_sigma2,2); + // x=5, y=5 + ch_mag_50_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,TWENTYFIVE_OVER_FOUR_SQRT_42); + ch_mag_50_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_50_over_42_with_sigma2,1); + // x=5, y=7 + ch_mag_74_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,THIRTYSEVEN_OVER_FOUR_SQRT_42); + ch_mag_74_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_74_over_42_with_sigma2,2); + // x=7, y=7 + ch_mag_98_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,FORTYNINE_OVER_FOUR_SQRT_42); + ch_mag_98_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_98_over_42_with_sigma2,2); + + // Computing Metrics + xmm0 = _mm_subs_epi16(psi_a_p7_p7, a_sq_p7_p7); + xmm1 = _mm_adds_epi16(xmm0, y0_p_7_7); + bit_met_p7_p7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_p5, a_sq_p7_p5); + xmm1 = _mm_adds_epi16(xmm0, y0_p_7_5); + bit_met_p7_p5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_p3, a_sq_p7_p3); + xmm1 = _mm_adds_epi16(xmm0, y0_p_7_3); + bit_met_p7_p3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_p1, a_sq_p7_p1); + xmm1 = _mm_adds_epi16(xmm0, y0_p_7_1); + bit_met_p7_p1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_m1, a_sq_p7_m1); + xmm1 = _mm_adds_epi16(xmm0, y0_m_7_1); + bit_met_p7_m1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_m3, a_sq_p7_m3); + xmm1 = _mm_adds_epi16(xmm0, y0_m_7_3); + bit_met_p7_m3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_m5, a_sq_p7_m5); + xmm1 = _mm_adds_epi16(xmm0, y0_m_7_5); + bit_met_p7_m5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_m7, a_sq_p7_m7); + xmm1 = _mm_adds_epi16(xmm0, y0_m_7_7); + bit_met_p7_m7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_p7, a_sq_p5_p7); + xmm1 = _mm_adds_epi16(xmm0, y0_p_5_7); + bit_met_p5_p7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_p5, a_sq_p5_p5); + xmm1 = _mm_adds_epi16(xmm0, y0_p_5_5); + bit_met_p5_p5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_p3, a_sq_p5_p3); + xmm1 = _mm_adds_epi16(xmm0, y0_p_5_3); + bit_met_p5_p3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_p1, a_sq_p5_p1); + xmm1 = _mm_adds_epi16(xmm0, y0_p_5_1); + bit_met_p5_p1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_m1, a_sq_p5_m1); + xmm1 = _mm_adds_epi16(xmm0, y0_m_5_1); + bit_met_p5_m1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_m3, a_sq_p5_m3); + xmm1 = _mm_adds_epi16(xmm0, y0_m_5_3); + bit_met_p5_m3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_m5, a_sq_p5_m5); + xmm1 = _mm_adds_epi16(xmm0, y0_m_5_5); + bit_met_p5_m5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_m7, a_sq_p5_m7); + xmm1 = _mm_adds_epi16(xmm0, y0_m_5_7); + bit_met_p5_m7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_p7, a_sq_p3_p7); + xmm1 = _mm_adds_epi16(xmm0, y0_p_3_7); + bit_met_p3_p7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_p5, a_sq_p3_p5); + xmm1 = _mm_adds_epi16(xmm0, y0_p_3_5); + bit_met_p3_p5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_p3, a_sq_p3_p3); + xmm1 = _mm_adds_epi16(xmm0, y0_p_3_3); + bit_met_p3_p3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_p1, a_sq_p3_p1); + xmm1 = _mm_adds_epi16(xmm0, y0_p_3_1); + bit_met_p3_p1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_m1, a_sq_p3_m1); + xmm1 = _mm_adds_epi16(xmm0, y0_m_3_1); + bit_met_p3_m1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_m3, a_sq_p3_m3); + xmm1 = _mm_adds_epi16(xmm0, y0_m_3_3); + bit_met_p3_m3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_m5, a_sq_p3_m5); + xmm1 = _mm_adds_epi16(xmm0, y0_m_3_5); + bit_met_p3_m5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_m7, a_sq_p3_m7); + xmm1 = _mm_adds_epi16(xmm0, y0_m_3_7); + bit_met_p3_m7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_p7, a_sq_p1_p7); + xmm1 = _mm_adds_epi16(xmm0, y0_p_1_7); + bit_met_p1_p7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_p5, a_sq_p1_p5); + xmm1 = _mm_adds_epi16(xmm0, y0_p_1_5); + bit_met_p1_p5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_p3, a_sq_p1_p3); + xmm1 = _mm_adds_epi16(xmm0, y0_p_1_3); + bit_met_p1_p3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_p1, a_sq_p1_p1); + xmm1 = _mm_adds_epi16(xmm0, y0_p_1_1); + bit_met_p1_p1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_m1, a_sq_p1_m1); + xmm1 = _mm_adds_epi16(xmm0, y0_m_1_1); + bit_met_p1_m1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_m3, a_sq_p1_m3); + xmm1 = _mm_adds_epi16(xmm0, y0_m_1_3); + bit_met_p1_m3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_m5, a_sq_p1_m5); + xmm1 = _mm_adds_epi16(xmm0, y0_m_1_5); + bit_met_p1_m5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_m7, a_sq_p1_m7); + xmm1 = _mm_adds_epi16(xmm0, y0_m_1_7); + bit_met_p1_m7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + + xmm0 = _mm_subs_epi16(psi_a_m1_p7, a_sq_m1_p7); + xmm1 = _mm_subs_epi16(xmm0, y0_m_1_7); + bit_met_m1_p7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_p5, a_sq_m1_p5); + xmm1 = _mm_subs_epi16(xmm0, y0_m_1_5); + bit_met_m1_p5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_p3, a_sq_m1_p3); + xmm1 = _mm_subs_epi16(xmm0, y0_m_1_3); + bit_met_m1_p3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_p1, a_sq_m1_p1); + xmm1 = _mm_subs_epi16(xmm0, y0_m_1_1); + bit_met_m1_p1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_m1, a_sq_m1_m1); + xmm1 = _mm_subs_epi16(xmm0, y0_p_1_1); + bit_met_m1_m1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_m3, a_sq_m1_m3); + xmm1 = _mm_subs_epi16(xmm0, y0_p_1_3); + bit_met_m1_m3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_m5, a_sq_m1_m5); + xmm1 = _mm_subs_epi16(xmm0, y0_p_1_5); + bit_met_m1_m5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_m7, a_sq_m1_m7); + xmm1 = _mm_subs_epi16(xmm0, y0_p_1_7); + bit_met_m1_m7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_p7, a_sq_m3_p7); + xmm1 = _mm_subs_epi16(xmm0, y0_m_3_7); + bit_met_m3_p7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_p5, a_sq_m3_p5); + xmm1 = _mm_subs_epi16(xmm0, y0_m_3_5); + bit_met_m3_p5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_p3, a_sq_m3_p3); + xmm1 = _mm_subs_epi16(xmm0, y0_m_3_3); + bit_met_m3_p3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_p1, a_sq_m3_p1); + xmm1 = _mm_subs_epi16(xmm0, y0_m_3_1); + bit_met_m3_p1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_m1, a_sq_m3_m1); + xmm1 = _mm_subs_epi16(xmm0, y0_p_3_1); + bit_met_m3_m1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_m3, a_sq_m3_m3); + xmm1 = _mm_subs_epi16(xmm0, y0_p_3_3); + bit_met_m3_m3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_m5, a_sq_m3_m5); + xmm1 = _mm_subs_epi16(xmm0, y0_p_3_5); + bit_met_m3_m5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_m7, a_sq_m3_m7); + xmm1 = _mm_subs_epi16(xmm0, y0_p_3_7); + bit_met_m3_m7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_p7, a_sq_m5_p7); + xmm1 = _mm_subs_epi16(xmm0, y0_m_5_7); + bit_met_m5_p7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_p5, a_sq_m5_p5); + xmm1 = _mm_subs_epi16(xmm0, y0_m_5_5); + bit_met_m5_p5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_p3, a_sq_m5_p3); + xmm1 = _mm_subs_epi16(xmm0, y0_m_5_3); + bit_met_m5_p3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_p1, a_sq_m5_p1); + xmm1 = _mm_subs_epi16(xmm0, y0_m_5_1); + bit_met_m5_p1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_m1, a_sq_m5_m1); + xmm1 = _mm_subs_epi16(xmm0, y0_p_5_1); + bit_met_m5_m1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_m3, a_sq_m5_m3); + xmm1 = _mm_subs_epi16(xmm0, y0_p_5_3); + bit_met_m5_m3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_m5, a_sq_m5_m5); + xmm1 = _mm_subs_epi16(xmm0, y0_p_5_5); + bit_met_m5_m5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_m7, a_sq_m5_m7); + xmm1 = _mm_subs_epi16(xmm0, y0_p_5_7); + bit_met_m5_m7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_p7, a_sq_m7_p7); + xmm1 = _mm_subs_epi16(xmm0, y0_m_7_7); + bit_met_m7_p7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_p5, a_sq_m7_p5); + xmm1 = _mm_subs_epi16(xmm0, y0_m_7_5); + bit_met_m7_p5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_p3, a_sq_m7_p3); + xmm1 = _mm_subs_epi16(xmm0, y0_m_7_3); + bit_met_m7_p3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_p1, a_sq_m7_p1); + xmm1 = _mm_subs_epi16(xmm0, y0_m_7_1); + bit_met_m7_p1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_m1, a_sq_m7_m1); + xmm1 = _mm_subs_epi16(xmm0, y0_p_7_1); + bit_met_m7_m1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_m3, a_sq_m7_m3); + xmm1 = _mm_subs_epi16(xmm0, y0_p_7_3); + bit_met_m7_m3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_m5, a_sq_m7_m5); + xmm1 = _mm_subs_epi16(xmm0, y0_p_7_5); + bit_met_m7_m5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_m7, a_sq_m7_m7); + xmm1 = _mm_subs_epi16(xmm0, y0_p_7_7); + bit_met_m7_m7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + + // Detection for 1st bit (LTE mapping) + // bit = 1 + xmm0 = _mm_max_epi16(bit_met_m7_p7, bit_met_m7_p5); + xmm1 = _mm_max_epi16(bit_met_m7_p3, bit_met_m7_p1); + xmm2 = _mm_max_epi16(bit_met_m7_m1, bit_met_m7_m3); + xmm3 = _mm_max_epi16(bit_met_m7_m5, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m5_p7, bit_met_m5_p5); + xmm1 = _mm_max_epi16(bit_met_m5_p3, bit_met_m5_p1); + xmm2 = _mm_max_epi16(bit_met_m5_m1, bit_met_m5_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m5_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_m3_p7, bit_met_m3_p5); + xmm1 = _mm_max_epi16(bit_met_m3_p3, bit_met_m3_p1); + xmm2 = _mm_max_epi16(bit_met_m3_m1, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m3_m5, bit_met_m3_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_m1_p7, bit_met_m1_p5); + xmm1 = _mm_max_epi16(bit_met_m1_p3, bit_met_m1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m1_m3); + xmm3 = _mm_max_epi16(bit_met_m1_m5, bit_met_m1_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + // bit = 0 + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p7_p5); + xmm1 = _mm_max_epi16(bit_met_p7_p3, bit_met_p7_p1); + xmm2 = _mm_max_epi16(bit_met_p7_m1, bit_met_p7_m3); + xmm3 = _mm_max_epi16(bit_met_p7_m5, bit_met_p7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p5_p7, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p5_p3, bit_met_p5_p1); + xmm2 = _mm_max_epi16(bit_met_p5_m1, bit_met_p5_m3); + xmm3 = _mm_max_epi16(bit_met_p5_m5, bit_met_p5_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p3_p7, bit_met_p3_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p3_p1); + xmm2 = _mm_max_epi16(bit_met_p3_m1, bit_met_p3_m3); + xmm3 = _mm_max_epi16(bit_met_p3_m5, bit_met_p3_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p1_p7, bit_met_p1_p5); + xmm1 = _mm_max_epi16(bit_met_p1_p3, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_p1_m1, bit_met_p1_m3); + xmm3 = _mm_max_epi16(bit_met_p1_m5, bit_met_p1_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y0r = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 2nd bit (LTE mapping) + // bit = 1 + xmm0 = _mm_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = _mm_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = _mm_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = _mm_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = _mm_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = _mm_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = _mm_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + // bit = 0 + xmm0 = _mm_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = _mm_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = _mm_max_epi16(bit_met_m5_p1, bit_met_m7_p1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = _mm_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = _mm_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m7_p5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = _mm_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = _mm_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = _mm_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y1r = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 3rd bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_m7_m7, bit_met_m7_m5); + xmm1 = _mm_max_epi16(bit_met_m7_m3, bit_met_m7_m1); + xmm2 = _mm_max_epi16(bit_met_m7_p1, bit_met_m7_p3); + xmm3 = _mm_max_epi16(bit_met_m7_p5, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m5_m7, bit_met_m5_m5); + xmm1 = _mm_max_epi16(bit_met_m5_m3, bit_met_m5_m1); + xmm2 = _mm_max_epi16(bit_met_m5_p1, bit_met_m5_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p5_m7, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p5_m3, bit_met_p5_m1); + xmm2 = _mm_max_epi16(bit_met_p5_p1, bit_met_p5_p3); + xmm3 = _mm_max_epi16(bit_met_p5_p5, bit_met_p5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p7_m5); + xmm1 = _mm_max_epi16(bit_met_p7_m3, bit_met_p7_m1); + xmm2 = _mm_max_epi16(bit_met_p7_p1, bit_met_p7_p3); + xmm3 = _mm_max_epi16(bit_met_p7_p5, bit_met_p7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_m3_m7, bit_met_m3_m5); + xmm1 = _mm_max_epi16(bit_met_m3_m3, bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_m3_p1, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m3_p5, bit_met_m3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m1_m7, bit_met_m1_m5); + xmm1 = _mm_max_epi16(bit_met_m1_m3, bit_met_m1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m1_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p5, bit_met_m1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p1_m7, bit_met_p1_m5); + xmm1 = _mm_max_epi16(bit_met_p1_m3, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_p1_p1, bit_met_p1_p3); + xmm3 = _mm_max_epi16(bit_met_p1_p5, bit_met_p1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p3_m7, bit_met_p3_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p3_m1); + xmm2 = _mm_max_epi16(bit_met_p3_p1, bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_p5, bit_met_p3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y2r = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 4th bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = _mm_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = _mm_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = _mm_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = _mm_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m5_p5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = _mm_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = _mm_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = _mm_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = _mm_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = _mm_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = _mm_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = _mm_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = _mm_max_epi16(bit_met_m5_p1, bit_met_m7_p1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = _mm_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y0i = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + + // Detection for 5th bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_m7_m7, bit_met_m7_m5); + xmm1 = _mm_max_epi16(bit_met_m7_m3, bit_met_m7_m1); + xmm2 = _mm_max_epi16(bit_met_m7_p1, bit_met_m7_p3); + xmm3 = _mm_max_epi16(bit_met_m7_p5, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m1_m7, bit_met_m1_m5); + xmm1 = _mm_max_epi16(bit_met_m1_m3, bit_met_m1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m1_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p5, bit_met_m1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p1_m7, bit_met_p1_m5); + xmm1 = _mm_max_epi16(bit_met_p1_m3, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_p1_p1, bit_met_p1_p3); + xmm3 = _mm_max_epi16(bit_met_p1_p5, bit_met_p1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p7_m5); + xmm1 = _mm_max_epi16(bit_met_p7_m3, bit_met_p7_m1); + xmm2 = _mm_max_epi16(bit_met_p7_p1, bit_met_p7_p3); + xmm3 = _mm_max_epi16(bit_met_p7_p5, bit_met_p7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_m5_m7, bit_met_m5_m5); + xmm1 = _mm_max_epi16(bit_met_m5_m3, bit_met_m5_m1); + xmm2 = _mm_max_epi16(bit_met_m5_p1, bit_met_m5_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m3_m7, bit_met_m3_m5); + xmm1 = _mm_max_epi16(bit_met_m3_m3, bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_m3_p1, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m3_p5, bit_met_m3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p3_m7, bit_met_p3_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p3_m1); + xmm2 = _mm_max_epi16(bit_met_p3_p1, bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_p5, bit_met_p3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p5_m7, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p5_m3, bit_met_p5_m1); + xmm2 = _mm_max_epi16(bit_met_p5_p1, bit_met_p5_p3); + xmm3 = _mm_max_epi16(bit_met_p5_p5, bit_met_p5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y1i = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 6th bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = _mm_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = _mm_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = _mm_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = _mm_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = _mm_max_epi16(bit_met_m5_p1, bit_met_m5_p1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = _mm_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = _mm_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = _mm_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = _mm_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = _mm_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = _mm_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = _mm_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = _mm_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m7_p5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y2i = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + + // map to output stream, difficult to do in SIMD since we have 6 16bit LLRs + // RE 1 + j = 24*i; + stream0_out[j + 0] = ((short *)&y0r)[0]; + stream0_out[j + 1] = ((short *)&y1r)[0]; + stream0_out[j + 2] = ((short *)&y2r)[0]; + stream0_out[j + 3] = ((short *)&y0i)[0]; + stream0_out[j + 4] = ((short *)&y1i)[0]; + stream0_out[j + 5] = ((short *)&y2i)[0]; + // RE 2 + stream0_out[j + 6] = ((short *)&y0r)[1]; + stream0_out[j + 7] = ((short *)&y1r)[1]; + stream0_out[j + 8] = ((short *)&y2r)[1]; + stream0_out[j + 9] = ((short *)&y0i)[1]; + stream0_out[j + 10] = ((short *)&y1i)[1]; + stream0_out[j + 11] = ((short *)&y2i)[1]; + // RE 3 + stream0_out[j + 12] = ((short *)&y0r)[2]; + stream0_out[j + 13] = ((short *)&y1r)[2]; + stream0_out[j + 14] = ((short *)&y2r)[2]; + stream0_out[j + 15] = ((short *)&y0i)[2]; + stream0_out[j + 16] = ((short *)&y1i)[2]; + stream0_out[j + 17] = ((short *)&y2i)[2]; + // RE 4 + stream0_out[j + 18] = ((short *)&y0r)[3]; + stream0_out[j + 19] = ((short *)&y1r)[3]; + stream0_out[j + 20] = ((short *)&y2r)[3]; + stream0_out[j + 21] = ((short *)&y0i)[3]; + stream0_out[j + 22] = ((short *)&y1i)[3]; + stream0_out[j + 23] = ((short *)&y2i)[3]; + // RE 5 + stream0_out[j + 24] = ((short *)&y0r)[4]; + stream0_out[j + 25] = ((short *)&y1r)[4]; + stream0_out[j + 26] = ((short *)&y2r)[4]; + stream0_out[j + 27] = ((short *)&y0i)[4]; + stream0_out[j + 28] = ((short *)&y1i)[4]; + stream0_out[j + 29] = ((short *)&y2i)[4]; + // RE 6 + stream0_out[j + 30] = ((short *)&y0r)[5]; + stream0_out[j + 31] = ((short *)&y1r)[5]; + stream0_out[j + 32] = ((short *)&y2r)[5]; + stream0_out[j + 33] = ((short *)&y0i)[5]; + stream0_out[j + 34] = ((short *)&y1i)[5]; + stream0_out[j + 35] = ((short *)&y2i)[5]; + // RE 7 + stream0_out[j + 36] = ((short *)&y0r)[6]; + stream0_out[j + 37] = ((short *)&y1r)[6]; + stream0_out[j + 38] = ((short *)&y2r)[6]; + stream0_out[j + 39] = ((short *)&y0i)[6]; + stream0_out[j + 40] = ((short *)&y1i)[6]; + stream0_out[j + 41] = ((short *)&y2i)[6]; + // RE 8 + stream0_out[j + 42] = ((short *)&y0r)[7]; + stream0_out[j + 43] = ((short *)&y1r)[7]; + stream0_out[j + 44] = ((short *)&y2r)[7]; + stream0_out[j + 45] = ((short *)&y0i)[7]; + stream0_out[j + 46] = ((short *)&y1i)[7]; + stream0_out[j + 47] = ((short *)&y2i)[7]; + +#elif defined(__arm__) + +#endif + } + +#if defined(__x86_64__) || defined(__i386__) + _mm_empty(); + _m_empty(); +#endif + +} + + +int nr_dlsch_64qam_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, + int32_t **rxdataF_comp, + int32_t **rxdataF_comp_i, + int32_t **dl_ch_mag, + int32_t **dl_ch_mag_i, + int32_t **rho_i, + int16_t *dlsch_llr, + uint8_t symbol, + uint8_t first_symbol_flag, + uint16_t nb_rb, + uint16_t pbch_pss_sss_adjust, + int16_t **llr16p) +{ + + int16_t *rxF = (int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rxF_i = (int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag = (int16_t*)&dl_ch_mag[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag_i = (int16_t*)&dl_ch_mag_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rho = (int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *llr16; + int len; + uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol; + + //first symbol has different structure due to more pilots + if (first_symbol_flag == 1) { + llr16 = (int16_t*)dlsch_llr; + } else { + llr16 = (int16_t*)(*llr16p); + } + + AssertFatal(llr16!=NULL,"nr_dlsch_16qam_64qam_llr:llr is null, symbol %d\n",symbol); + + if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) { + // if symbol has pilots + if (frame_parms->nb_antenna_ports_eNB!=1) + // in 2 antenna ports we have 8 REs per symbol per RB + len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3); + else + // for 1 antenna port we have 10 REs per symbol per RB + len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6); + } else { + // symbol has no pilots + len = (nb_rb*12) - pbch_pss_sss_adjust; + } + + qam64_qam16((short *)rxF, + (short *)rxF_i, + (short *)ch_mag, + (short *)ch_mag_i, + (short *)llr16, + (short *)rho, + len); + + llr16 += (6*len); + *llr16p = (short *)llr16; + return(0); +} + +void qam64_qam64(short *stream0_in, + short *stream1_in, + short *ch_mag, + short *ch_mag_i, + short *stream0_out, + short *rho01, + int length + ) +{ + + /* + Author: S. Wagner + Date: 31-07-12 + + Input: + stream0_in: MF filter for 1st stream, i.e., y0=h0'*y + stream1_in: MF filter for 2nd stream, i.e., y1=h1'*y + ch_mag: 4*h0/sqrt(42), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + ch_mag_i: 4*h1/sqrt(42), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + rho01: Channel cross correlation, i.e., h1'*h0 + + Output: + stream0_out: output LLRs for 1st stream + */ + +#if defined(__x86_64__) || defined(__i386__) + + __m128i *rho01_128i = (__m128i *)rho01; + __m128i *stream0_128i_in = (__m128i *)stream0_in; + __m128i *stream1_128i_in = (__m128i *)stream1_in; + __m128i *ch_mag_128i = (__m128i *)ch_mag; + __m128i *ch_mag_128i_i = (__m128i *)ch_mag_i; + + __m128i ONE_OVER_SQRT_42 = _mm_set1_epi16(10112); // round(1/sqrt(42)*2^16) + __m128i THREE_OVER_SQRT_42 = _mm_set1_epi16(30337); // round(3/sqrt(42)*2^16) + __m128i FIVE_OVER_SQRT_42 = _mm_set1_epi16(25281); // round(5/sqrt(42)*2^15) + __m128i SEVEN_OVER_SQRT_42 = _mm_set1_epi16(17697); // round(7/sqrt(42)*2^14) Q2.14 + __m128i ONE_OVER_SQRT_2 = _mm_set1_epi16(23170); // round(1/sqrt(2)*2^15) + __m128i ONE_OVER_SQRT_2_42 = _mm_set1_epi16(3575); // round(1/sqrt(2*42)*2^15) + __m128i THREE_OVER_SQRT_2_42 = _mm_set1_epi16(10726); // round(3/sqrt(2*42)*2^15) + __m128i FIVE_OVER_SQRT_2_42 = _mm_set1_epi16(17876); // round(5/sqrt(2*42)*2^15) + __m128i SEVEN_OVER_SQRT_2_42 = _mm_set1_epi16(25027); // round(7/sqrt(2*42)*2^15) + __m128i FORTYNINE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(30969); // round(49/(4*sqrt(42))*2^14), Q2.14 + __m128i THIRTYSEVEN_OVER_FOUR_SQRT_42 = _mm_set1_epi16(23385); // round(37/(4*sqrt(42))*2^14), Q2.14 + __m128i TWENTYFIVE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(31601); // round(25/(4*sqrt(42))*2^15) + __m128i TWENTYNINE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(18329); // round(29/(4*sqrt(42))*2^15), Q2.14 + __m128i SEVENTEEN_OVER_FOUR_SQRT_42 = _mm_set1_epi16(21489); // round(17/(4*sqrt(42))*2^15) + __m128i NINE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(11376); // round(9/(4*sqrt(42))*2^15) + __m128i THIRTEEN_OVER_FOUR_SQRT_42 = _mm_set1_epi16(16433); // round(13/(4*sqrt(42))*2^15) + __m128i FIVE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(6320); // round(5/(4*sqrt(42))*2^15) + __m128i ONE_OVER_FOUR_SQRT_42 = _mm_set1_epi16(1264); // round(1/(4*sqrt(42))*2^15) + __m128i SQRT_42_OVER_FOUR = _mm_set1_epi16(13272); // round(sqrt(42)/4*2^13), Q3.12 + + __m128i ch_mag_des; + __m128i ch_mag_int; + __m128i ch_mag_98_over_42_with_sigma2; + __m128i ch_mag_74_over_42_with_sigma2; + __m128i ch_mag_58_over_42_with_sigma2; + __m128i ch_mag_50_over_42_with_sigma2; + __m128i ch_mag_34_over_42_with_sigma2; + __m128i ch_mag_18_over_42_with_sigma2; + __m128i ch_mag_26_over_42_with_sigma2; + __m128i ch_mag_10_over_42_with_sigma2; + __m128i ch_mag_2_over_42_with_sigma2; + __m128i y0r_one_over_sqrt_21; + __m128i y0r_three_over_sqrt_21; + __m128i y0r_five_over_sqrt_21; + __m128i y0r_seven_over_sqrt_21; + __m128i y0i_one_over_sqrt_21; + __m128i y0i_three_over_sqrt_21; + __m128i y0i_five_over_sqrt_21; + __m128i y0i_seven_over_sqrt_21; + __m128i ch_mag_int_with_sigma2; + __m128i two_ch_mag_int_with_sigma2; + __m128i three_ch_mag_int_with_sigma2; +#elif defined(__arm__) + +#endif + + int i,j; + + + for (i=0; i>2; i+=2) { + +#if defined(__x86_64__) || defined(__i386__) + + // Get rho + xmm0 = rho01_128i[i]; + xmm1 = rho01_128i[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho) + xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho) + rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho) + rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho) + + // Compute the different rhos + rho_rpi_1_1 = _mm_mulhi_epi16(rho_rpi, ONE_OVER_SQRT_42); + rho_rmi_1_1 = _mm_mulhi_epi16(rho_rmi, ONE_OVER_SQRT_42); + rho_rpi_3_3 = _mm_mulhi_epi16(rho_rpi, THREE_OVER_SQRT_42); + rho_rmi_3_3 = _mm_mulhi_epi16(rho_rmi, THREE_OVER_SQRT_42); + rho_rpi_5_5 = _mm_mulhi_epi16(rho_rpi, FIVE_OVER_SQRT_42); + rho_rmi_5_5 = _mm_mulhi_epi16(rho_rmi, FIVE_OVER_SQRT_42); + rho_rpi_7_7 = _mm_mulhi_epi16(rho_rpi, SEVEN_OVER_SQRT_42); + rho_rmi_7_7 = _mm_mulhi_epi16(rho_rmi, SEVEN_OVER_SQRT_42); + + rho_rpi_5_5 = _mm_slli_epi16(rho_rpi_5_5, 1); + rho_rmi_5_5 = _mm_slli_epi16(rho_rmi_5_5, 1); + rho_rpi_7_7 = _mm_slli_epi16(rho_rpi_7_7, 2); + rho_rmi_7_7 = _mm_slli_epi16(rho_rmi_7_7, 2); + + xmm4 = _mm_mulhi_epi16(xmm2, ONE_OVER_SQRT_42); + xmm5 = _mm_mulhi_epi16(xmm3, ONE_OVER_SQRT_42); + xmm6 = _mm_mulhi_epi16(xmm3, THREE_OVER_SQRT_42); + xmm7 = _mm_mulhi_epi16(xmm3, FIVE_OVER_SQRT_42); + xmm8 = _mm_mulhi_epi16(xmm3, SEVEN_OVER_SQRT_42); + xmm7 = _mm_slli_epi16(xmm7, 1); + xmm8 = _mm_slli_epi16(xmm8, 2); + + rho_rpi_1_3 = _mm_adds_epi16(xmm4, xmm6); + rho_rmi_1_3 = _mm_subs_epi16(xmm4, xmm6); + rho_rpi_1_5 = _mm_adds_epi16(xmm4, xmm7); + rho_rmi_1_5 = _mm_subs_epi16(xmm4, xmm7); + rho_rpi_1_7 = _mm_adds_epi16(xmm4, xmm8); + rho_rmi_1_7 = _mm_subs_epi16(xmm4, xmm8); + + xmm4 = _mm_mulhi_epi16(xmm2, THREE_OVER_SQRT_42); + rho_rpi_3_1 = _mm_adds_epi16(xmm4, xmm5); + rho_rmi_3_1 = _mm_subs_epi16(xmm4, xmm5); + rho_rpi_3_5 = _mm_adds_epi16(xmm4, xmm7); + rho_rmi_3_5 = _mm_subs_epi16(xmm4, xmm7); + rho_rpi_3_7 = _mm_adds_epi16(xmm4, xmm8); + rho_rmi_3_7 = _mm_subs_epi16(xmm4, xmm8); + + xmm4 = _mm_mulhi_epi16(xmm2, FIVE_OVER_SQRT_42); + xmm4 = _mm_slli_epi16(xmm4, 1); + rho_rpi_5_1 = _mm_adds_epi16(xmm4, xmm5); + rho_rmi_5_1 = _mm_subs_epi16(xmm4, xmm5); + rho_rpi_5_3 = _mm_adds_epi16(xmm4, xmm6); + rho_rmi_5_3 = _mm_subs_epi16(xmm4, xmm6); + rho_rpi_5_7 = _mm_adds_epi16(xmm4, xmm8); + rho_rmi_5_7 = _mm_subs_epi16(xmm4, xmm8); + + xmm4 = _mm_mulhi_epi16(xmm2, SEVEN_OVER_SQRT_42); + xmm4 = _mm_slli_epi16(xmm4, 2); + rho_rpi_7_1 = _mm_adds_epi16(xmm4, xmm5); + rho_rmi_7_1 = _mm_subs_epi16(xmm4, xmm5); + rho_rpi_7_3 = _mm_adds_epi16(xmm4, xmm6); + rho_rmi_7_3 = _mm_subs_epi16(xmm4, xmm6); + rho_rpi_7_5 = _mm_adds_epi16(xmm4, xmm7); + rho_rmi_7_5 = _mm_subs_epi16(xmm4, xmm7); + + // Rearrange interfering MF output + xmm0 = stream1_128i_in[i]; + xmm1 = stream1_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)] + y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)] + + // Psi_r calculation from rho_rpi or rho_rmi + xmm0 = _mm_setzero_si128(); // ZERO for abs_pi16 + xmm2 = _mm_subs_epi16(rho_rpi_7_7, y1r); + psi_r_p7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_5, y1r); + psi_r_p7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_3, y1r); + psi_r_p7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_1, y1r); + psi_r_p7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_1, y1r); + psi_r_p7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_3, y1r); + psi_r_p7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_5, y1r); + psi_r_p7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_7, y1r); + psi_r_p7_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_7, y1r); + psi_r_p5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_5, y1r); + psi_r_p5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_3, y1r); + psi_r_p5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_1, y1r); + psi_r_p5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_1, y1r); + psi_r_p5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_3, y1r); + psi_r_p5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_5, y1r); + psi_r_p5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_7, y1r); + psi_r_p5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_7, y1r); + psi_r_p3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_5, y1r); + psi_r_p3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_3, y1r); + psi_r_p3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_1, y1r); + psi_r_p3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_1, y1r); + psi_r_p3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_3, y1r); + psi_r_p3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_5, y1r); + psi_r_p3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_7, y1r); + psi_r_p3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_7, y1r); + psi_r_p1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_5, y1r); + psi_r_p1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_3, y1r); + psi_r_p1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_1, y1r); + psi_r_p1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_1, y1r); + psi_r_p1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_3, y1r); + psi_r_p1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_5, y1r); + psi_r_p1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_7, y1r); + psi_r_p1_m7 = _mm_abs_epi16(xmm2); + + xmm2 = _mm_adds_epi16(rho_rmi_1_7, y1r); + psi_r_m1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_5, y1r); + psi_r_m1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_3, y1r); + psi_r_m1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_1, y1r); + psi_r_m1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_1, y1r); + psi_r_m1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_3, y1r); + psi_r_m1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_5, y1r); + psi_r_m1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_7, y1r); + psi_r_m1_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_7, y1r); + psi_r_m3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_5, y1r); + psi_r_m3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_3, y1r); + psi_r_m3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_1, y1r); + psi_r_m3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_1, y1r); + psi_r_m3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_3, y1r); + psi_r_m3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_5, y1r); + psi_r_m3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_7, y1r); + psi_r_m3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_7, y1r); + psi_r_m5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_5, y1r); + psi_r_m5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_3, y1r); + psi_r_m5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_1, y1r); + psi_r_m5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_1, y1r); + psi_r_m5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_3, y1r); + psi_r_m5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_5, y1r); + psi_r_m5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_7, y1r); + psi_r_m5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_7, y1r); + psi_r_m7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_5, y1r); + psi_r_m7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_3, y1r); + psi_r_m7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_1, y1r); + psi_r_m7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_1, y1r); + psi_r_m7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_3, y1r); + psi_r_m7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_5, y1r); + psi_r_m7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_7, y1r); + psi_r_m7_m7 = _mm_abs_epi16(xmm2); + + // Psi_i calculation from rho_rpi or rho_rmi + xmm2 = _mm_subs_epi16(rho_rmi_7_7, y1i); + psi_i_p7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_7, y1i); + psi_i_p7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_7, y1i); + psi_i_p7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_7, y1i); + psi_i_p7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_7, y1i); + psi_i_p7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_7, y1i); + psi_i_p7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_7, y1i); + psi_i_p7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_7, y1i); + psi_i_p7_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_5, y1i); + psi_i_p5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_5, y1i); + psi_i_p5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_5, y1i); + psi_i_p5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_5, y1i); + psi_i_p5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_5, y1i); + psi_i_p5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_5, y1i); + psi_i_p5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_5, y1i); + psi_i_p5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_5, y1i); + psi_i_p5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_3, y1i); + psi_i_p3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_3, y1i); + psi_i_p3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_3, y1i); + psi_i_p3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_3, y1i); + psi_i_p3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_3, y1i); + psi_i_p3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_3, y1i); + psi_i_p3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_3, y1i); + psi_i_p3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_3, y1i); + psi_i_p3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_7_1, y1i); + psi_i_p1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_5_1, y1i); + psi_i_p1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_3_1, y1i); + psi_i_p1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rmi_1_1, y1i); + psi_i_p1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_1_1, y1i); + psi_i_p1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_3_1, y1i); + psi_i_p1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_5_1, y1i); + psi_i_p1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rpi_7_1, y1i); + psi_i_p1_m7 = _mm_abs_epi16(xmm2); + + xmm2 = _mm_subs_epi16(rho_rpi_7_1, y1i); + psi_i_m1_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_1, y1i); + psi_i_m1_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_1, y1i); + psi_i_m1_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_1, y1i); + psi_i_m1_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_1, y1i); + psi_i_m1_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_1, y1i); + psi_i_m1_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_1, y1i); + psi_i_m1_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_1, y1i); + psi_i_m1_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_3, y1i); + psi_i_m3_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_3, y1i); + psi_i_m3_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_3, y1i); + psi_i_m3_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_3, y1i); + psi_i_m3_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_3, y1i); + psi_i_m3_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_3, y1i); + psi_i_m3_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_3, y1i); + psi_i_m3_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_3, y1i); + psi_i_m3_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_5, y1i); + psi_i_m5_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_5, y1i); + psi_i_m5_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_5, y1i); + psi_i_m5_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_5, y1i); + psi_i_m5_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_5, y1i); + psi_i_m5_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_5, y1i); + psi_i_m5_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_5, y1i); + psi_i_m5_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_5, y1i); + psi_i_m5_m7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_7_7, y1i); + psi_i_m7_p7 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_5_7, y1i); + psi_i_m7_p5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_3_7, y1i); + psi_i_m7_p3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_subs_epi16(rho_rpi_1_7, y1i); + psi_i_m7_p1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_1_7, y1i); + psi_i_m7_m1 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_3_7, y1i); + psi_i_m7_m3 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_5_7, y1i); + psi_i_m7_m5 = _mm_abs_epi16(xmm2); + xmm2 = _mm_adds_epi16(rho_rmi_7_7, y1i); + psi_i_m7_m7 = _mm_abs_epi16(xmm2); + + + // Rearrange desired MF output + xmm0 = stream0_128i_in[i]; + xmm1 = stream0_128i_in[i+1]; + xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3)); + //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)] + y0i = _mm_unpackhi_epi64(xmm0,xmm1); + + // Rearrange desired channel magnitudes + xmm2 = ch_mag_128i[i]; // = [|h|^2(1),|h|^2(1),|h|^2(2),|h|^2(2)]*(2/sqrt(10)) + xmm3 = ch_mag_128i[i+1]; // = [|h|^2(3),|h|^2(3),|h|^2(4),|h|^2(4)]*(2/sqrt(10)) + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + ch_mag_des = _mm_unpacklo_epi64(xmm2,xmm3); + + // Rearrange interfering channel magnitudes + xmm2 = ch_mag_128i_i[i]; + xmm3 = ch_mag_128i_i[i+1]; + xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); + ch_mag_int = _mm_unpacklo_epi64(xmm2,xmm3); + + y0r_one_over_sqrt_21 = _mm_mulhi_epi16(y0r, ONE_OVER_SQRT_42); + y0r_three_over_sqrt_21 = _mm_mulhi_epi16(y0r, THREE_OVER_SQRT_42); + y0r_five_over_sqrt_21 = _mm_mulhi_epi16(y0r, FIVE_OVER_SQRT_42); + y0r_five_over_sqrt_21 = _mm_slli_epi16(y0r_five_over_sqrt_21, 1); + y0r_seven_over_sqrt_21 = _mm_mulhi_epi16(y0r, SEVEN_OVER_SQRT_42); + y0r_seven_over_sqrt_21 = _mm_slli_epi16(y0r_seven_over_sqrt_21, 2); // Q2.14 + + y0i_one_over_sqrt_21 = _mm_mulhi_epi16(y0i, ONE_OVER_SQRT_42); + y0i_three_over_sqrt_21 = _mm_mulhi_epi16(y0i, THREE_OVER_SQRT_42); + y0i_five_over_sqrt_21 = _mm_mulhi_epi16(y0i, FIVE_OVER_SQRT_42); + y0i_five_over_sqrt_21 = _mm_slli_epi16(y0i_five_over_sqrt_21, 1); + y0i_seven_over_sqrt_21 = _mm_mulhi_epi16(y0i, SEVEN_OVER_SQRT_42); + y0i_seven_over_sqrt_21 = _mm_slli_epi16(y0i_seven_over_sqrt_21, 2); // Q2.14 + + y0_p_7_1 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_7_3 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_7_5 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_7_7 = _mm_adds_epi16(y0r_seven_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_p_5_1 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_5_3 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_5_5 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_5_7 = _mm_adds_epi16(y0r_five_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_p_3_1 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_3_3 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_3_5 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_3_7 = _mm_adds_epi16(y0r_three_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_p_1_1 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_one_over_sqrt_21); + y0_p_1_3 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_three_over_sqrt_21); + y0_p_1_5 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_five_over_sqrt_21); + y0_p_1_7 = _mm_adds_epi16(y0r_one_over_sqrt_21, y0i_seven_over_sqrt_21); + + y0_m_1_1 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_1_3 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_1_5 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_1_7 = _mm_subs_epi16(y0r_one_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_m_3_1 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_3_3 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_3_5 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_3_7 = _mm_subs_epi16(y0r_three_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_m_5_1 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_5_3 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_5_5 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_5_7 = _mm_subs_epi16(y0r_five_over_sqrt_21, y0i_seven_over_sqrt_21); + y0_m_7_1 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_one_over_sqrt_21); + y0_m_7_3 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_three_over_sqrt_21); + y0_m_7_5 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_five_over_sqrt_21); + y0_m_7_7 = _mm_subs_epi16(y0r_seven_over_sqrt_21, y0i_seven_over_sqrt_21); + + // Detection of interference term + ch_mag_int_with_sigma2 = _mm_srai_epi16(ch_mag_int, 1); // *2 + two_ch_mag_int_with_sigma2 = ch_mag_int; // *4 + three_ch_mag_int_with_sigma2 = _mm_adds_epi16(ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2); // *6 + + interference_abs_64qam_epi16(psi_r_p7_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p7_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p7_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p7_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p7_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p7_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p7_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p7_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p7_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p7_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p7_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p7_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p7_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p7_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p7_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p7_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p5_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p5_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p5_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p5_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p5_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p5_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p5_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p5_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p5_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p5_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p5_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p5_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p5_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p5_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p5_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p5_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p3_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p3_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_p1_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m1_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m3_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m3_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m5_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m5_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m5_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m5_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m5_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m5_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m5_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m5_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m5_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m5_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m5_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m5_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m5_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m5_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m5_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m5_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m7_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m7_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m7_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m7_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m7_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m7_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m7_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m7_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m7_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m7_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m7_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m7_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m7_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m7_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_r_m7_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m7_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + + interference_abs_64qam_epi16(psi_i_p7_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p7_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p7_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p7_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p7_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p7_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p7_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p7_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p7_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p7_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p7_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p7_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p7_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p7_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p7_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p7_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p5_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p5_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p5_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p5_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p5_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p5_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p5_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p5_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p5_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p5_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p5_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p5_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p5_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p5_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p5_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p5_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p3_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p3_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_p1_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m1_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m3_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m3_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m5_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m5_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m5_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m5_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m5_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m5_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m5_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m5_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m5_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m5_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m5_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m5_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m5_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m5_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m5_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m5_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m7_p7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m7_p7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m7_p5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m7_p5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m7_p3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m7_p3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m7_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m7_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m7_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m7_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m7_m3, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m7_m3, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m7_m5, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m7_m5, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + interference_abs_64qam_epi16(psi_i_m7_m7, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m7_m7, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, + SEVEN_OVER_SQRT_2_42); + + // Calculation of a group of two terms in the bit metric involving product of psi and interference + prodsum_psi_a_epi16(psi_r_p7_p7, a_r_p7_p7, psi_i_p7_p7, a_i_p7_p7, psi_a_p7_p7); + prodsum_psi_a_epi16(psi_r_p7_p5, a_r_p7_p5, psi_i_p7_p5, a_i_p7_p5, psi_a_p7_p5); + prodsum_psi_a_epi16(psi_r_p7_p3, a_r_p7_p3, psi_i_p7_p3, a_i_p7_p3, psi_a_p7_p3); + prodsum_psi_a_epi16(psi_r_p7_p1, a_r_p7_p1, psi_i_p7_p1, a_i_p7_p1, psi_a_p7_p1); + prodsum_psi_a_epi16(psi_r_p7_m1, a_r_p7_m1, psi_i_p7_m1, a_i_p7_m1, psi_a_p7_m1); + prodsum_psi_a_epi16(psi_r_p7_m3, a_r_p7_m3, psi_i_p7_m3, a_i_p7_m3, psi_a_p7_m3); + prodsum_psi_a_epi16(psi_r_p7_m5, a_r_p7_m5, psi_i_p7_m5, a_i_p7_m5, psi_a_p7_m5); + prodsum_psi_a_epi16(psi_r_p7_m7, a_r_p7_m7, psi_i_p7_m7, a_i_p7_m7, psi_a_p7_m7); + prodsum_psi_a_epi16(psi_r_p5_p7, a_r_p5_p7, psi_i_p5_p7, a_i_p5_p7, psi_a_p5_p7); + prodsum_psi_a_epi16(psi_r_p5_p5, a_r_p5_p5, psi_i_p5_p5, a_i_p5_p5, psi_a_p5_p5); + prodsum_psi_a_epi16(psi_r_p5_p3, a_r_p5_p3, psi_i_p5_p3, a_i_p5_p3, psi_a_p5_p3); + prodsum_psi_a_epi16(psi_r_p5_p1, a_r_p5_p1, psi_i_p5_p1, a_i_p5_p1, psi_a_p5_p1); + prodsum_psi_a_epi16(psi_r_p5_m1, a_r_p5_m1, psi_i_p5_m1, a_i_p5_m1, psi_a_p5_m1); + prodsum_psi_a_epi16(psi_r_p5_m3, a_r_p5_m3, psi_i_p5_m3, a_i_p5_m3, psi_a_p5_m3); + prodsum_psi_a_epi16(psi_r_p5_m5, a_r_p5_m5, psi_i_p5_m5, a_i_p5_m5, psi_a_p5_m5); + prodsum_psi_a_epi16(psi_r_p5_m7, a_r_p5_m7, psi_i_p5_m7, a_i_p5_m7, psi_a_p5_m7); + prodsum_psi_a_epi16(psi_r_p3_p7, a_r_p3_p7, psi_i_p3_p7, a_i_p3_p7, psi_a_p3_p7); + prodsum_psi_a_epi16(psi_r_p3_p5, a_r_p3_p5, psi_i_p3_p5, a_i_p3_p5, psi_a_p3_p5); + prodsum_psi_a_epi16(psi_r_p3_p3, a_r_p3_p3, psi_i_p3_p3, a_i_p3_p3, psi_a_p3_p3); + prodsum_psi_a_epi16(psi_r_p3_p1, a_r_p3_p1, psi_i_p3_p1, a_i_p3_p1, psi_a_p3_p1); + prodsum_psi_a_epi16(psi_r_p3_m1, a_r_p3_m1, psi_i_p3_m1, a_i_p3_m1, psi_a_p3_m1); + prodsum_psi_a_epi16(psi_r_p3_m3, a_r_p3_m3, psi_i_p3_m3, a_i_p3_m3, psi_a_p3_m3); + prodsum_psi_a_epi16(psi_r_p3_m5, a_r_p3_m5, psi_i_p3_m5, a_i_p3_m5, psi_a_p3_m5); + prodsum_psi_a_epi16(psi_r_p3_m7, a_r_p3_m7, psi_i_p3_m7, a_i_p3_m7, psi_a_p3_m7); + prodsum_psi_a_epi16(psi_r_p1_p7, a_r_p1_p7, psi_i_p1_p7, a_i_p1_p7, psi_a_p1_p7); + prodsum_psi_a_epi16(psi_r_p1_p5, a_r_p1_p5, psi_i_p1_p5, a_i_p1_p5, psi_a_p1_p5); + prodsum_psi_a_epi16(psi_r_p1_p3, a_r_p1_p3, psi_i_p1_p3, a_i_p1_p3, psi_a_p1_p3); + prodsum_psi_a_epi16(psi_r_p1_p1, a_r_p1_p1, psi_i_p1_p1, a_i_p1_p1, psi_a_p1_p1); + prodsum_psi_a_epi16(psi_r_p1_m1, a_r_p1_m1, psi_i_p1_m1, a_i_p1_m1, psi_a_p1_m1); + prodsum_psi_a_epi16(psi_r_p1_m3, a_r_p1_m3, psi_i_p1_m3, a_i_p1_m3, psi_a_p1_m3); + prodsum_psi_a_epi16(psi_r_p1_m5, a_r_p1_m5, psi_i_p1_m5, a_i_p1_m5, psi_a_p1_m5); + prodsum_psi_a_epi16(psi_r_p1_m7, a_r_p1_m7, psi_i_p1_m7, a_i_p1_m7, psi_a_p1_m7); + prodsum_psi_a_epi16(psi_r_m1_p7, a_r_m1_p7, psi_i_m1_p7, a_i_m1_p7, psi_a_m1_p7); + prodsum_psi_a_epi16(psi_r_m1_p5, a_r_m1_p5, psi_i_m1_p5, a_i_m1_p5, psi_a_m1_p5); + prodsum_psi_a_epi16(psi_r_m1_p3, a_r_m1_p3, psi_i_m1_p3, a_i_m1_p3, psi_a_m1_p3); + prodsum_psi_a_epi16(psi_r_m1_p1, a_r_m1_p1, psi_i_m1_p1, a_i_m1_p1, psi_a_m1_p1); + prodsum_psi_a_epi16(psi_r_m1_m1, a_r_m1_m1, psi_i_m1_m1, a_i_m1_m1, psi_a_m1_m1); + prodsum_psi_a_epi16(psi_r_m1_m3, a_r_m1_m3, psi_i_m1_m3, a_i_m1_m3, psi_a_m1_m3); + prodsum_psi_a_epi16(psi_r_m1_m5, a_r_m1_m5, psi_i_m1_m5, a_i_m1_m5, psi_a_m1_m5); + prodsum_psi_a_epi16(psi_r_m1_m7, a_r_m1_m7, psi_i_m1_m7, a_i_m1_m7, psi_a_m1_m7); + prodsum_psi_a_epi16(psi_r_m3_p7, a_r_m3_p7, psi_i_m3_p7, a_i_m3_p7, psi_a_m3_p7); + prodsum_psi_a_epi16(psi_r_m3_p5, a_r_m3_p5, psi_i_m3_p5, a_i_m3_p5, psi_a_m3_p5); + prodsum_psi_a_epi16(psi_r_m3_p3, a_r_m3_p3, psi_i_m3_p3, a_i_m3_p3, psi_a_m3_p3); + prodsum_psi_a_epi16(psi_r_m3_p1, a_r_m3_p1, psi_i_m3_p1, a_i_m3_p1, psi_a_m3_p1); + prodsum_psi_a_epi16(psi_r_m3_m1, a_r_m3_m1, psi_i_m3_m1, a_i_m3_m1, psi_a_m3_m1); + prodsum_psi_a_epi16(psi_r_m3_m3, a_r_m3_m3, psi_i_m3_m3, a_i_m3_m3, psi_a_m3_m3); + prodsum_psi_a_epi16(psi_r_m3_m5, a_r_m3_m5, psi_i_m3_m5, a_i_m3_m5, psi_a_m3_m5); + prodsum_psi_a_epi16(psi_r_m3_m7, a_r_m3_m7, psi_i_m3_m7, a_i_m3_m7, psi_a_m3_m7); + prodsum_psi_a_epi16(psi_r_m5_p7, a_r_m5_p7, psi_i_m5_p7, a_i_m5_p7, psi_a_m5_p7); + prodsum_psi_a_epi16(psi_r_m5_p5, a_r_m5_p5, psi_i_m5_p5, a_i_m5_p5, psi_a_m5_p5); + prodsum_psi_a_epi16(psi_r_m5_p3, a_r_m5_p3, psi_i_m5_p3, a_i_m5_p3, psi_a_m5_p3); + prodsum_psi_a_epi16(psi_r_m5_p1, a_r_m5_p1, psi_i_m5_p1, a_i_m5_p1, psi_a_m5_p1); + prodsum_psi_a_epi16(psi_r_m5_m1, a_r_m5_m1, psi_i_m5_m1, a_i_m5_m1, psi_a_m5_m1); + prodsum_psi_a_epi16(psi_r_m5_m3, a_r_m5_m3, psi_i_m5_m3, a_i_m5_m3, psi_a_m5_m3); + prodsum_psi_a_epi16(psi_r_m5_m5, a_r_m5_m5, psi_i_m5_m5, a_i_m5_m5, psi_a_m5_m5); + prodsum_psi_a_epi16(psi_r_m5_m7, a_r_m5_m7, psi_i_m5_m7, a_i_m5_m7, psi_a_m5_m7); + prodsum_psi_a_epi16(psi_r_m7_p7, a_r_m7_p7, psi_i_m7_p7, a_i_m7_p7, psi_a_m7_p7); + prodsum_psi_a_epi16(psi_r_m7_p5, a_r_m7_p5, psi_i_m7_p5, a_i_m7_p5, psi_a_m7_p5); + prodsum_psi_a_epi16(psi_r_m7_p3, a_r_m7_p3, psi_i_m7_p3, a_i_m7_p3, psi_a_m7_p3); + prodsum_psi_a_epi16(psi_r_m7_p1, a_r_m7_p1, psi_i_m7_p1, a_i_m7_p1, psi_a_m7_p1); + prodsum_psi_a_epi16(psi_r_m7_m1, a_r_m7_m1, psi_i_m7_m1, a_i_m7_m1, psi_a_m7_m1); + prodsum_psi_a_epi16(psi_r_m7_m3, a_r_m7_m3, psi_i_m7_m3, a_i_m7_m3, psi_a_m7_m3); + prodsum_psi_a_epi16(psi_r_m7_m5, a_r_m7_m5, psi_i_m7_m5, a_i_m7_m5, psi_a_m7_m5); + prodsum_psi_a_epi16(psi_r_m7_m7, a_r_m7_m7, psi_i_m7_m7, a_i_m7_m7, psi_a_m7_m7); + + // Multiply by sqrt(2) + psi_a_p7_p7 = _mm_mulhi_epi16(psi_a_p7_p7, ONE_OVER_SQRT_2); + psi_a_p7_p7 = _mm_slli_epi16(psi_a_p7_p7, 2); + psi_a_p7_p5 = _mm_mulhi_epi16(psi_a_p7_p5, ONE_OVER_SQRT_2); + psi_a_p7_p5 = _mm_slli_epi16(psi_a_p7_p5, 2); + psi_a_p7_p3 = _mm_mulhi_epi16(psi_a_p7_p3, ONE_OVER_SQRT_2); + psi_a_p7_p3 = _mm_slli_epi16(psi_a_p7_p3, 2); + psi_a_p7_p1 = _mm_mulhi_epi16(psi_a_p7_p1, ONE_OVER_SQRT_2); + psi_a_p7_p1 = _mm_slli_epi16(psi_a_p7_p1, 2); + psi_a_p7_m1 = _mm_mulhi_epi16(psi_a_p7_m1, ONE_OVER_SQRT_2); + psi_a_p7_m1 = _mm_slli_epi16(psi_a_p7_m1, 2); + psi_a_p7_m3 = _mm_mulhi_epi16(psi_a_p7_m3, ONE_OVER_SQRT_2); + psi_a_p7_m3 = _mm_slli_epi16(psi_a_p7_m3, 2); + psi_a_p7_m5 = _mm_mulhi_epi16(psi_a_p7_m5, ONE_OVER_SQRT_2); + psi_a_p7_m5 = _mm_slli_epi16(psi_a_p7_m5, 2); + psi_a_p7_m7 = _mm_mulhi_epi16(psi_a_p7_m7, ONE_OVER_SQRT_2); + psi_a_p7_m7 = _mm_slli_epi16(psi_a_p7_m7, 2); + psi_a_p5_p7 = _mm_mulhi_epi16(psi_a_p5_p7, ONE_OVER_SQRT_2); + psi_a_p5_p7 = _mm_slli_epi16(psi_a_p5_p7, 2); + psi_a_p5_p5 = _mm_mulhi_epi16(psi_a_p5_p5, ONE_OVER_SQRT_2); + psi_a_p5_p5 = _mm_slli_epi16(psi_a_p5_p5, 2); + psi_a_p5_p3 = _mm_mulhi_epi16(psi_a_p5_p3, ONE_OVER_SQRT_2); + psi_a_p5_p3 = _mm_slli_epi16(psi_a_p5_p3, 2); + psi_a_p5_p1 = _mm_mulhi_epi16(psi_a_p5_p1, ONE_OVER_SQRT_2); + psi_a_p5_p1 = _mm_slli_epi16(psi_a_p5_p1, 2); + psi_a_p5_m1 = _mm_mulhi_epi16(psi_a_p5_m1, ONE_OVER_SQRT_2); + psi_a_p5_m1 = _mm_slli_epi16(psi_a_p5_m1, 2); + psi_a_p5_m3 = _mm_mulhi_epi16(psi_a_p5_m3, ONE_OVER_SQRT_2); + psi_a_p5_m3 = _mm_slli_epi16(psi_a_p5_m3, 2); + psi_a_p5_m5 = _mm_mulhi_epi16(psi_a_p5_m5, ONE_OVER_SQRT_2); + psi_a_p5_m5 = _mm_slli_epi16(psi_a_p5_m5, 2); + psi_a_p5_m7 = _mm_mulhi_epi16(psi_a_p5_m7, ONE_OVER_SQRT_2); + psi_a_p5_m7 = _mm_slli_epi16(psi_a_p5_m7, 2); + psi_a_p3_p7 = _mm_mulhi_epi16(psi_a_p3_p7, ONE_OVER_SQRT_2); + psi_a_p3_p7 = _mm_slli_epi16(psi_a_p3_p7, 2); + psi_a_p3_p5 = _mm_mulhi_epi16(psi_a_p3_p5, ONE_OVER_SQRT_2); + psi_a_p3_p5 = _mm_slli_epi16(psi_a_p3_p5, 2); + psi_a_p3_p3 = _mm_mulhi_epi16(psi_a_p3_p3, ONE_OVER_SQRT_2); + psi_a_p3_p3 = _mm_slli_epi16(psi_a_p3_p3, 2); + psi_a_p3_p1 = _mm_mulhi_epi16(psi_a_p3_p1, ONE_OVER_SQRT_2); + psi_a_p3_p1 = _mm_slli_epi16(psi_a_p3_p1, 2); + psi_a_p3_m1 = _mm_mulhi_epi16(psi_a_p3_m1, ONE_OVER_SQRT_2); + psi_a_p3_m1 = _mm_slli_epi16(psi_a_p3_m1, 2); + psi_a_p3_m3 = _mm_mulhi_epi16(psi_a_p3_m3, ONE_OVER_SQRT_2); + psi_a_p3_m3 = _mm_slli_epi16(psi_a_p3_m3, 2); + psi_a_p3_m5 = _mm_mulhi_epi16(psi_a_p3_m5, ONE_OVER_SQRT_2); + psi_a_p3_m5 = _mm_slli_epi16(psi_a_p3_m5, 2); + psi_a_p3_m7 = _mm_mulhi_epi16(psi_a_p3_m7, ONE_OVER_SQRT_2); + psi_a_p3_m7 = _mm_slli_epi16(psi_a_p3_m7, 2); + psi_a_p1_p7 = _mm_mulhi_epi16(psi_a_p1_p7, ONE_OVER_SQRT_2); + psi_a_p1_p7 = _mm_slli_epi16(psi_a_p1_p7, 2); + psi_a_p1_p5 = _mm_mulhi_epi16(psi_a_p1_p5, ONE_OVER_SQRT_2); + psi_a_p1_p5 = _mm_slli_epi16(psi_a_p1_p5, 2); + psi_a_p1_p3 = _mm_mulhi_epi16(psi_a_p1_p3, ONE_OVER_SQRT_2); + psi_a_p1_p3 = _mm_slli_epi16(psi_a_p1_p3, 2); + psi_a_p1_p1 = _mm_mulhi_epi16(psi_a_p1_p1, ONE_OVER_SQRT_2); + psi_a_p1_p1 = _mm_slli_epi16(psi_a_p1_p1, 2); + psi_a_p1_m1 = _mm_mulhi_epi16(psi_a_p1_m1, ONE_OVER_SQRT_2); + psi_a_p1_m1 = _mm_slli_epi16(psi_a_p1_m1, 2); + psi_a_p1_m3 = _mm_mulhi_epi16(psi_a_p1_m3, ONE_OVER_SQRT_2); + psi_a_p1_m3 = _mm_slli_epi16(psi_a_p1_m3, 2); + psi_a_p1_m5 = _mm_mulhi_epi16(psi_a_p1_m5, ONE_OVER_SQRT_2); + psi_a_p1_m5 = _mm_slli_epi16(psi_a_p1_m5, 2); + psi_a_p1_m7 = _mm_mulhi_epi16(psi_a_p1_m7, ONE_OVER_SQRT_2); + psi_a_p1_m7 = _mm_slli_epi16(psi_a_p1_m7, 2); + psi_a_m1_p7 = _mm_mulhi_epi16(psi_a_m1_p7, ONE_OVER_SQRT_2); + psi_a_m1_p7 = _mm_slli_epi16(psi_a_m1_p7, 2); + psi_a_m1_p5 = _mm_mulhi_epi16(psi_a_m1_p5, ONE_OVER_SQRT_2); + psi_a_m1_p5 = _mm_slli_epi16(psi_a_m1_p5, 2); + psi_a_m1_p3 = _mm_mulhi_epi16(psi_a_m1_p3, ONE_OVER_SQRT_2); + psi_a_m1_p3 = _mm_slli_epi16(psi_a_m1_p3, 2); + psi_a_m1_p1 = _mm_mulhi_epi16(psi_a_m1_p1, ONE_OVER_SQRT_2); + psi_a_m1_p1 = _mm_slli_epi16(psi_a_m1_p1, 2); + psi_a_m1_m1 = _mm_mulhi_epi16(psi_a_m1_m1, ONE_OVER_SQRT_2); + psi_a_m1_m1 = _mm_slli_epi16(psi_a_m1_m1, 2); + psi_a_m1_m3 = _mm_mulhi_epi16(psi_a_m1_m3, ONE_OVER_SQRT_2); + psi_a_m1_m3 = _mm_slli_epi16(psi_a_m1_m3, 2); + psi_a_m1_m5 = _mm_mulhi_epi16(psi_a_m1_m5, ONE_OVER_SQRT_2); + psi_a_m1_m5 = _mm_slli_epi16(psi_a_m1_m5, 2); + psi_a_m1_m7 = _mm_mulhi_epi16(psi_a_m1_m7, ONE_OVER_SQRT_2); + psi_a_m1_m7 = _mm_slli_epi16(psi_a_m1_m7, 2); + psi_a_m3_p7 = _mm_mulhi_epi16(psi_a_m3_p7, ONE_OVER_SQRT_2); + psi_a_m3_p7 = _mm_slli_epi16(psi_a_m3_p7, 2); + psi_a_m3_p5 = _mm_mulhi_epi16(psi_a_m3_p5, ONE_OVER_SQRT_2); + psi_a_m3_p5 = _mm_slli_epi16(psi_a_m3_p5, 2); + psi_a_m3_p3 = _mm_mulhi_epi16(psi_a_m3_p3, ONE_OVER_SQRT_2); + psi_a_m3_p3 = _mm_slli_epi16(psi_a_m3_p3, 2); + psi_a_m3_p1 = _mm_mulhi_epi16(psi_a_m3_p1, ONE_OVER_SQRT_2); + psi_a_m3_p1 = _mm_slli_epi16(psi_a_m3_p1, 2); + psi_a_m3_m1 = _mm_mulhi_epi16(psi_a_m3_m1, ONE_OVER_SQRT_2); + psi_a_m3_m1 = _mm_slli_epi16(psi_a_m3_m1, 2); + psi_a_m3_m3 = _mm_mulhi_epi16(psi_a_m3_m3, ONE_OVER_SQRT_2); + psi_a_m3_m3 = _mm_slli_epi16(psi_a_m3_m3, 2); + psi_a_m3_m5 = _mm_mulhi_epi16(psi_a_m3_m5, ONE_OVER_SQRT_2); + psi_a_m3_m5 = _mm_slli_epi16(psi_a_m3_m5, 2); + psi_a_m3_m7 = _mm_mulhi_epi16(psi_a_m3_m7, ONE_OVER_SQRT_2); + psi_a_m3_m7 = _mm_slli_epi16(psi_a_m3_m7, 2); + psi_a_m5_p7 = _mm_mulhi_epi16(psi_a_m5_p7, ONE_OVER_SQRT_2); + psi_a_m5_p7 = _mm_slli_epi16(psi_a_m5_p7, 2); + psi_a_m5_p5 = _mm_mulhi_epi16(psi_a_m5_p5, ONE_OVER_SQRT_2); + psi_a_m5_p5 = _mm_slli_epi16(psi_a_m5_p5, 2); + psi_a_m5_p3 = _mm_mulhi_epi16(psi_a_m5_p3, ONE_OVER_SQRT_2); + psi_a_m5_p3 = _mm_slli_epi16(psi_a_m5_p3, 2); + psi_a_m5_p1 = _mm_mulhi_epi16(psi_a_m5_p1, ONE_OVER_SQRT_2); + psi_a_m5_p1 = _mm_slli_epi16(psi_a_m5_p1, 2); + psi_a_m5_m1 = _mm_mulhi_epi16(psi_a_m5_m1, ONE_OVER_SQRT_2); + psi_a_m5_m1 = _mm_slli_epi16(psi_a_m5_m1, 2); + psi_a_m5_m3 = _mm_mulhi_epi16(psi_a_m5_m3, ONE_OVER_SQRT_2); + psi_a_m5_m3 = _mm_slli_epi16(psi_a_m5_m3, 2); + psi_a_m5_m5 = _mm_mulhi_epi16(psi_a_m5_m5, ONE_OVER_SQRT_2); + psi_a_m5_m5 = _mm_slli_epi16(psi_a_m5_m5, 2); + psi_a_m5_m7 = _mm_mulhi_epi16(psi_a_m5_m7, ONE_OVER_SQRT_2); + psi_a_m5_m7 = _mm_slli_epi16(psi_a_m5_m7, 2); + psi_a_m7_p7 = _mm_mulhi_epi16(psi_a_m7_p7, ONE_OVER_SQRT_2); + psi_a_m7_p7 = _mm_slli_epi16(psi_a_m7_p7, 2); + psi_a_m7_p5 = _mm_mulhi_epi16(psi_a_m7_p5, ONE_OVER_SQRT_2); + psi_a_m7_p5 = _mm_slli_epi16(psi_a_m7_p5, 2); + psi_a_m7_p3 = _mm_mulhi_epi16(psi_a_m7_p3, ONE_OVER_SQRT_2); + psi_a_m7_p3 = _mm_slli_epi16(psi_a_m7_p3, 2); + psi_a_m7_p1 = _mm_mulhi_epi16(psi_a_m7_p1, ONE_OVER_SQRT_2); + psi_a_m7_p1 = _mm_slli_epi16(psi_a_m7_p1, 2); + psi_a_m7_m1 = _mm_mulhi_epi16(psi_a_m7_m1, ONE_OVER_SQRT_2); + psi_a_m7_m1 = _mm_slli_epi16(psi_a_m7_m1, 2); + psi_a_m7_m3 = _mm_mulhi_epi16(psi_a_m7_m3, ONE_OVER_SQRT_2); + psi_a_m7_m3 = _mm_slli_epi16(psi_a_m7_m3, 2); + psi_a_m7_m5 = _mm_mulhi_epi16(psi_a_m7_m5, ONE_OVER_SQRT_2); + psi_a_m7_m5 = _mm_slli_epi16(psi_a_m7_m5, 2); + psi_a_m7_m7 = _mm_mulhi_epi16(psi_a_m7_m7, ONE_OVER_SQRT_2); + psi_a_m7_m7 = _mm_slli_epi16(psi_a_m7_m7, 2); + + // Calculation of a group of two terms in the bit metric involving squares of interference + square_a_64qam_epi16(a_r_p7_p7, a_i_p7_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_p7); + square_a_64qam_epi16(a_r_p7_p5, a_i_p7_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_p5); + square_a_64qam_epi16(a_r_p7_p3, a_i_p7_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_p3); + square_a_64qam_epi16(a_r_p7_p1, a_i_p7_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_p1); + square_a_64qam_epi16(a_r_p7_m1, a_i_p7_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_m1); + square_a_64qam_epi16(a_r_p7_m3, a_i_p7_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_m3); + square_a_64qam_epi16(a_r_p7_m5, a_i_p7_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_m5); + square_a_64qam_epi16(a_r_p7_m7, a_i_p7_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_m7); + square_a_64qam_epi16(a_r_p5_p7, a_i_p5_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_p7); + square_a_64qam_epi16(a_r_p5_p5, a_i_p5_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_p5); + square_a_64qam_epi16(a_r_p5_p3, a_i_p5_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_p3); + square_a_64qam_epi16(a_r_p5_p1, a_i_p5_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_p1); + square_a_64qam_epi16(a_r_p5_m1, a_i_p5_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_m1); + square_a_64qam_epi16(a_r_p5_m3, a_i_p5_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_m3); + square_a_64qam_epi16(a_r_p5_m5, a_i_p5_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_m5); + square_a_64qam_epi16(a_r_p5_m7, a_i_p5_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_m7); + square_a_64qam_epi16(a_r_p3_p7, a_i_p3_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_p7); + square_a_64qam_epi16(a_r_p3_p5, a_i_p3_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_p5); + square_a_64qam_epi16(a_r_p3_p3, a_i_p3_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_p3); + square_a_64qam_epi16(a_r_p3_p1, a_i_p3_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_p1); + square_a_64qam_epi16(a_r_p3_m1, a_i_p3_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_m1); + square_a_64qam_epi16(a_r_p3_m3, a_i_p3_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_m3); + square_a_64qam_epi16(a_r_p3_m5, a_i_p3_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_m5); + square_a_64qam_epi16(a_r_p3_m7, a_i_p3_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_m7); + square_a_64qam_epi16(a_r_p1_p7, a_i_p1_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_p7); + square_a_64qam_epi16(a_r_p1_p5, a_i_p1_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_p5); + square_a_64qam_epi16(a_r_p1_p3, a_i_p1_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_p3); + square_a_64qam_epi16(a_r_p1_p1, a_i_p1_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_p1); + square_a_64qam_epi16(a_r_p1_m1, a_i_p1_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_m1); + square_a_64qam_epi16(a_r_p1_m3, a_i_p1_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_m3); + square_a_64qam_epi16(a_r_p1_m5, a_i_p1_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_m5); + square_a_64qam_epi16(a_r_p1_m7, a_i_p1_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_m7); + square_a_64qam_epi16(a_r_m1_p7, a_i_m1_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_p7); + square_a_64qam_epi16(a_r_m1_p5, a_i_m1_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_p5); + square_a_64qam_epi16(a_r_m1_p3, a_i_m1_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_p3); + square_a_64qam_epi16(a_r_m1_p1, a_i_m1_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_p1); + square_a_64qam_epi16(a_r_m1_m1, a_i_m1_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_m1); + square_a_64qam_epi16(a_r_m1_m3, a_i_m1_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_m3); + square_a_64qam_epi16(a_r_m1_m5, a_i_m1_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_m5); + square_a_64qam_epi16(a_r_m1_m7, a_i_m1_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_m7); + square_a_64qam_epi16(a_r_m3_p7, a_i_m3_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_p7); + square_a_64qam_epi16(a_r_m3_p5, a_i_m3_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_p5); + square_a_64qam_epi16(a_r_m3_p3, a_i_m3_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_p3); + square_a_64qam_epi16(a_r_m3_p1, a_i_m3_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_p1); + square_a_64qam_epi16(a_r_m3_m1, a_i_m3_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_m1); + square_a_64qam_epi16(a_r_m3_m3, a_i_m3_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_m3); + square_a_64qam_epi16(a_r_m3_m5, a_i_m3_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_m5); + square_a_64qam_epi16(a_r_m3_m7, a_i_m3_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_m7); + square_a_64qam_epi16(a_r_m5_p7, a_i_m5_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_p7); + square_a_64qam_epi16(a_r_m5_p5, a_i_m5_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_p5); + square_a_64qam_epi16(a_r_m5_p3, a_i_m5_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_p3); + square_a_64qam_epi16(a_r_m5_p1, a_i_m5_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_p1); + square_a_64qam_epi16(a_r_m5_m1, a_i_m5_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_m1); + square_a_64qam_epi16(a_r_m5_m3, a_i_m5_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_m3); + square_a_64qam_epi16(a_r_m5_m5, a_i_m5_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_m5); + square_a_64qam_epi16(a_r_m5_m7, a_i_m5_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_m7); + square_a_64qam_epi16(a_r_m7_p7, a_i_m7_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_p7); + square_a_64qam_epi16(a_r_m7_p5, a_i_m7_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_p5); + square_a_64qam_epi16(a_r_m7_p3, a_i_m7_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_p3); + square_a_64qam_epi16(a_r_m7_p1, a_i_m7_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_p1); + square_a_64qam_epi16(a_r_m7_m1, a_i_m7_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_m1); + square_a_64qam_epi16(a_r_m7_m3, a_i_m7_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_m3); + square_a_64qam_epi16(a_r_m7_m5, a_i_m7_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_m5); + square_a_64qam_epi16(a_r_m7_m7, a_i_m7_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_m7); + + // Computing different multiples of ||h0||^2 + // x=1, y=1 + ch_mag_2_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,ONE_OVER_FOUR_SQRT_42); + ch_mag_2_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_2_over_42_with_sigma2,1); + // x=1, y=3 + ch_mag_10_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,FIVE_OVER_FOUR_SQRT_42); + ch_mag_10_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_10_over_42_with_sigma2,1); + // x=1, x=5 + ch_mag_26_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,THIRTEEN_OVER_FOUR_SQRT_42); + ch_mag_26_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_26_over_42_with_sigma2,1); + // x=1, y=7 + ch_mag_50_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,TWENTYFIVE_OVER_FOUR_SQRT_42); + ch_mag_50_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_50_over_42_with_sigma2,1); + // x=3, y=3 + ch_mag_18_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,NINE_OVER_FOUR_SQRT_42); + ch_mag_18_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_18_over_42_with_sigma2,1); + // x=3, y=5 + ch_mag_34_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,SEVENTEEN_OVER_FOUR_SQRT_42); + ch_mag_34_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_34_over_42_with_sigma2,1); + // x=3, y=7 + ch_mag_58_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,TWENTYNINE_OVER_FOUR_SQRT_42); + ch_mag_58_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_58_over_42_with_sigma2,2); + // x=5, y=5 + ch_mag_50_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,TWENTYFIVE_OVER_FOUR_SQRT_42); + ch_mag_50_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_50_over_42_with_sigma2,1); + // x=5, y=7 + ch_mag_74_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,THIRTYSEVEN_OVER_FOUR_SQRT_42); + ch_mag_74_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_74_over_42_with_sigma2,2); + // x=7, y=7 + ch_mag_98_over_42_with_sigma2 = _mm_mulhi_epi16(ch_mag_des,FORTYNINE_OVER_FOUR_SQRT_42); + ch_mag_98_over_42_with_sigma2 = _mm_slli_epi16(ch_mag_98_over_42_with_sigma2,2); + + // Computing Metrics + xmm0 = _mm_subs_epi16(psi_a_p7_p7, a_sq_p7_p7); + xmm1 = _mm_adds_epi16(xmm0, y0_p_7_7); + bit_met_p7_p7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_p5, a_sq_p7_p5); + xmm1 = _mm_adds_epi16(xmm0, y0_p_7_5); + bit_met_p7_p5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_p3, a_sq_p7_p3); + xmm1 = _mm_adds_epi16(xmm0, y0_p_7_3); + bit_met_p7_p3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_p1, a_sq_p7_p1); + xmm1 = _mm_adds_epi16(xmm0, y0_p_7_1); + bit_met_p7_p1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_m1, a_sq_p7_m1); + xmm1 = _mm_adds_epi16(xmm0, y0_m_7_1); + bit_met_p7_m1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_m3, a_sq_p7_m3); + xmm1 = _mm_adds_epi16(xmm0, y0_m_7_3); + bit_met_p7_m3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_m5, a_sq_p7_m5); + xmm1 = _mm_adds_epi16(xmm0, y0_m_7_5); + bit_met_p7_m5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p7_m7, a_sq_p7_m7); + xmm1 = _mm_adds_epi16(xmm0, y0_m_7_7); + bit_met_p7_m7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_p7, a_sq_p5_p7); + xmm1 = _mm_adds_epi16(xmm0, y0_p_5_7); + bit_met_p5_p7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_p5, a_sq_p5_p5); + xmm1 = _mm_adds_epi16(xmm0, y0_p_5_5); + bit_met_p5_p5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_p3, a_sq_p5_p3); + xmm1 = _mm_adds_epi16(xmm0, y0_p_5_3); + bit_met_p5_p3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_p1, a_sq_p5_p1); + xmm1 = _mm_adds_epi16(xmm0, y0_p_5_1); + bit_met_p5_p1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_m1, a_sq_p5_m1); + xmm1 = _mm_adds_epi16(xmm0, y0_m_5_1); + bit_met_p5_m1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_m3, a_sq_p5_m3); + xmm1 = _mm_adds_epi16(xmm0, y0_m_5_3); + bit_met_p5_m3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_m5, a_sq_p5_m5); + xmm1 = _mm_adds_epi16(xmm0, y0_m_5_5); + bit_met_p5_m5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p5_m7, a_sq_p5_m7); + xmm1 = _mm_adds_epi16(xmm0, y0_m_5_7); + bit_met_p5_m7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_p7, a_sq_p3_p7); + xmm1 = _mm_adds_epi16(xmm0, y0_p_3_7); + bit_met_p3_p7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_p5, a_sq_p3_p5); + xmm1 = _mm_adds_epi16(xmm0, y0_p_3_5); + bit_met_p3_p5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_p3, a_sq_p3_p3); + xmm1 = _mm_adds_epi16(xmm0, y0_p_3_3); + bit_met_p3_p3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_p1, a_sq_p3_p1); + xmm1 = _mm_adds_epi16(xmm0, y0_p_3_1); + bit_met_p3_p1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_m1, a_sq_p3_m1); + xmm1 = _mm_adds_epi16(xmm0, y0_m_3_1); + bit_met_p3_m1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_m3, a_sq_p3_m3); + xmm1 = _mm_adds_epi16(xmm0, y0_m_3_3); + bit_met_p3_m3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_m5, a_sq_p3_m5); + xmm1 = _mm_adds_epi16(xmm0, y0_m_3_5); + bit_met_p3_m5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p3_m7, a_sq_p3_m7); + xmm1 = _mm_adds_epi16(xmm0, y0_m_3_7); + bit_met_p3_m7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_p7, a_sq_p1_p7); + xmm1 = _mm_adds_epi16(xmm0, y0_p_1_7); + bit_met_p1_p7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_p5, a_sq_p1_p5); + xmm1 = _mm_adds_epi16(xmm0, y0_p_1_5); + bit_met_p1_p5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_p3, a_sq_p1_p3); + xmm1 = _mm_adds_epi16(xmm0, y0_p_1_3); + bit_met_p1_p3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_p1, a_sq_p1_p1); + xmm1 = _mm_adds_epi16(xmm0, y0_p_1_1); + bit_met_p1_p1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_m1, a_sq_p1_m1); + xmm1 = _mm_adds_epi16(xmm0, y0_m_1_1); + bit_met_p1_m1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_m3, a_sq_p1_m3); + xmm1 = _mm_adds_epi16(xmm0, y0_m_1_3); + bit_met_p1_m3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_m5, a_sq_p1_m5); + xmm1 = _mm_adds_epi16(xmm0, y0_m_1_5); + bit_met_p1_m5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_p1_m7, a_sq_p1_m7); + xmm1 = _mm_adds_epi16(xmm0, y0_m_1_7); + bit_met_p1_m7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + + xmm0 = _mm_subs_epi16(psi_a_m1_p7, a_sq_m1_p7); + xmm1 = _mm_subs_epi16(xmm0, y0_m_1_7); + bit_met_m1_p7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_p5, a_sq_m1_p5); + xmm1 = _mm_subs_epi16(xmm0, y0_m_1_5); + bit_met_m1_p5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_p3, a_sq_m1_p3); + xmm1 = _mm_subs_epi16(xmm0, y0_m_1_3); + bit_met_m1_p3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_p1, a_sq_m1_p1); + xmm1 = _mm_subs_epi16(xmm0, y0_m_1_1); + bit_met_m1_p1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_m1, a_sq_m1_m1); + xmm1 = _mm_subs_epi16(xmm0, y0_p_1_1); + bit_met_m1_m1 = _mm_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_m3, a_sq_m1_m3); + xmm1 = _mm_subs_epi16(xmm0, y0_p_1_3); + bit_met_m1_m3 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_m5, a_sq_m1_m5); + xmm1 = _mm_subs_epi16(xmm0, y0_p_1_5); + bit_met_m1_m5 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m1_m7, a_sq_m1_m7); + xmm1 = _mm_subs_epi16(xmm0, y0_p_1_7); + bit_met_m1_m7 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_p7, a_sq_m3_p7); + xmm1 = _mm_subs_epi16(xmm0, y0_m_3_7); + bit_met_m3_p7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_p5, a_sq_m3_p5); + xmm1 = _mm_subs_epi16(xmm0, y0_m_3_5); + bit_met_m3_p5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_p3, a_sq_m3_p3); + xmm1 = _mm_subs_epi16(xmm0, y0_m_3_3); + bit_met_m3_p3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_p1, a_sq_m3_p1); + xmm1 = _mm_subs_epi16(xmm0, y0_m_3_1); + bit_met_m3_p1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_m1, a_sq_m3_m1); + xmm1 = _mm_subs_epi16(xmm0, y0_p_3_1); + bit_met_m3_m1 = _mm_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_m3, a_sq_m3_m3); + xmm1 = _mm_subs_epi16(xmm0, y0_p_3_3); + bit_met_m3_m3 = _mm_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_m5, a_sq_m3_m5); + xmm1 = _mm_subs_epi16(xmm0, y0_p_3_5); + bit_met_m3_m5 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m3_m7, a_sq_m3_m7); + xmm1 = _mm_subs_epi16(xmm0, y0_p_3_7); + bit_met_m3_m7 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_p7, a_sq_m5_p7); + xmm1 = _mm_subs_epi16(xmm0, y0_m_5_7); + bit_met_m5_p7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_p5, a_sq_m5_p5); + xmm1 = _mm_subs_epi16(xmm0, y0_m_5_5); + bit_met_m5_p5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_p3, a_sq_m5_p3); + xmm1 = _mm_subs_epi16(xmm0, y0_m_5_3); + bit_met_m5_p3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_p1, a_sq_m5_p1); + xmm1 = _mm_subs_epi16(xmm0, y0_m_5_1); + bit_met_m5_p1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_m1, a_sq_m5_m1); + xmm1 = _mm_subs_epi16(xmm0, y0_p_5_1); + bit_met_m5_m1 = _mm_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_m3, a_sq_m5_m3); + xmm1 = _mm_subs_epi16(xmm0, y0_p_5_3); + bit_met_m5_m3 = _mm_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_m5, a_sq_m5_m5); + xmm1 = _mm_subs_epi16(xmm0, y0_p_5_5); + bit_met_m5_m5 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m5_m7, a_sq_m5_m7); + xmm1 = _mm_subs_epi16(xmm0, y0_p_5_7); + bit_met_m5_m7 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_p7, a_sq_m7_p7); + xmm1 = _mm_subs_epi16(xmm0, y0_m_7_7); + bit_met_m7_p7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_p5, a_sq_m7_p5); + xmm1 = _mm_subs_epi16(xmm0, y0_m_7_5); + bit_met_m7_p5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_p3, a_sq_m7_p3); + xmm1 = _mm_subs_epi16(xmm0, y0_m_7_3); + bit_met_m7_p3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_p1, a_sq_m7_p1); + xmm1 = _mm_subs_epi16(xmm0, y0_m_7_1); + bit_met_m7_p1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_m1, a_sq_m7_m1); + xmm1 = _mm_subs_epi16(xmm0, y0_p_7_1); + bit_met_m7_m1 = _mm_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_m3, a_sq_m7_m3); + xmm1 = _mm_subs_epi16(xmm0, y0_p_7_3); + bit_met_m7_m3 = _mm_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_m5, a_sq_m7_m5); + xmm1 = _mm_subs_epi16(xmm0, y0_p_7_5); + bit_met_m7_m5 = _mm_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = _mm_subs_epi16(psi_a_m7_m7, a_sq_m7_m7); + xmm1 = _mm_subs_epi16(xmm0, y0_p_7_7); + bit_met_m7_m7 = _mm_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + + // Detection for 1st bit (LTE mapping) + // bit = 1 + xmm0 = _mm_max_epi16(bit_met_m7_p7, bit_met_m7_p5); + xmm1 = _mm_max_epi16(bit_met_m7_p3, bit_met_m7_p1); + xmm2 = _mm_max_epi16(bit_met_m7_m1, bit_met_m7_m3); + xmm3 = _mm_max_epi16(bit_met_m7_m5, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m5_p7, bit_met_m5_p5); + xmm1 = _mm_max_epi16(bit_met_m5_p3, bit_met_m5_p1); + xmm2 = _mm_max_epi16(bit_met_m5_m1, bit_met_m5_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m5_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_m3_p7, bit_met_m3_p5); + xmm1 = _mm_max_epi16(bit_met_m3_p3, bit_met_m3_p1); + xmm2 = _mm_max_epi16(bit_met_m3_m1, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m3_m5, bit_met_m3_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_m1_p7, bit_met_m1_p5); + xmm1 = _mm_max_epi16(bit_met_m1_p3, bit_met_m1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m1_m3); + xmm3 = _mm_max_epi16(bit_met_m1_m5, bit_met_m1_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + // bit = 0 + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p7_p5); + xmm1 = _mm_max_epi16(bit_met_p7_p3, bit_met_p7_p1); + xmm2 = _mm_max_epi16(bit_met_p7_m1, bit_met_p7_m3); + xmm3 = _mm_max_epi16(bit_met_p7_m5, bit_met_p7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p5_p7, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p5_p3, bit_met_p5_p1); + xmm2 = _mm_max_epi16(bit_met_p5_m1, bit_met_p5_m3); + xmm3 = _mm_max_epi16(bit_met_p5_m5, bit_met_p5_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p3_p7, bit_met_p3_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p3_p1); + xmm2 = _mm_max_epi16(bit_met_p3_m1, bit_met_p3_m3); + xmm3 = _mm_max_epi16(bit_met_p3_m5, bit_met_p3_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p1_p7, bit_met_p1_p5); + xmm1 = _mm_max_epi16(bit_met_p1_p3, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_p1_m1, bit_met_p1_m3); + xmm3 = _mm_max_epi16(bit_met_p1_m5, bit_met_p1_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y0r = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 2nd bit (LTE mapping) + // bit = 1 + xmm0 = _mm_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = _mm_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = _mm_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = _mm_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = _mm_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = _mm_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = _mm_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + // bit = 0 + xmm0 = _mm_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = _mm_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = _mm_max_epi16(bit_met_m5_p1, bit_met_m7_p1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = _mm_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = _mm_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m7_p5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = _mm_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = _mm_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = _mm_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y1r = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 3rd bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_m7_m7, bit_met_m7_m5); + xmm1 = _mm_max_epi16(bit_met_m7_m3, bit_met_m7_m1); + xmm2 = _mm_max_epi16(bit_met_m7_p1, bit_met_m7_p3); + xmm3 = _mm_max_epi16(bit_met_m7_p5, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m5_m7, bit_met_m5_m5); + xmm1 = _mm_max_epi16(bit_met_m5_m3, bit_met_m5_m1); + xmm2 = _mm_max_epi16(bit_met_m5_p1, bit_met_m5_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p5_m7, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p5_m3, bit_met_p5_m1); + xmm2 = _mm_max_epi16(bit_met_p5_p1, bit_met_p5_p3); + xmm3 = _mm_max_epi16(bit_met_p5_p5, bit_met_p5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p7_m5); + xmm1 = _mm_max_epi16(bit_met_p7_m3, bit_met_p7_m1); + xmm2 = _mm_max_epi16(bit_met_p7_p1, bit_met_p7_p3); + xmm3 = _mm_max_epi16(bit_met_p7_p5, bit_met_p7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_m3_m7, bit_met_m3_m5); + xmm1 = _mm_max_epi16(bit_met_m3_m3, bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_m3_p1, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m3_p5, bit_met_m3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m1_m7, bit_met_m1_m5); + xmm1 = _mm_max_epi16(bit_met_m1_m3, bit_met_m1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m1_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p5, bit_met_m1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p1_m7, bit_met_p1_m5); + xmm1 = _mm_max_epi16(bit_met_p1_m3, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_p1_p1, bit_met_p1_p3); + xmm3 = _mm_max_epi16(bit_met_p1_p5, bit_met_p1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p3_m7, bit_met_p3_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p3_m1); + xmm2 = _mm_max_epi16(bit_met_p3_p1, bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_p5, bit_met_p3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y2r = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 4th bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = _mm_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = _mm_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = _mm_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = _mm_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m5_p5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = _mm_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = _mm_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = _mm_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = _mm_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = _mm_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = _mm_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = _mm_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = _mm_max_epi16(bit_met_m5_p1, bit_met_m7_p1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = _mm_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y0i = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + + // Detection for 5th bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_m7_m7, bit_met_m7_m5); + xmm1 = _mm_max_epi16(bit_met_m7_m3, bit_met_m7_m1); + xmm2 = _mm_max_epi16(bit_met_m7_p1, bit_met_m7_p3); + xmm3 = _mm_max_epi16(bit_met_m7_p5, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m1_m7, bit_met_m1_m5); + xmm1 = _mm_max_epi16(bit_met_m1_m3, bit_met_m1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m1_p3); + xmm3 = _mm_max_epi16(bit_met_m1_p5, bit_met_m1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p1_m7, bit_met_p1_m5); + xmm1 = _mm_max_epi16(bit_met_p1_m3, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_p1_p1, bit_met_p1_p3); + xmm3 = _mm_max_epi16(bit_met_p1_p5, bit_met_p1_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p7_m5); + xmm1 = _mm_max_epi16(bit_met_p7_m3, bit_met_p7_m1); + xmm2 = _mm_max_epi16(bit_met_p7_p1, bit_met_p7_p3); + xmm3 = _mm_max_epi16(bit_met_p7_p5, bit_met_p7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_m5_m7, bit_met_m5_m5); + xmm1 = _mm_max_epi16(bit_met_m5_m3, bit_met_m5_m1); + xmm2 = _mm_max_epi16(bit_met_m5_p1, bit_met_m5_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_m3_m7, bit_met_m3_m5); + xmm1 = _mm_max_epi16(bit_met_m3_m3, bit_met_m3_m1); + xmm2 = _mm_max_epi16(bit_met_m3_p1, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m3_p5, bit_met_m3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p3_m7, bit_met_p3_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p3_m1); + xmm2 = _mm_max_epi16(bit_met_p3_p1, bit_met_p3_p3); + xmm3 = _mm_max_epi16(bit_met_p3_p5, bit_met_p3_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p5_m7, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p5_m3, bit_met_p5_m1); + xmm2 = _mm_max_epi16(bit_met_p5_p1, bit_met_p5_p3); + xmm3 = _mm_max_epi16(bit_met_p5_p5, bit_met_p5_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y1i = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 6th bit (LTE mapping) + xmm0 = _mm_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = _mm_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = _mm_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = _mm_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = _mm_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = _mm_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = _mm_max_epi16(bit_met_m5_p1, bit_met_m5_p1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = _mm_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = _mm_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = _mm_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = _mm_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = _mm_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = _mm_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = _mm_max_epi16(logmax_den_re0, xmm5); + + xmm0 = _mm_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = _mm_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = _mm_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = _mm_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(xmm4, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = _mm_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = _mm_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = _mm_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = _mm_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = _mm_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = _mm_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + xmm0 = _mm_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = _mm_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = _mm_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = _mm_max_epi16(bit_met_m5_p5, bit_met_m7_p5); + xmm4 = _mm_max_epi16(xmm0, xmm1); + xmm5 = _mm_max_epi16(xmm2, xmm3); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = _mm_max_epi16(logmax_num_re0, xmm5); + + y2i = _mm_subs_epi16(logmax_num_re0, logmax_den_re0); + + + // map to output stream, difficult to do in SIMD since we have 6 16bit LLRs + // RE 1 + j = 24*i; + stream0_out[j + 0] = ((short *)&y0r)[0]; + stream0_out[j + 1] = ((short *)&y1r)[0]; + stream0_out[j + 2] = ((short *)&y2r)[0]; + stream0_out[j + 3] = ((short *)&y0i)[0]; + stream0_out[j + 4] = ((short *)&y1i)[0]; + stream0_out[j + 5] = ((short *)&y2i)[0]; + // RE 2 + stream0_out[j + 6] = ((short *)&y0r)[1]; + stream0_out[j + 7] = ((short *)&y1r)[1]; + stream0_out[j + 8] = ((short *)&y2r)[1]; + stream0_out[j + 9] = ((short *)&y0i)[1]; + stream0_out[j + 10] = ((short *)&y1i)[1]; + stream0_out[j + 11] = ((short *)&y2i)[1]; + // RE 3 + stream0_out[j + 12] = ((short *)&y0r)[2]; + stream0_out[j + 13] = ((short *)&y1r)[2]; + stream0_out[j + 14] = ((short *)&y2r)[2]; + stream0_out[j + 15] = ((short *)&y0i)[2]; + stream0_out[j + 16] = ((short *)&y1i)[2]; + stream0_out[j + 17] = ((short *)&y2i)[2]; + // RE 4 + stream0_out[j + 18] = ((short *)&y0r)[3]; + stream0_out[j + 19] = ((short *)&y1r)[3]; + stream0_out[j + 20] = ((short *)&y2r)[3]; + stream0_out[j + 21] = ((short *)&y0i)[3]; + stream0_out[j + 22] = ((short *)&y1i)[3]; + stream0_out[j + 23] = ((short *)&y2i)[3]; + // RE 5 + stream0_out[j + 24] = ((short *)&y0r)[4]; + stream0_out[j + 25] = ((short *)&y1r)[4]; + stream0_out[j + 26] = ((short *)&y2r)[4]; + stream0_out[j + 27] = ((short *)&y0i)[4]; + stream0_out[j + 28] = ((short *)&y1i)[4]; + stream0_out[j + 29] = ((short *)&y2i)[4]; + // RE 6 + stream0_out[j + 30] = ((short *)&y0r)[5]; + stream0_out[j + 31] = ((short *)&y1r)[5]; + stream0_out[j + 32] = ((short *)&y2r)[5]; + stream0_out[j + 33] = ((short *)&y0i)[5]; + stream0_out[j + 34] = ((short *)&y1i)[5]; + stream0_out[j + 35] = ((short *)&y2i)[5]; + // RE 7 + stream0_out[j + 36] = ((short *)&y0r)[6]; + stream0_out[j + 37] = ((short *)&y1r)[6]; + stream0_out[j + 38] = ((short *)&y2r)[6]; + stream0_out[j + 39] = ((short *)&y0i)[6]; + stream0_out[j + 40] = ((short *)&y1i)[6]; + stream0_out[j + 41] = ((short *)&y2i)[6]; + // RE 8 + stream0_out[j + 42] = ((short *)&y0r)[7]; + stream0_out[j + 43] = ((short *)&y1r)[7]; + stream0_out[j + 44] = ((short *)&y2r)[7]; + stream0_out[j + 45] = ((short *)&y0i)[7]; + stream0_out[j + 46] = ((short *)&y1i)[7]; + stream0_out[j + 47] = ((short *)&y2i)[7]; + +#elif defined(__arm__) + +#endif + + } + +#if defined(__x86_64__) || defined(__i386__) + _mm_empty(); + _m_empty(); +#endif +} + + +int nr_dlsch_64qam_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, + int32_t **rxdataF_comp, + int32_t **rxdataF_comp_i, + int32_t **dl_ch_mag, + int32_t **dl_ch_mag_i, + int32_t **rho_i, + int16_t *dlsch_llr, + uint8_t symbol, + uint8_t first_symbol_flag, + uint16_t nb_rb, + uint16_t pbch_pss_sss_adjust, + //int16_t **llr16p, + uint32_t llr_offset) +{ + + int16_t *rxF = (int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rxF_i = (int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag = (int16_t*)&dl_ch_mag[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *ch_mag_i = (int16_t*)&dl_ch_mag_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *rho = (int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)]; + int16_t *llr16; + int8_t *pllr_symbol; // pointer where llrs should filled for this ofdm symbol + int len; + uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol; + + //first symbol has different structure due to more pilots + /*if (first_symbol_flag == 1) { + llr16 = (int16_t*)dlsch_llr; + } else { + llr16 = (int16_t*)(*llr16p); + }*/ + + llr16 = (int16_t*)dlsch_llr; + + AssertFatal(llr16!=NULL,"nr_dlsch_16qam_64qam_llr:llr is null, symbol %d\n",symbol); + + + if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) { + // if symbol has pilots + if (frame_parms->nb_antenna_ports_eNB!=1) + // in 2 antenna ports we have 8 REs per symbol per RB + len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3); + else + // for 1 antenna port we have 10 REs per symbol per RB + len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6); + } else { + // symbol has no pilots + len = (nb_rb*12) - pbch_pss_sss_adjust; + } + + pllr_symbol = (int8_t*)dlsch_llr; + pllr_symbol += llr_offset; + //printf("nr_dlsch_64qam_64qam_llr: symbol %d,nb_rb %d, len %d,pbch_pss_sss_adjust %d\n",symbol,nb_rb,len,pbch_pss_sss_adjust); + /*LOG_I(PHY,"nr_dlsch_64qam_64qam_llr [symb %d / FirstSym %d / Length %d / LLR Offset %d]: @LLR Buff %x, @LLR Buff(symb) %x, , @Compute LLR Buff(symb) %x \n", + symbol, + first_symbol_flag, + len, + llr_offset, + (int16_t*)dlsch_llr, + llr16, + pllr_symbol);*/ + +#ifdef __AVX2__ + + // Round length up to multiple of 16 words + uint32_t len256i = ((len+16)>>4)*16; + int32_t *rxF_256i = (int32_t*) malloc16_clear(len256i*4); + int32_t *rxF_i_256i = (int32_t*) malloc16_clear(len256i*4); + int32_t *ch_mag_256i = (int32_t*) malloc16_clear(len256i*4); + int32_t *ch_mag_i_256i = (int32_t*) malloc16_clear(len256i*4); + int32_t *rho_256i = (int32_t*) malloc16_clear(len256i*4); + + memcpy(rxF_256i, rxF, len*4); + memcpy(rxF_i_256i, rxF_i, len*4); + memcpy(ch_mag_256i, ch_mag, len*4); + memcpy(ch_mag_i_256i, ch_mag_i, len*4); + memcpy(rho_256i, rho, len*4); + +#if 0 + qam64_qam16_avx2((short *)rxF_256i, + (short *)rxF_i_256i, + (short *)ch_mag_256i, + (short *)ch_mag_i_256i, + (short *)llr16, + (short *) rho_256i, + len); +#else + qam64_qam64_avx2((int32_t *)rxF_256i, + (int32_t *)rxF_i_256i, + (int32_t *)ch_mag_256i, + (int32_t *)ch_mag_i_256i, + (int16_t *)llr16, + (int32_t *) rho_256i, + len); +#endif + + free16(rxF_256i, sizeof(rxF_256i)); + free16(rxF_i_256i, sizeof(rxF_i_256i)); + free16(ch_mag_256i, sizeof(ch_mag_256i)); + free16(ch_mag_i_256i, sizeof(ch_mag_i_256i)); + free16(rho_256i, sizeof(rho_256i)); + +#else + qam64_qam64((short *)rxF, + (short *)rxF_i, + (short *)ch_mag, + (short *)ch_mag_i, + (short *)llr16, + (short *)rho, + len); +#endif + + llr16 += (6*len); + //*llr16p = (short *)llr16; + + return(0); +} diff --git a/openair1/PHY/NR_UE_TRANSPORT/nr_initial_sync.c b/openair1/PHY/NR_UE_TRANSPORT/nr_initial_sync.c index 80c71bd665..50a283f465 100644 --- a/openair1/PHY/NR_UE_TRANSPORT/nr_initial_sync.c +++ b/openair1/PHY/NR_UE_TRANSPORT/nr_initial_sync.c @@ -175,7 +175,6 @@ int nr_initial_sync(PHY_VARS_NR_UE *ue, runmode_t mode) * sync_pos SS/PBCH block */ - cnt++; if (1){ // (cnt>100) cnt =0; @@ -183,17 +182,20 @@ int nr_initial_sync(PHY_VARS_NR_UE *ue, runmode_t mode) /* process pss search on received buffer */ sync_pos = pss_synchro_nr(ue, NO_RATE_CHANGE); + sync_pos_slot = (fp->samples_per_subframe/fp->slots_per_subframe) - 10*(fp->ofdm_symbol_size + fp->nb_prefix_samples); - if (sync_pos >= fp->nb_prefix_samples) - ue->ssb_offset = sync_pos - fp->nb_prefix_samples; - else - ue->ssb_offset = sync_pos + (fp->samples_per_subframe * 10) - fp->nb_prefix_samples; + if (sync_pos >= fp->nb_prefix_samples){ + ue->ssb_offset = sync_pos - fp->nb_prefix_samples;} + else{ + ue->ssb_offset = sync_pos + (fp->samples_per_subframe * 10) - fp->nb_prefix_samples;} + + ue->rx_offset = ue->ssb_offset - sync_pos_slot; //write_output("rxdata1.m","rxd1",ue->common_vars.rxdata[0],10*fp->samples_per_subframe,1,1); #ifdef DEBUG_INITIAL_SYNCH LOG_I(PHY,"[UE%d] Initial sync : Estimated PSS position %d, Nid2 %d\n", ue->Mod_id, sync_pos,ue->common_vars.eNb_id); - LOG_I(PHY,"sync_pos %d ssb_offset %d\n",sync_pos,ue->ssb_offset); + LOG_I(PHY,"sync_pos %d ssb_offset %d sync_pos_slot %d \n",sync_pos,ue->ssb_offset,sync_pos_slot); #endif /* check that SSS/PBCH block is continuous inside the received buffer */ diff --git a/openair1/PHY/NR_UE_TRANSPORT/nr_pbch.c b/openair1/PHY/NR_UE_TRANSPORT/nr_pbch.c index 3e255ab03a..3e959fa42f 100644 --- a/openair1/PHY/NR_UE_TRANSPORT/nr_pbch.c +++ b/openair1/PHY/NR_UE_TRANSPORT/nr_pbch.c @@ -283,6 +283,7 @@ void nr_pbch_channel_compensation(int **rxdataF_ext, uint16_t nb_re=180; uint8_t aarx; + #if defined(__x86_64__) || defined(__i386__) __m128i *dl_ch128,*rxdataF128,*rxdataF_comp128; #elif defined(__arm__) @@ -569,7 +570,6 @@ int nr_rx_pbch( PHY_VARS_NR_UE *ue, for (symbol=first_symbol; symbol<(first_symbol+3); symbol++) { - nr_pbch_extract(nr_ue_common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[subframe_rx]].rxdataF, nr_ue_common_vars->common_vars_rx_data_per_thread[ue->current_thread_id[subframe_rx]].dl_ch_estimates[eNB_id], nr_ue_pbch_vars->rxdataF_ext, @@ -681,6 +681,7 @@ int nr_rx_pbch( PHY_VARS_NR_UE *ue, M = (Lmax == 64)? (NR_POLAR_PBCH_PAYLOAD_BITS - 6) : (NR_POLAR_PBCH_PAYLOAD_BITS - 3); nushift = ((nr_ue_pbch_vars->pbch_a_prime>>24)&1) ^ (((nr_ue_pbch_vars->pbch_a_prime>>6)&1)<<1); nr_pbch_unscrambling(nr_ue_pbch_vars,frame_parms->Nid_cell,nushift,M,NR_POLAR_PBCH_PAYLOAD_BITS,1,unscrambling_mask); + //printf("nushift %d sfn 3rd %d 2nd %d", nushift,((nr_ue_pbch_vars->pbch_a_prime>>6)&1), ((nr_ue_pbch_vars->pbch_a_prime>>24)&1) ); //payload deinterleaving //uint32_t in=0; diff --git a/openair1/PHY/NR_UE_TRANSPORT/nr_transport_proto_ue.h b/openair1/PHY/NR_UE_TRANSPORT/nr_transport_proto_ue.h index e656686310..3c272b9c17 100644 --- a/openair1/PHY/NR_UE_TRANSPORT/nr_transport_proto_ue.h +++ b/openair1/PHY/NR_UE_TRANSPORT/nr_transport_proto_ue.h @@ -32,6 +32,7 @@ #ifndef __NR_TRANSPORT_PROTO_UE__H__ #define __NR_TRANSPORT_PROTO_UE__H__ #include "PHY/defs_nr_UE.h" +#include "SCHED_NR_UE/defs.h" //#include "PHY/LTE_TRANSPORT/transport_common_proto.h" #include #include "nfapi_interface.h" @@ -46,7 +47,7 @@ \brief This function frees memory allocated for a particular DLSCH at UE @param dlsch Pointer to DLSCH to be removed */ -void free_ue_dlsch(NR_UE_DLSCH_t *dlsch); +void free_nr_ue_dlsch(NR_UE_DLSCH_t *dlsch); /** \fn new_ue_dlsch(uint8_t Kmimo,uint8_t Mdlharq,uint32_t Nsoft,uint8_t abstraction_flag) \brief This function allocates structures for a particular DLSCH at UE @@ -57,7 +58,7 @@ void free_ue_dlsch(NR_UE_DLSCH_t *dlsch); @params N_RB_DL total number of resource blocks (determine the operating BW) @param abstraction_flag Flag to indicate abstracted interface */ -NR_UE_DLSCH_t *new_ue_dlsch(uint8_t Kmimo,uint8_t Mdlharq,uint32_t Nsoft,uint8_t max_turbo_iterations,uint8_t N_RB_DL, uint8_t abstraction_flag); +NR_UE_DLSCH_t *new_nr_ue_dlsch(uint8_t Kmimo,uint8_t Mdlharq,uint32_t Nsoft,uint8_t max_turbo_iterations,uint8_t N_RB_DL, uint8_t abstraction_flag); void free_ue_ulsch(NR_UE_ULSCH_t *ulsch); @@ -106,7 +107,7 @@ void qpsk_qpsk(int16_t *stream0_in, @param nb_rb number of RBs for this allocation @param pbch_pss_sss_adj Number of channel bits taken by PBCH/PSS/SSS @param llr128p pointer to pointer to symbol in dlsch_llr*/ -int32_t dlsch_qpsk_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, +int32_t nr_dlsch_qpsk_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, int32_t **rxdataF_comp, int32_t **rxdataF_comp_i, int32_t **rho_i, @@ -142,7 +143,7 @@ void qpsk_qam16(int16_t *stream0_in, @param nb_rb number of RBs for this allocation @param pbch_pss_sss_adj Number of channel bits taken by PBCH/PSS/SSS @param llr128p pointer to pointer to symbol in dlsch_llr*/ -int32_t dlsch_qpsk_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, +int32_t nr_dlsch_qpsk_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, int32_t **rxdataF_comp, int32_t **rxdataF_comp_i, int **dl_ch_mag_i, //|h_1|^2*(2/sqrt{10}) @@ -154,6 +155,7 @@ int32_t dlsch_qpsk_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, uint16_t pbch_pss_sss_adj, int16_t **llr128p); + /** \brief This function computes the LLRs for ML (max-logsum approximation) dual-stream QPSK/64QAM reception. @param stream0_in Input from channel compensated (MR combined) stream 0 @param stream1_in Input from channel compensated (MR combined) stream 1 @@ -179,7 +181,7 @@ void qpsk_qam64(int16_t *stream0_in, @param nb_rb number of RBs for this allocation @param pbch_pss_sss_adj Number of channel bits taken by PBCH/PSS/SSS @param llr128p pointer to pointer to symbol in dlsch_llr*/ -int32_t dlsch_qpsk_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, +int32_t nr_dlsch_qpsk_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, int32_t **rxdataF_comp, int32_t **rxdataF_comp_i, int **dl_ch_mag_i, //|h_1|^2*(2/sqrt{10}) @@ -217,7 +219,7 @@ void qam16_qpsk(short *stream0_in, @param nb_rb number of RBs for this allocation @param pbch_pss_sss_adj Number of channel bits taken by PBCH/PSS/SSS @param llr16p pointer to pointer to symbol in dlsch_llr*/ -int dlsch_16qam_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, +int nr_dlsch_16qam_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, int **rxdataF_comp, int **rxdataF_comp_i, int **dl_ch_mag, //|h_0|^2*(2/sqrt{10}) @@ -258,7 +260,7 @@ void qam16_qam16(short *stream0_in, @param nb_rb number of RBs for this allocation @param pbch_pss_sss_adj Number of channel bits taken by PBCH/PSS/SSS @param llr16p pointer to pointer to symbol in dlsch_llr*/ -int dlsch_16qam_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, +int nr_dlsch_16qam_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, int **rxdataF_comp, int **rxdataF_comp_i, int **dl_ch_mag, //|h_0|^2*(2/sqrt{10}) @@ -300,7 +302,7 @@ void qam16_qam64(short *stream0_in, @param nb_rb number of RBs for this allocation @param pbch_pss_sss_adj Number of channel bits taken by PBCH/PSS/SSS @param llr16p pointer to pointer to symbol in dlsch_llr*/ -int dlsch_16qam_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, +int nr_dlsch_16qam_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, int **rxdataF_comp, int **rxdataF_comp_i, int **dl_ch_mag, //|h_0|^2*(2/sqrt{10}) @@ -339,7 +341,7 @@ void qam64_qpsk(short *stream0_in, @param nb_rb number of RBs for this allocation @param pbch_pss_sss_adj Number of channel bits taken by PBCH/PSS/SSS @param llr16p pointer to pointer to symbol in dlsch_llr*/ -int dlsch_64qam_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, +int nr_dlsch_64qam_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, int **rxdataF_comp, int **rxdataF_comp_i, int **dl_ch_mag, @@ -396,7 +398,7 @@ void qam64_qam16_avx2(short *stream0_in, @param nb_rb number of RBs for this allocation @param pbch_pss_sss_adj Number of channel bits taken by PBCH/PSS/SSS @param llr16p pointer to pointer to symbol in dlsch_llr*/ -int dlsch_64qam_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, +int nr_dlsch_64qam_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, int **rxdataF_comp, int **rxdataF_comp_i, int **dl_ch_mag, @@ -454,7 +456,7 @@ void qam64_qam64_avx2(int32_t *stream0_in, @param nb_rb number of RBs for this allocation @param pbch_pss_sss_adj Number of channel bits taken by PBCH/PSS/SSS @param llr16p pointer to pointer to symbol in dlsch_llr*/ -int dlsch_64qam_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, +int nr_dlsch_64qam_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, int **rxdataF_comp, int **rxdataF_comp_i, int **dl_ch_mag, @@ -480,15 +482,13 @@ int dlsch_64qam_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, @param llr128p pointer to pointer to symbol in dlsch_llr @param beamforming_mode beamforming mode */ -int32_t dlsch_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, - int32_t **rxdataF_comp, - int16_t *dlsch_llr, - uint8_t symbol, - uint8_t first_symbol_flag, - uint16_t nb_rb, - uint16_t pbch_pss_sss_adj, - //int16_t **llr128p, - uint8_t beamforming_mode); +int32_t nr_dlsch_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, + int32_t **rxdataF_comp, + int16_t *dlsch_llr, + uint8_t symbol, + uint8_t first_symbol_flag, + uint16_t nb_rb, + uint8_t beamforming_mode); /** \brief This function generates log-likelihood ratios (decoder input) for single-stream 16QAM received waveforms @@ -504,7 +504,7 @@ int32_t dlsch_qpsk_llr(NR_DL_FRAME_PARMS *frame_parms, @param beamforming_mode beamforming mode */ -int32_t dlsch_qpsk_llr_SIC(NR_DL_FRAME_PARMS *frame_parms, +int32_t nr_dlsch_qpsk_llr_SIC(NR_DL_FRAME_PARMS *frame_parms, int **rxdataF_comp, int32_t **sic_buffer, int **rho_i, @@ -515,15 +515,14 @@ int32_t dlsch_qpsk_llr_SIC(NR_DL_FRAME_PARMS *frame_parms, uint16_t mod_order_0, uint32_t rb_alloc); -void dlsch_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, +void nr_dlsch_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, int32_t **rxdataF_comp, int16_t *dlsch_llr, int32_t **dl_ch_mag, uint8_t symbol, uint8_t first_symbol_flag, uint16_t nb_rb, - uint16_t pbch_pss_sss_adjust, - int16_t **llr128p, + int16_t **llr32p, uint8_t beamforming_mode); /** \brief This function generates log-likelihood ratios (decoder input) for single-stream 16QAM received waveforms @@ -538,7 +537,7 @@ void dlsch_16qam_llr(NR_DL_FRAME_PARMS *frame_parms, @param pbch_pss_sss_adjust PBCH/PSS/SSS RE adjustment (in REs) @param beamforming_mode beamforming mode */ -void dlsch_16qam_llr_SIC (NR_DL_FRAME_PARMS *frame_parms, +void nr_dlsch_16qam_llr_SIC (NR_DL_FRAME_PARMS *frame_parms, int32_t **rxdataF_comp, int32_t **sic_buffer, //Q15 int32_t **rho_i, @@ -563,8 +562,7 @@ void dlsch_64qam_llr_SIC(NR_DL_FRAME_PARMS *frame_parms, uint16_t mod_order_0, uint32_t rb_alloc); - -void dlsch_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, +void nr_dlsch_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, int32_t **rxdataF_comp, int16_t *dlsch_llr, int32_t **dl_ch_mag, @@ -572,8 +570,6 @@ void dlsch_64qam_llr(NR_DL_FRAME_PARMS *frame_parms, uint8_t symbol, uint8_t first_symbol_flag, uint16_t nb_rb, - uint16_t pbch_pss_sss_adjust, - //int16_t **llr_save, uint32_t llr_offset, uint8_t beamforming_mode); @@ -708,7 +704,7 @@ void dlsch_detection_mrc_TM34(NR_DL_FRAME_PARMS *frame_parms, @param high_speed_flag @param frame_parms Pointer to frame descriptor */ -uint16_t dlsch_extract_rbs_single(int32_t **rxdataF, +/*uint16_t nr_dlsch_extract_rbs_single(int32_t **rxdataF, int32_t **dl_ch_estimates, int32_t **rxdataF_ext, int32_t **dl_ch_estimates_ext, @@ -718,7 +714,21 @@ uint16_t dlsch_extract_rbs_single(int32_t **rxdataF, uint8_t symbol, uint8_t subframe, uint32_t high_speed_flag, - NR_DL_FRAME_PARMS *frame_parms); + NR_DL_FRAME_PARMS *frame_parms);*/ + +unsigned short nr_dlsch_extract_rbs_single(int **rxdataF, + int **dl_ch_estimates, + int **rxdataF_ext, + int **dl_ch_estimates_ext, + unsigned short pmi, + unsigned char *pmi_ext, + unsigned int *rb_alloc, + unsigned char symbol, + unsigned short start_rb, + unsigned short nb_pdsch_rb, + unsigned char nr_tti_rx, + uint32_t high_speed_flag, + NR_DL_FRAME_PARMS *frame_parms); /** \fn dlsch_extract_rbs_dual(int32_t **rxdataF, int32_t **dl_ch_estimates, @@ -802,7 +812,7 @@ uint16_t dlsch_extract_rbs_TM7(int32_t **rxdataF, @param output_shift Rescaling for compensated output (should be energy-normalizing) @param phy_measurements Pointer to UE PHY measurements */ -void dlsch_channel_compensation(int32_t **rxdataF_ext, +void nr_dlsch_channel_compensation(int32_t **rxdataF_ext, int32_t **dl_ch_estimates_ext, int32_t **dl_ch_mag, int32_t **dl_ch_magb, @@ -816,6 +826,12 @@ void dlsch_channel_compensation(int32_t **rxdataF_ext, uint8_t output_shift, PHY_NR_MEASUREMENTS *phy_measurements); +void nr_dlsch_deinterleaving(uint8_t symbol, + uint16_t L, + uint16_t *llr, + uint16_t *llr_deint, + uint16_t nb_rb_pdsch); + void dlsch_dual_stream_correlation(NR_DL_FRAME_PARMS *frame_parms, unsigned char symbol, unsigned short nb_rb, @@ -923,7 +939,7 @@ void dlsch_channel_compensation_TM34(NR_DL_FRAME_PARMS *frame_parms, @param pilots_flag Flag to indicate pilots in symbol @param nb_rb Number of allocated RBs */ -void dlsch_channel_level(int32_t **dl_ch_estimates_ext, +void nr_dlsch_channel_level(int32_t **dl_ch_estimates_ext, NR_DL_FRAME_PARMS *frame_parms, int32_t *avg, uint8_t pilots_flag, @@ -953,7 +969,7 @@ void dlsch_channel_level_TM7(int32_t **dl_bf_ch_estimates_ext, uint8_t pilots_flag, uint16_t nb_rb); -void dlsch_scale_channel(int32_t **dl_ch_estimates_ext, +void nr_dlsch_scale_channel(int32_t **dl_ch_estimates_ext, NR_DL_FRAME_PARMS *frame_parms, NR_UE_DLSCH_t **dlsch_ue, uint8_t symbol_mod, @@ -977,17 +993,42 @@ void dlsch_scale_channel(int32_t **dl_ch_estimates_ext, @param llr8_flag If 1, indicate that the 8-bit turbo decoder should be used @returns 0 on success, 1 on unsuccessful decoding */ -uint32_t dlsch_decoding(PHY_VARS_NR_UE *phy_vars_ue, - int16_t *dlsch_llr, - NR_DL_FRAME_PARMS *lte_frame_parms, - NR_UE_DLSCH_t *dlsch, - NR_DL_UE_HARQ_t *harq_process, - uint32_t frame, - uint8_t subframe, - uint8_t harq_pid, - uint8_t is_crnti, - uint8_t llr8_flag); +uint32_t nr_dlsch_decoding(PHY_VARS_NR_UE *phy_vars_ue, + short *dlsch_llr, + NR_DL_FRAME_PARMS *frame_parms, + NR_UE_DLSCH_t *dlsch, + NR_DL_UE_HARQ_t *harq_process, + uint32_t frame, + uint16_t nb_symb_sch, + uint8_t nr_tti_rx, + uint8_t harq_pid, + uint8_t is_crnti, + uint8_t llr8_flag); + +uint32_t nr_dlsch_decoding_mthread(PHY_VARS_NR_UE *phy_vars_ue, + UE_nr_rxtx_proc_t *proc, + int eNB_id, + short *dlsch_llr, + NR_DL_FRAME_PARMS *frame_parms, + NR_UE_DLSCH_t *dlsch, + NR_DL_UE_HARQ_t *harq_process, + uint32_t frame, + uint16_t nb_symb_sch, + uint8_t nr_tti_rx, + uint8_t harq_pid, + uint8_t is_crnti, + uint8_t llr8_flag); + +void *nr_dlsch_decoding_2thread0(void *arg); + +void *nr_dlsch_decoding_2thread1(void *arg); + +void nr_dlsch_unscrambling(int16_t* llr, + uint32_t size, + uint8_t q, + uint32_t Nid, + uint32_t n_RNTI); uint32_t dlsch_decoding_emul(PHY_VARS_NR_UE *phy_vars_ue, uint8_t subframe, @@ -1011,7 +1052,7 @@ uint32_t dlsch_decoding_emul(PHY_VARS_NR_UE *phy_vars_ue, @param rx_type. rx_type=RX_IC_single_stream will enable interference cancellation of a second stream when decoding the first stream. In case of TM1, 2, 5, and this can cancel interference from a neighbouring cell given by eNB_id_i. In case of TM5, eNB_id_i should be set to n_connected_eNB to perform multi-user interference cancellation. In case of TM3, eNB_id_i should be set to eNB_id to perform co-channel interference cancellation; this option should be used together with an interference cancellation step [...]. In case of TM3, if rx_type=RX_IC_dual_stream, both streams will be decoded by applying the IC single stream receiver twice. @param i_mod Modulation order of the interfering stream */ -int32_t rx_pdsch(PHY_VARS_NR_UE *phy_vars_ue, +int32_t nr_rx_pdsch(PHY_VARS_NR_UE *phy_vars_ue, PDSCH_t type, uint8_t eNB_id, uint8_t eNB_id_i, @@ -1023,12 +1064,16 @@ int32_t rx_pdsch(PHY_VARS_NR_UE *phy_vars_ue, uint8_t i_mod, uint8_t harq_pid); -int32_t rx_pdcch(PHY_VARS_NR_UE *ue, - uint32_t frame, - uint8_t subframe, - uint8_t eNB_id, - MIMO_mode_t mimo_mode, - uint32_t high_speed_flag); +int32_t nr_rx_pdcch(PHY_VARS_NR_UE *ue, + uint32_t frame, + uint8_t nr_tti_rx, + uint8_t eNB_id, + MIMO_mode_t mimo_mode, + uint32_t high_speed_flag, + uint8_t is_secondary_ue, + int nb_coreset_active, + uint16_t symbol_mon, + NR_SEARCHSPACE_TYPE_t searchSpaceType); /*! \brief Extract PSS and SSS resource elements @param phy_vars_ue Pointer to UE variables @@ -1386,37 +1431,6 @@ uint32_t ulsch_encoding(uint8_t *a, uint8_t control_only_flag, uint8_t Nbundled); - - - -/* \brief This routine demodulates the PHICH and updates PUSCH/ULSCH parameters. - @param phy_vars_ue Pointer to UE variables - @param proc Pointer to RXN_TXNp4 proc - @param subframe Subframe of received PDCCH/PHICH - @param eNB_id Index of eNB -*/ - -void rx_phich(PHY_VARS_NR_UE *phy_vars_ue, - UE_nr_rxtx_proc_t *proc, - uint8_t subframe, - uint8_t eNB_id); - - -/** \brief This routine provides the relationship between a PHICH TXOp and its corresponding PUSCH subframe (Table 8.3.-1 from 36.213). - @param frame_parms Pointer to DL frame configuration parameters - @param subframe Subframe of received/transmitted PHICH - @returns subframe of PUSCH transmission -*/ -uint8_t phich_subframe2_pusch_subframe(NR_DL_FRAME_PARMS *frame_parms,uint8_t subframe); - -/** \brief This routine provides the relationship between a PHICH TXOp and its corresponding PUSCH frame (Table 8.3.-1 from 36.213). - @param frame_parms Pointer to DL frame configuration parameters - @param frame Frame of received/transmitted PHICH - @param subframe Subframe of received/transmitted PHICH - @returns frame of PUSCH transmission -*/ -int phich_frame2_pusch_frame(NR_DL_FRAME_PARMS *frame_parms, int frame, int subframe); - void print_CQI(void *o,UCI_format_t uci_format,uint8_t eNB_id,int N_RB_DL); void fill_CQI(NR_UE_ULSCH_t *ulsch,PHY_NR_MEASUREMENTS *meas,uint8_t eNB_id, uint8_t harq_pid,int N_RB_DL, rnti_t rnti, uint8_t trans_mode,double sinr_eff); @@ -1643,5 +1657,66 @@ void nr_pdcch_unscrambling(uint16_t crnti, NR_DL_FRAME_PARMS *frame_parms, uint8 uint32_t lte_gold_generic(uint32_t *x1, uint32_t *x2, uint8_t reset); +uint8_t nr_dci_decoding_procedure(int s, + int p, + PHY_VARS_NR_UE *ue, + NR_DCI_ALLOC_t *dci_alloc, + NR_SEARCHSPACE_TYPE_t searchSpacetype, + int16_t eNB_id, + uint8_t nr_tti_rx, + uint8_t dci_fields_sizes_cnt[MAX_NR_DCI_DECODED_SLOT][NBR_NR_DCI_FIELDS][NBR_NR_FORMATS], + uint16_t n_RB_ULBWP, + uint16_t n_RB_DLBWP, + crc_scrambled_t *crc_scrambled, + format_found_t *format_found, + uint16_t crc_scrambled_values[TOTAL_NBR_SCRAMBLED_VALUES]); + +int nr_generate_ue_ul_dlsch_params_from_dci(PHY_VARS_NR_UE *ue, + uint8_t eNB_id, + int frame, + uint8_t nr_tti_rx, + uint32_t dci_pdu[4], + uint16_t rnti, + uint8_t dci_length, + NR_DCI_format_t dci_format, + NR_UE_PDCCH *pdcch_vars, + NR_UE_PDSCH *pdsch_vars, + NR_UE_DLSCH_t **dlsch, + NR_UE_ULSCH_t *ulsch, + NR_DL_FRAME_PARMS *frame_parms, + PDSCH_CONFIG_DEDICATED *pdsch_config_dedicated, + uint8_t beamforming_mode, + uint8_t dci_fields_sizes[NBR_NR_DCI_FIELDS][NBR_NR_FORMATS], + uint16_t n_RB_ULBWP, + uint16_t n_RB_DLBWP, + uint16_t crc_scrambled_values[TOTAL_NBR_SCRAMBLED_VALUES], + NR_DCI_INFO_EXTRACTED_t *nr_dci_info_extracted); + +int nr_rx_pdsch(PHY_VARS_NR_UE *ue, + PDSCH_t type, + unsigned char eNB_id, + unsigned char eNB_id_i, //if this == ue->n_connected_eNB, we assume MU interference + uint32_t frame, + uint8_t nr_tti_rx, + unsigned char symbol, + unsigned char first_symbol_flag, + RX_type_t rx_type, + unsigned char i_mod, + unsigned char harq_pid); + +uint32_t nr_get_G(uint16_t nb_rb, uint16_t nb_symb_sch,uint8_t nb_re_dmrs,uint16_t length_dmrs, uint8_t Qm, uint8_t Nl) ; + +uint32_t nr_dlsch_decoding(PHY_VARS_NR_UE *phy_vars_ue, + short *dlsch_llr, + NR_DL_FRAME_PARMS *frame_parms, + NR_UE_DLSCH_t *dlsch, + NR_DL_UE_HARQ_t *harq_process, + uint32_t frame, + uint16_t nb_symb_sch, + uint8_t nr_tti_rx, + uint8_t harq_pid, + uint8_t is_crnti, + uint8_t llr8_flag); + /**@}*/ #endif diff --git a/openair1/PHY/NR_UE_TRANSPORT/nr_transport_ue.h b/openair1/PHY/NR_UE_TRANSPORT/nr_transport_ue.h index 28667df2ca..f89816c782 100644 --- a/openair1/PHY/NR_UE_TRANSPORT/nr_transport_ue.h +++ b/openair1/PHY/NR_UE_TRANSPORT/nr_transport_ue.h @@ -37,10 +37,7 @@ //#include "PHY/defs_nr_UE.h" -//#include "../LTE_TRANSPORT/dci.h" -//#include "../LTE_TRANSPORT/mdci.h" -//#include "../LTE_TRANSPORT/uci_common.h" -//#include "../LTE_TRANSPORT/transport_common.h" +#include "../NR_TRANSPORT/nr_transport_common_proto.h" #ifndef STANDALONE_COMPILE #include "UTIL/LISTS/list.h" #endif @@ -117,39 +114,31 @@ typedef struct { uint8_t *b; /// Pointers to transport block segments uint8_t *c[MAX_NUM_ULSCH_SEGMENTS]; - /// RTC values for each segment (for definition see 36-212 V8.6 2009-03, p.15) - uint32_t RTC[MAX_NUM_ULSCH_SEGMENTS]; /// Index of current HARQ round for this ULSCH uint8_t round; /// MCS format of this ULSCH uint8_t mcs; /// Redundancy-version of the current sub-frame uint8_t rvidx; - /// Turbo-code outputs (36-212 V8.6 2009-03, p.12 - uint8_t d[MAX_NUM_ULSCH_SEGMENTS][(96+3+(3*6144))]; - /// Sub-block interleaver outputs (36-212 V8.6 2009-03, p.16-17) - uint8_t w[MAX_NUM_ULSCH_SEGMENTS][3*6144]; - /// Number of code segments (for definition see 36-212 V8.6 2009-03, p.9) + /// LDPC-code outputs + uint8_t d[MAX_NUM_ULSCH_SEGMENTS][3*8448]; + /// Interleaver outputs + uint8_t w[MAX_NUM_ULSCH_SEGMENTS][3*8448]; + /// Number of code segments uint32_t C; - /// Number of "small" code segments (for definition see 36-212 V8.6 2009-03, p.10) - uint32_t Cminus; - /// Number of "large" code segments (for definition see 36-212 V8.6 2009-03, p.10) - uint32_t Cplus; - /// Number of bits in "small" code segments (<6144) (for definition see 36-212 V8.6 2009-03, p.10) - uint32_t Kminus; - /// Number of bits in "large" code segments (<6144) (for definition see 36-212 V8.6 2009-03, p.10) - uint32_t Kplus; + /// Number of bits in code segments + uint32_t K; /// Total number of bits across all segments uint32_t sumKr; - /// Number of "Filler" bits (for definition see 36-212 V8.6 2009-03, p.10) + /// Number of "Filler" bits uint32_t F; - /// Msc_initial, Initial number of subcarriers for ULSCH (36-212, v8.6 2009-03, p.26-27) + /// Msc_initial, Initial number of subcarriers for ULSCH uint16_t Msc_initial; - /// Nsymb_initial, Initial number of symbols for ULSCH (36-212, v8.6 2009-03, p.26-27) + /// Nsymb_initial, Initial number of symbols for ULSCH uint8_t Nsymb_initial; - /// n_DMRS for cyclic shift of DMRS (36.213 Table 9.1.2-2) + /// n_DMRS for cyclic shift of DMRS uint8_t n_DMRS; - /// n_DMRS2 for cyclic shift of DMRS (36.211 Table 5.5.1.1.-1) + /// n_DMRS2 for cyclic shift of DMRS uint8_t n_DMRS2; /// Flag to indicate that this is a control only ULSCH (i.e. no MAC SDU) uint8_t control_only; @@ -261,9 +250,7 @@ typedef struct { /// Pointer to the payload uint8_t *b; /// Pointers to transport block segments - uint8_t *c[MAX_NUM_DLSCH_SEGMENTS]; - /// RTC values for each segment (for definition see 36-212 V8.6 2009-03, p.15) - uint32_t RTC[MAX_NUM_DLSCH_SEGMENTS]; + uint8_t *c[MAX_NUM_NR_DLSCH_SEGMENTS]; /// Index of current HARQ round for this DLSCH uint8_t round; /// MCS format for this DLSCH @@ -275,24 +262,20 @@ typedef struct { /// MIMO mode for this DLSCH MIMO_mode_t mimo_mode; /// soft bits for each received segment ("w"-sequence)(for definition see 36-212 V8.6 2009-03, p.15) - int16_t w[MAX_NUM_DLSCH_SEGMENTS][3*(6144+64)]; + int16_t w[MAX_NUM_NR_DLSCH_SEGMENTS][3*8448]; /// for abstraction soft bits for each received segment ("w"-sequence)(for definition see 36-212 V8.6 2009-03, p.15) - double w_abs[MAX_NUM_DLSCH_SEGMENTS][3*(6144+64)]; + double w_abs[MAX_NUM_NR_DLSCH_SEGMENTS][3*8448]; /// soft bits for each received segment ("d"-sequence)(for definition see 36-212 V8.6 2009-03, p.15) - int16_t *d[MAX_NUM_DLSCH_SEGMENTS]; - /// Number of code segments (for definition see 36-212 V8.6 2009-03, p.9) + int16_t *d[MAX_NUM_NR_DLSCH_SEGMENTS]; + /// Number of code segments uint32_t C; - /// Number of "small" code segments (for definition see 36-212 V8.6 2009-03, p.10) - uint32_t Cminus; - /// Number of "large" code segments (for definition see 36-212 V8.6 2009-03, p.10) - uint32_t Cplus; - /// Number of bits in "small" code segments (<6144) (for definition see 36-212 V8.6 2009-03, p.10) - uint32_t Kminus; - /// Number of bits in "large" code segments (<6144) (for definition see 36-212 V8.6 2009-03, p.10) - uint32_t Kplus; - /// Number of "Filler" bits (for definition see 36-212 V8.6 2009-03, p.10) + /// Number of bits in code segments + uint32_t K; + /// Number of "Filler" bits uint32_t F; - /// Number of MIMO layers (streams) (for definition see 36-212 V8.6 2009-03, p.17) + /// LDPC lifting factor + uint32_t Z; + /// Number of MIMO layers (streams) uint8_t Nl; /// current delta_pucch int8_t delta_PUCCH; @@ -353,25 +336,22 @@ typedef struct { uint32_t cqi_alloc2; /// saved subband PMI allocation from last PUSCH/PUCCH report uint16_t pmi_alloc; -//#if defined(UPGRADE_RAT_NR) -#if 1 /// Pointers to up to HARQ processes - NR_DL_UE_HARQ_t harq_processes[NR_MAX_DLSCH_HARQ_PROCESSES]; + NR_DL_UE_HARQ_t *harq_processes[NR_MAX_DLSCH_HARQ_PROCESSES]; // DL number of harq processes uint8_t number_harq_processes_for_pdsch; /* higher layer parameter for reception of two transport blocks TS 38.213 9.1.3.1 Type-2 HARQ-ACK codebook dtermination */ uint8_t Number_MCS_HARQ_DL_DCI; /* spatial bundling of PUCCH */ uint8_t HARQ_ACK_spatial_bundling_PUCCH; -#endif /// Maximum number of HARQ processes(for definition see 36-212 V8.6 2009-03, p.17 uint8_t Mdlharq; /// MIMO transmission mode indicator for this sub-frame (for definition see 36-212 V8.6 2009-03, p.17) uint8_t Kmimo; /// Nsoft parameter related to UE Category uint32_t Nsoft; - /// Maximum number of Turbo iterations - uint8_t max_turbo_iterations; + /// Maximum number of LDPC iterations + uint8_t max_ldpc_iterations; /// number of iterations used in last turbo decoding uint8_t last_iteration_cnt; diff --git a/openair1/PHY/NR_UE_TRANSPORT/pss_nr.c b/openair1/PHY/NR_UE_TRANSPORT/pss_nr.c index 682d0233d8..5ac83b5280 100644 --- a/openair1/PHY/NR_UE_TRANSPORT/pss_nr.c +++ b/openair1/PHY/NR_UE_TRANSPORT/pss_nr.c @@ -383,8 +383,8 @@ void init_context_pss_nr(NR_DL_FRAME_PARMS *frame_parms_ue) bzero( primary_synchro_nr[i], sizePss); } else { - msg("Fatal memory allocation problem \n"); - assert(0); + LOG_E(PHY,"Fatal memory allocation problem \n"); + assert(0); } p = malloc(LENGTH_PSS_NR*2); if (p != NULL) { @@ -397,7 +397,7 @@ void init_context_pss_nr(NR_DL_FRAME_PARMS *frame_parms_ue) bzero( primary_synchro_time_nr[i], size); } else { - msg("Fatal memory allocation problem \n"); + LOG_E(PHY,"Fatal memory allocation problem \n"); assert(0); } @@ -408,7 +408,7 @@ void init_context_pss_nr(NR_DL_FRAME_PARMS *frame_parms_ue) bzero( pss_corr_ue[i], size); } else { - msg("Fatal memory allocation problem \n"); + LOG_E(PHY,"Fatal memory allocation problem \n"); assert(0); } @@ -437,7 +437,7 @@ void free_context_pss_nr(void) primary_synchro_time_nr[i] = NULL; } else { - msg("Fatal memory deallocation problem \n"); + LOG_E(PHY,"Fatal memory deallocation problem \n"); assert(0); } @@ -446,7 +446,7 @@ void free_context_pss_nr(void) primary_synchro_nr[i] = NULL; } else { - msg("Fatal memory deallocation problem \n"); + LOG_E(PHY,"Fatal memory deallocation problem \n"); assert(0); } @@ -455,7 +455,7 @@ void free_context_pss_nr(void) pss_corr_ue[i] = NULL; } else { - msg("Fatal memory deallocation problem \n"); + LOG_E(PHY,"Fatal memory deallocation problem \n"); assert(0); } } @@ -480,13 +480,13 @@ void init_context_synchro_nr(NR_DL_FRAME_PARMS *frame_parms_ue) /* initialise global buffers for synchronisation */ synchroF_tmp = malloc16(SYNCF_TMP_SIZE); if (synchroF_tmp == NULL) { - msg("Fatal memory allocation problem \n"); + LOG_E(PHY,"Fatal memory allocation problem \n"); assert(0); } synchro_tmp = malloc16(SYNC_TMP_SIZE); if (synchro_tmp == NULL) { - msg("Fatal memory allocation problem \n"); + LOG_E(PHY,"Fatal memory allocation problem \n"); assert(0); } @@ -518,7 +518,7 @@ void free_context_synchro_nr(void) synchroF_tmp = NULL; } else { - msg("Fatal memory deallocation problem \n"); + LOG_E(PHY,"Fatal memory deallocation problem \n"); assert(0); } @@ -527,7 +527,7 @@ void free_context_synchro_nr(void) synchro_tmp = NULL; } else { - msg("Fatal memory deallocation problem \n"); + LOG_E(PHY,"Fatal memory deallocation problem \n"); assert(0); } @@ -838,6 +838,7 @@ int pss_search_time_nr(int **rxdata, ///rx data in time domain /* Search pss in the received buffer each 4 samples which ensures a memory alignment on 128 bits (32 bits x 4 ) */ /* This is required by SIMD (single instruction Multiple Data) Extensions of Intel processors. */ /* Correlation computation is based on a a dot product which is realized thank to SIMS extensions */ + for (int pss_index = 0; pss_index < NUMBER_PSS_SEQUENCE; pss_index++) { avg[pss_index]=0; memset(pss_corr_ue[pss_index],0,length*sizeof(int64_t)); diff --git a/openair1/PHY/TOOLS/nr_phy_scope.c b/openair1/PHY/TOOLS/nr_phy_scope.c index 4da49702d8..0100dd44c4 100644 --- a/openair1/PHY/TOOLS/nr_phy_scope.c +++ b/openair1/PHY/TOOLS/nr_phy_scope.c @@ -419,7 +419,7 @@ FD_lte_phy_scope_ue *create_lte_phy_scope_ue( void ) fl_set_object_lcolor( fdui->pbch_llr, FL_WHITE ); // Label color fl_set_xyplot_symbolsize( fdui->pbch_llr,2); fl_set_xyplot_xgrid( fdui->pbch_llr,FL_GRID_MAJOR); - fl_set_xyplot_xbounds( fdui->pbch_llr,0,1920); + //fl_set_xyplot_xbounds( fdui->pbch_llr,0,1920); // I/Q PBCH comp fdui->pbch_comp = fl_add_xyplot( FL_POINTS_XYPLOT, 540, 260, 240, 100, "PBCH I/Q of MF Output" ); @@ -501,21 +501,29 @@ void phy_scope_UE(FD_lte_phy_scope_ue *form, /* int16_t **chest_t; int16_t **chest_f; + */ int16_t *pdsch_llr; int16_t *pdsch_comp; - int16_t *pdsch_mag; + //int16_t *pdsch_mag; int8_t *pdcch_llr; int16_t *pdcch_comp; - int8_t *pbch_llr; + int16_t *pbch_llr; int16_t *pbch_comp; - float Re,Im,ymax=1; - int num_pdcch_symbols=3; - float *llr, *bit, *chest_f_abs, llr_pbch[1920], bit_pbch[1920], *llr_pdcch, *bit_pdcch; + float llr_pbch[1920], bit_pbch[1920]; + float *llr, *bit; + float *llr_pdcch, *bit_pdcch; float *I, *Q; - float **chest_t_abs; + int num_pdcch_symbols=2; + int num_re = 4500; + int Qm = 2; + int coded_bits_per_codeword = num_re*Qm; + int symbol, first_symbol,nb_re; + int nb_rb_pdsch =50; + /* + float Re,Im,ymax=1; + float **chest_t_abs, *chest_f_abs; float freq[nsymb_ce*nb_antennas_rx*nb_antennas_tx]; int frame = phy_vars_ue->proc.proc_rxtx[0].frame_rx; - int coded_bits_per_codeword = 0; int mcs = 0; unsigned char harq_pid = 0; */ @@ -554,9 +562,10 @@ void phy_scope_UE(FD_lte_phy_scope_ue *form, } else { coded_bits_per_codeword = 0; //frame_parms->N_RB_DL*12*get_Qm(mcs)*(frame_parms->symbols_per_tti); } - - I = (float*) calloc(frame_parms->ofdm_symbol_size*frame_parms->symbols_per_tti*2,sizeof(float)); - Q = (float*) calloc(frame_parms->ofdm_symbol_size*frame_parms->symbols_per_tti*2,sizeof(float)); + */ + I = (float*) calloc(frame_parms->ofdm_symbol_size*frame_parms->symbols_per_slot*2,sizeof(float)); + Q = (float*) calloc(frame_parms->ofdm_symbol_size*frame_parms->symbols_per_slot*2,sizeof(float)); + /* chest_t_abs = (float**) malloc(nb_antennas_rx*sizeof(float*)); for (arx=0; arxN_RB_DL*num_pdcch_symbols*2,sizeof(float)); // init to zero bit_pdcch = (float*) calloc(12*frame_parms->N_RB_DL*num_pdcch_symbols*2,sizeof(float)); - */ rxsig_t = (int16_t**) phy_vars_ue->common_vars.rxdata; rxsig_t_dB = calloc(nb_antennas_rx,sizeof(float*)); @@ -577,19 +588,20 @@ void phy_scope_UE(FD_lte_phy_scope_ue *form, } time = calloc(samples_per_frame,sizeof(float)); corr = calloc(samples_per_frame,sizeof(float)); - + /* chest_t = (int16_t**) phy_vars_ue->common_vars.common_vars_rx_data_per_thread[phy_vars_ue->current_thread_id[subframe]].dl_ch_estimates_time[eNB_id]; chest_f = (int16_t**) phy_vars_ue->common_vars.common_vars_rx_data_per_thread[phy_vars_ue->current_thread_id[subframe]].dl_ch_estimates[eNB_id]; - pbch_llr = (int8_t*) phy_vars_ue->pbch_vars[eNB_id]->llr; + */ + pbch_llr = (int16_t*) phy_vars_ue->pbch_vars[eNB_id]->llr; pbch_comp = (int16_t*) phy_vars_ue->pbch_vars[eNB_id]->rxdataF_comp[0]; + pdcch_llr = (int8_t*) phy_vars_ue->pdcch_vars[phy_vars_ue->current_thread_id[subframe]][eNB_id]->llr; pdcch_comp = (int16_t*) phy_vars_ue->pdcch_vars[phy_vars_ue->current_thread_id[subframe]][eNB_id]->rxdataF_comp[0]; - pdsch_llr = (int16_t*) phy_vars_ue->pdsch_vars[phy_vars_ue->current_thread_id[subframe]][eNB_id]->llr[0]; // stream 0 + pdsch_llr = (int16_t*) phy_vars_ue->pdsch_vars[phy_vars_ue->current_thread_id[subframe]][eNB_id]->llr[0]; // stream 0 // pdsch_llr = (int16_t*) phy_vars_ue->lte_ue_pdsch_vars_SI[eNB_id]->llr[0]; // stream 0 pdsch_comp = (int16_t*) phy_vars_ue->pdsch_vars[phy_vars_ue->current_thread_id[subframe]][eNB_id]->rxdataF_comp0[0]; - pdsch_mag = (int16_t*) phy_vars_ue->pdsch_vars[phy_vars_ue->current_thread_id[subframe]][eNB_id]->dl_ch_mag0[0]; - */ + //pdsch_mag = (int16_t*) phy_vars_ue->pdsch_vars[phy_vars_ue->current_thread_id[subframe]][eNB_id]->dl_ch_mag0[0]; // Received signal in time domain of receive antenna 0 if (rxsig_t != NULL) { @@ -615,6 +627,7 @@ void phy_scope_UE(FD_lte_phy_scope_ue *form, */ } + if (phy_vars_ue->is_synchronized==0) { for (ind=0;ind<3;ind++) { if (pss_corr_ue[ind]) { for (i=0; ichest_t,ind,time,corr,samples_per_frame,rx_antenna_colors[ind]); } } - + } /* // Channel Impulse Response (still repeated format) @@ -710,59 +723,55 @@ void phy_scope_UE(FD_lte_phy_scope_ue *form, } } } - + */ + // PBCH LLRs if (pbch_llr != NULL) { - for (i=0; i<1920; i++) { + for (i=0; i<864; i++) { llr_pbch[i] = (float) pbch_llr[i]; bit_pbch[i] = (float) i; } - fl_set_xyplot_data(form->pbch_llr,bit_pbch,llr_pbch,1920,"","",""); + fl_set_xyplot_data(form->pbch_llr,bit_pbch,llr_pbch,864,"","",""); } + if (phy_vars_ue->is_synchronized==1) + first_symbol=5; + else + first_symbol=1; + // PBCH I/Q of MF Output if (pbch_comp!=NULL) { - for (i=0; i<72*2; i++) { - I[i] = pbch_comp[2*i]; - Q[i] = pbch_comp[2*i+1]; + for (symbol=first_symbol; symbol<(first_symbol+3); symbol++) { + if (symbol == 2 || symbol == 6) + nb_re = 72; + else + nb_re = 180; + for (i=0; ipbch_comp,I,Q,72*2,"","",""); + fl_set_xyplot_data(form->pbch_comp,I,Q,432,"","",""); } // PDCCH LLRs if (pdcch_llr != NULL) { - for (i=0; i<12*frame_parms->N_RB_DL*2*num_pdcch_symbols; i++) { - llr_pdcch[i] = (float) pdcch_llr[i]; + for (i=0; i<100; i++) { //12*frame_parms->N_RB_DL*2*num_pdcch_symbols + llr_pdcch[i] = (float) pdcch_llr[2*24*9 +i]; bit_pdcch[i] = (float) i; } - fl_set_xyplot_xbounds(form->pdcch_llr,0,12*frame_parms->N_RB_DL*2*3); - if (frame_parms->N_RB_DL != 100) - { - fl_set_xyplot_data(form->pdcch_llr,bit_pdcch,llr_pdcch,12*frame_parms->N_RB_DL*2*num_pdcch_symbols,"","",""); - } - else - { - LOG_D(PHY,"UE PDCCH LLR plot is bugged in 20 MHz BW, to be fixed !!!\n"); - } + fl_set_xyplot_data(form->pdcch_llr,bit_pdcch,llr_pdcch,12*frame_parms->N_RB_DL*num_pdcch_symbols,"","",""); } // PDCCH I/Q of MF Output if (pdcch_comp!=NULL) { - for (i=0; i<12*frame_parms->N_RB_DL*num_pdcch_symbols; i++) { - I[i] = pdcch_comp[2*i]; - Q[i] = pdcch_comp[2*i+1]; - } - if (frame_parms->N_RB_DL != 100) - { - fl_set_xyplot_data(form->pdcch_comp,I,Q,12*frame_parms->N_RB_DL*num_pdcch_symbols,"","",""); - } - else - { - LOG_D(PHY,"UE PDCCH COMP plot is bugged in 20 MHz BW, to be fixed !!!\n"); + for (i=0; i<100; i++) { + I[i] = pdcch_comp[2*50*12+2*i]; + Q[i] = pdcch_comp[2*50*12+2*i+1]; } + fl_set_xyplot_data(form->pdcch_comp,I,Q,12*frame_parms->N_RB_DL*num_pdcch_symbols,"","",""); } // PDSCH LLRs @@ -772,32 +781,20 @@ void phy_scope_UE(FD_lte_phy_scope_ue *form, bit[i] = (float) i; } - fl_set_xyplot_xbounds(form->pdsch_llr,0,coded_bits_per_codeword); - if (frame_parms->N_RB_DL != 100) - { - fl_set_xyplot_data(form->pdsch_llr,bit,llr,coded_bits_per_codeword,"","",""); - } - else - { - LOG_D(PHY,"UE PDSCH LLR plot is bugged in 20 MHz BW, to be fixed !!!\n"); - } + //fl_set_xyplot_xbounds(form->pdsch_llr,0,coded_bits_per_codeword); + fl_set_xyplot_data(form->pdsch_llr,bit,llr,coded_bits_per_codeword,"","",""); } // PDSCH I/Q of MF Output if (pdsch_comp!=NULL) { - ind=0; - - for (k=0; ksymbols_per_tti; k++) { - for (i=0; i<12*frame_parms->N_RB_DL/2; i++) { - int j = (2*frame_parms->N_RB_DL*12*k)+4*i; - I[ind] = (pdsch_mag[j ]!=0? 1.0/pdsch_mag[j ]: 0.0) * pdsch_comp[j ]*1.0; - Q[ind] = (pdsch_mag[j+1]!=0? 1.0/pdsch_mag[j+1]: 0.0) * pdsch_comp[j+1]*1.0; - ind++; - } + for (i=0; ipdsch_comp,I,Q,ind,"","",""); + + fl_set_xyplot_data(form->pdsch_comp,I,Q,nb_rb_pdsch*12,"","",""); } + /* // PDSCH Throughput memmove( tput_time_ue[UE_id], &tput_time_ue[UE_id][1], (TPUT_WINDOW_LENGTH-1)*sizeof(float) ); @@ -824,19 +821,18 @@ void phy_scope_UE(FD_lte_phy_scope_ue *form, } free(rxsig_t_dB); - /* free(I); free(Q); - free(chest_f_abs); free(llr); free(bit); free(bit_pdcch); free(llr_pdcch); + /* + free(chest_f_abs); for (arx=0; arx +#include "TOOLS/tools_defs.h" + #include "targets/COMMON/openairinterface5g_limits.h" +#include "common/utils/LOG/log.h" #include "types.h" #include "nfapi_interface.h" @@ -102,6 +105,7 @@ #ifdef OCP_FRAMEWORK #include "enums.h" #else + typedef enum {TDD=1,FDD=0} lte_frame_type_t; typedef enum {EXTENDED=1,NORMAL=0} lte_prefix_type_t; diff --git a/openair1/PHY/defs_eNB.h b/openair1/PHY/defs_eNB.h index ab88fd133c..d23586c634 100644 --- a/openair1/PHY/defs_eNB.h +++ b/openair1/PHY/defs_eNB.h @@ -710,4 +710,5 @@ typedef struct PHY_VARS_eNB_s { int32_t pusch_stats_bsr[NUMBER_OF_UE_MAX][10240]; int32_t pusch_stats_BO[NUMBER_OF_UE_MAX][10240]; } PHY_VARS_eNB; + #endif // __PHY_DEFS_eNB_H__ diff --git a/openair1/PHY/defs_gNB.h b/openair1/PHY/defs_gNB.h index b5be517d9b..24a9fd45c8 100644 --- a/openair1/PHY/defs_gNB.h +++ b/openair1/PHY/defs_gNB.h @@ -37,6 +37,7 @@ #include "defs_nr_common.h" #include "CODING/nrPolar_tools/nr_polar_pbch_defs.h" #include "openair2/NR_PHY_INTERFACE/NR_IF_Module.h" +#include "PHY/NR_TRANSPORT/nr_transport_common_proto.h" #define MAX_NUM_RU_PER_gNB MAX_NUM_RU_PER_eNB @@ -61,6 +62,8 @@ typedef struct { uint16_t size; /// Aggregation level uint8_t L; + /// HARQ PID + uint8_t harq_pid; /// PDCCH parameters nfapi_nr_dl_config_pdcch_parameters_rel15_t pdcch_params; /// CCE list @@ -71,9 +74,92 @@ typedef struct { typedef struct { uint8_t num_dci; + uint8_t num_pdsch_rnti; NR_gNB_DCI_ALLOC_t dci_alloc[256]; } NR_gNB_PDCCH; + +typedef struct { + /// Nfapi DLSCH PDU + nfapi_nr_dl_config_dlsch_pdu dlsch_pdu; + /// pointer to pdu from MAC interface (this is "a" in 36.212) + uint8_t *pdu; + /// The payload + CRC size in bits, "B" from 36-212 + uint32_t B; + /// Pointer to the payload + uint8_t *b; + /// Pointers to transport block segments + uint8_t *c[MAX_NUM_NR_DLSCH_SEGMENTS]; + /// Frame where current HARQ round was sent + uint32_t frame; + /// Subframe where current HARQ round was sent + uint32_t subframe; + /// Index of current HARQ round for this DLSCH + uint8_t round; + /// MIMO mode for this DLSCH + MIMO_mode_t mimo_mode; + /// Concatenated sequences + uint8_t e[MAX_NUM_NR_CHANNEL_BITS] __attribute__((aligned(32))); + /// LDPC-code outputs + uint8_t *d[MAX_NUM_NR_DLSCH_SEGMENTS]; + /// Interleaver outputs + uint8_t f[MAX_NUM_NR_CHANNEL_BITS] __attribute__((aligned(32))); + /// Number of code segments + uint32_t C; + /// Number of bits in "small" code segments + uint32_t K; + /// Number of "Filler" bits + uint32_t F; +} NR_DL_gNB_HARQ_t; + + +typedef struct { + + /// Pointers to 16 HARQ processes for the DLSCH + NR_DL_gNB_HARQ_t *harq_processes[NR_MAX_NB_HARQ_PROCESSES]; + /// TX buffers for UE-spec transmission (antenna ports 5 or 7..14, prior to precoding) + int32_t *txdataF[NR_MAX_NB_LAYERS]; + /// Modulated symbols buffer + int32_t *mod_symbs[NR_MAX_NB_CODEWORDS]; + /// beamforming weights for UE-spec transmission (antenna ports 5 or 7..14), for each codeword, maximum 4 layers? + int32_t **ue_spec_bf_weights[NR_MAX_NB_LAYERS]; + /// dl channel estimates (estimated from ul channel estimates) + int32_t **calib_dl_ch_estimates; + /// Allocated RNTI (0 means DLSCH_t is not currently used) + uint16_t rnti; + /// Active flag for baseband transmitter processing + uint8_t active; + /// HARQ process mask, indicates which processes are currently active + uint16_t harq_mask; + /// Indicator of TX activation per subframe. Used during PUCCH detection for ACK/NAK. + uint8_t subframe_tx[10]; + /// First CCE of last PDSCH scheduling per subframe. Again used during PUCCH detection for ACK/NAK. + uint8_t nCCE[10]; + /// Process ID's per subframe. Used to associate received ACKs on PUSCH/PUCCH to DLSCH harq process ids + uint8_t harq_ids[10]; + /// Window size (in outgoing transport blocks) for fine-grain rate adaptation + uint8_t ra_window_size; + /// First-round error threshold for fine-grain rate adaptation + uint8_t error_threshold; + /// Number of soft channel bits + uint32_t G; + /// Codebook index for this dlsch (0,1,2,3) + uint8_t codebook_index; + /// Maximum number of HARQ processes + uint8_t Mdlharq; + /// Maximum number of HARQ rounds + uint8_t Mlimit; + /// MIMO transmission mode indicator for this sub-frame + uint8_t Kmimo; + /// Nsoft parameter related to UE Category + uint32_t Nsoft; + /// amplitude of PDSCH (compared to RS) in symbols without pilots + int16_t sqrt_rho_a; + /// amplitude of PDSCH (compared to RS) in symbols containing pilots + int16_t sqrt_rho_b; +} NR_gNB_DLSCH_t; + + typedef struct { /// \brief Pointers (dynamic) to the received data in the time domain. /// - first index: rx antenna [0..nb_antennas_rx[ @@ -313,11 +399,11 @@ typedef struct PHY_VARS_gNB_s { LTE_eNB_SRS srs_vars[NUMBER_OF_UE_MAX]; LTE_eNB_PUSCH *pusch_vars[NUMBER_OF_UE_MAX]; LTE_eNB_PRACH prach_vars; - LTE_eNB_DLSCH_t *dlsch[NUMBER_OF_UE_MAX][2]; // Nusers times two spatial streams + NR_gNB_DLSCH_t *dlsch[NUMBER_OF_UE_MAX][2]; // Nusers times two spatial streams LTE_eNB_ULSCH_t *ulsch[NUMBER_OF_UE_MAX+1]; // Nusers + number of RA - LTE_eNB_DLSCH_t *dlsch_SI,*dlsch_ra,*dlsch_p; - LTE_eNB_DLSCH_t *dlsch_MCH; - LTE_eNB_DLSCH_t *dlsch_PCH; + NR_gNB_DLSCH_t *dlsch_SI,*dlsch_ra,*dlsch_p; + NR_gNB_DLSCH_t *dlsch_MCH; + NR_gNB_DLSCH_t *dlsch_PCH; LTE_eNB_UE_stats UE_stats[NUMBER_OF_UE_MAX]; LTE_eNB_UE_stats *UE_stats_ptr[NUMBER_OF_UE_MAX]; @@ -338,6 +424,9 @@ typedef struct PHY_VARS_gNB_s { /// PDCCH DMRS sequence uint32_t ***nr_gold_pdcch_dmrs; + /// PDSCH DMRS sequence + uint32_t ****nr_gold_pdsch_dmrs; + /// Indicator set to 0 after first SR uint8_t first_sr[NUMBER_OF_UE_MAX]; diff --git a/openair1/PHY/defs_nr_UE.h b/openair1/PHY/defs_nr_UE.h index 6a40eaa822..c11ba8933f 100644 --- a/openair1/PHY/defs_nr_UE.h +++ b/openair1/PHY/defs_nr_UE.h @@ -57,7 +57,6 @@ # include "COMMON/ral_messages_types.h" # include "UTIL/queue.h" # endif -# include "common/utils/LOG/log.h" # define msg(aRGS...) LOG_D(PHY, ##aRGS) # else # define msg printf @@ -864,16 +863,16 @@ typedef struct { int32_t **rho; /// \brief Pointer to llrs, 4-bit resolution. /// - first index: ? [0..48*N_RB_DL[ - uint16_t *llr; + int16_t *llr; /// \brief Pointer to llrs, 16-bit resolution. /// - first index: ? [0..96*N_RB_DL[ - uint16_t *llr16; + int16_t *llr16; /// \brief \f$\overline{w}\f$ from 36-211. /// - first index: ? [0..48*N_RB_DL[ - uint16_t *wbar; + int16_t *wbar; /// \brief PDCCH/DCI e-sequence (input to rate matching). /// - first index: ? [0..96*N_RB_DL[ - int8_t *e_rx; + int16_t *e_rx; /// number of PDCCH symbols in current subframe uint8_t num_pdcch_symbols; /// Allocated CRNTI for UE diff --git a/openair1/PHY/defs_nr_common.h b/openair1/PHY/defs_nr_common.h index 72cb5a05f2..58a8833b18 100644 --- a/openair1/PHY/defs_nr_common.h +++ b/openair1/PHY/defs_nr_common.h @@ -54,9 +54,12 @@ #define ONE_OVER_SQRT2_Q15 23170 #define ONE_OVER_TWO_Q15 16384 -#define NR_MOD_TABLE_SIZE_SHORT 14 +#define NR_MOD_TABLE_SIZE_SHORT 686 #define NR_MOD_TABLE_BPSK_OFFSET 1 #define NR_MOD_TABLE_QPSK_OFFSET 3 +#define NR_MOD_TABLE_QAM16_OFFSET 7 +#define NR_MOD_TABLE_QAM64_OFFSET 23 +#define NR_MOD_TABLE_QAM256_OFFSET 87 #define NR_PSS_LENGTH 127 #define NR_SSS_LENGTH 127 @@ -71,6 +74,9 @@ /*used for the resource mapping*/ #define NR_MAX_PDCCH_DMRS_LENGTH 576 // 16(L)*2(QPSK)*3(3 DMRS symbs per REG)*6(REG per CCE) +#define NR_MAX_PDSCH_DMRS_LENGTH 3300 //275*6(k)*2(QPSK) +#define NR_MAX_PDSCH_DMRS_INIT_LENGTH_DWORD 104 // ceil(NR_MAX_PDSCH_DMRS_LENGTH/32) + #define NR_MAX_DCI_PAYLOAD_SIZE 64 #define NR_MAX_DCI_SIZE 1728 //16(L)*2(QPSK)*9(12 RE per REG - 3(DMRS))*6(REG per CCE) #define NR_MAX_DCI_SIZE_DWORD 54 // ceil(NR_MAX_DCI_SIZE/32) @@ -80,6 +86,13 @@ #define NR_MAX_PDCCH_AGG_LEVEL 16 #define NR_MAX_CSET_DURATION 3 +#define NR_MAX_NB_RBG 18 +#define NR_MAX_NB_LAYERS 8 +#define NR_MAX_NB_CODEWORDS 2 +#define NR_MAX_NB_HARQ_PROCESSES 16 +#define NR_MAX_PDSCH_ENCODED_LENGTH 950984 +#define NR_MAX_PDSCH_TBS 3824 + typedef enum { NR_MU_0=0, NR_MU_1, @@ -109,6 +122,36 @@ typedef enum { nr_FR2 } nr_frequency_range_e; +typedef enum { + MOD_BPSK=0, + MOD_QPSK, + MOD_QAM16, + MOD_QAM64, + MOD_QAM256 +}nr_mod_t; + +typedef struct { + /// Size of first RBG + uint8_t start_size; + /// Nominal size + uint8_t P; + /// Size of last RBG + uint8_t end_size; + /// Number of RBG + uint8_t N_RBG; +}nr_rbg_parms_t; + +typedef struct { + /// Size of first PRG + uint8_t start_size; + /// Nominal size + uint8_t P_prime; + /// Size of last PRG + uint8_t end_size; + /// Number of PRG + uint8_t N_PRG; +} nr_prg_parms_t; + typedef struct NR_BWP_PARMS { /// BWP ID uint8_t bwp_id; @@ -122,6 +165,10 @@ typedef struct NR_BWP_PARMS { uint16_t ofdm_symbol_size; /// Cyclic prefix uint8_t cyclic_prefix; + /// RBG params + nr_rbg_parms_t rbg_parms; + /// PRG params + nr_prg_parms_t prg_parms; } NR_BWP_PARMS; typedef struct { diff --git a/openair1/PHY/phy_extern.h b/openair1/PHY/phy_extern.h index fe888c8d2e..623a4bb0c6 100644 --- a/openair1/PHY/phy_extern.h +++ b/openair1/PHY/phy_extern.h @@ -38,6 +38,7 @@ extern unsigned int TX_DMA_BUFFER[4][NB_ANTENNAS_TX]; extern unsigned int DAQ_MBOX; extern int number_of_cards; +extern short conjugate[8],conjugate2[8]; #ifndef OCP_FRAMEWORK //extern PHY_VARS_eNB ***PHY_vars_eNB_g; diff --git a/openair1/PHY/phy_extern_nr_ue.h b/openair1/PHY/phy_extern_nr_ue.h index 2fa9ef511e..66dbd82fc4 100644 --- a/openair1/PHY/phy_extern_nr_ue.h +++ b/openair1/PHY/phy_extern_nr_ue.h @@ -19,8 +19,8 @@ * contact@openairinterface.org */ -#ifndef __PHY_EXTERN_UE__H__ -#define __PHY_EXTERN_UE__H__ +#ifndef __PHY_EXTERN_NR_UE__H__ +#define __PHY_EXTERN_NR_UE__H__ #include "PHY/defs_nr_UE.h" //#include "common/ran_context.h" @@ -34,6 +34,7 @@ extern unsigned int TX_DMA_BUFFER[4][NB_ANTENNAS_TX]; //#include "PHY/LTE_TRANSPORT/transport_extern.h" +extern short conjugate[8],conjugate2[8]; extern int number_of_cards; diff --git a/openair1/PHY/phy_extern_ue.h b/openair1/PHY/phy_extern_ue.h index 8ae424f5f7..596686f261 100644 --- a/openair1/PHY/phy_extern_ue.h +++ b/openair1/PHY/phy_extern_ue.h @@ -36,6 +36,7 @@ extern unsigned int TX_DMA_BUFFER[4][NB_ANTENNAS_TX]; extern int number_of_cards; +extern short conjugate[8],conjugate2[8]; #ifndef OCP_FRAMEWORK extern PHY_VARS_UE ***PHY_vars_UE_g; diff --git a/openair1/PHY/phy_vars.h b/openair1/PHY/phy_vars.h index e8588ce0b8..897d826ae2 100644 --- a/openair1/PHY/phy_vars.h +++ b/openair1/PHY/phy_vars.h @@ -59,6 +59,8 @@ unsigned short rev[2048],rev_times4[8192],rev_half[1024]; unsigned short rev256[256],rev512[512],rev1024[1024],rev4096[4096],rev2048[2048],rev8192[8192]; +short conjugate[8]__attribute__((aligned(16))) = {-1,1,-1,1,-1,1,-1,1}; +short conjugate2[8]__attribute__((aligned(16))) = {1,-1,1,-1,1,-1,1,-1}; #include "SIMULATION/ETH_TRANSPORT/vars.h" diff --git a/openair1/PHY/phy_vars_nr_ue.h b/openair1/PHY/phy_vars_nr_ue.h index 93a489178a..0e7e4f79d1 100644 --- a/openair1/PHY/phy_vars_nr_ue.h +++ b/openair1/PHY/phy_vars_nr_ue.h @@ -19,8 +19,8 @@ * contact@openairinterface.org */ -#ifndef __PHY_VARS_NR_H__ -#define __PHY_VARS_NR_H__ +#ifndef __PHY_VARS_NR_UE_H__ +#define __PHY_VARS_NR_UE_H__ #include "PHY/types.h" #include "PHY/defs_nr_UE.h" @@ -55,6 +55,9 @@ unsigned short rev[2048],rev_times4[8192],rev_half[1024]; unsigned short rev256[256],rev512[512],rev1024[1024],rev4096[4096],rev2048[2048],rev8192[8192]; +short conjugate[8]__attribute__((aligned(16))) = {-1,1,-1,1,-1,1,-1,1}; +short conjugate2[8]__attribute__((aligned(16))) = {1,-1,1,-1,1,-1,1,-1}; + #include "SIMULATION/ETH_TRANSPORT/vars.h" @@ -143,5 +146,5 @@ int16_t unscrambling_lut[65536*16] __attribute__((aligned(32))); /// lookup table for scrambling in TX uint8_t scrambling_lut[65536*16] __attribute__((aligned(32))); -uint8_t max_turbo_iterations=4; +uint8_t max_ldpc_iterations=4; #endif /*__PHY_VARS_H__ */ diff --git a/openair1/PHY/phy_vars_ue.h b/openair1/PHY/phy_vars_ue.h index e00b302dd7..af3657ce2c 100644 --- a/openair1/PHY/phy_vars_ue.h +++ b/openair1/PHY/phy_vars_ue.h @@ -24,7 +24,6 @@ #include "PHY/types.h" #include "PHY/defs_UE.h" -#include "PHY/phy_vars_ue.h" #include "common/ran_context.h" @@ -54,6 +53,10 @@ unsigned short rev[2048],rev_times4[8192],rev_half[1024]; unsigned short rev256[256],rev512[512],rev1024[1024],rev4096[4096],rev2048[2048],rev8192[8192]; +short conjugate[8]__attribute__((aligned(16))) = {-1,1,-1,1,-1,1,-1,1}; +short conjugate2[8]__attribute__((aligned(16))) = {1,-1,1,-1,1,-1,1,-1}; + + #include "SIMULATION/ETH_TRANSPORT/vars.h" diff --git a/openair1/SCHED_NR/fapi_nr_l1.c b/openair1/SCHED_NR/fapi_nr_l1.c index 94e1127bc5..d663a95c35 100644 --- a/openair1/SCHED_NR/fapi_nr_l1.c +++ b/openair1/SCHED_NR/fapi_nr_l1.c @@ -57,7 +57,8 @@ void handle_nr_nfapi_bch_pdu(PHY_VARS_gNB *gNB, void handle_nfapi_nr_dci_dl_pdu(PHY_VARS_gNB *gNB, int frame, int slot, gNB_L1_rxtx_proc_t *proc, - nfapi_nr_dl_config_request_pdu_t *dl_config_pdu) + nfapi_nr_dl_config_request_pdu_t *dl_config_pdu, + nfapi_nr_dl_config_request_pdu_t *dl_config_dlsch_pdu) { int idx = slot&1; NR_gNB_PDCCH *pdcch_vars = &gNB->pdcch_vars; @@ -65,7 +66,7 @@ void handle_nfapi_nr_dci_dl_pdu(PHY_VARS_gNB *gNB, LOG_I(PHY,"Frame %d, Slot %d: DCI processing - populating pdcch_vars->dci_alloc[%d] proc:slot_tx:%d idx:%d pdcch_vars->num_dci:%d\n",frame,slot, pdcch_vars->num_dci, proc->slot_tx, idx, pdcch_vars->num_dci); // copy dci configuration into gNB structure - nr_fill_dci_and_dlsch(gNB,frame,slot,proc,&pdcch_vars->dci_alloc[pdcch_vars->num_dci],dl_config_pdu); + nr_fill_dci_and_dlsch(gNB,frame,slot,proc,&pdcch_vars->dci_alloc[pdcch_vars->num_dci],&dl_config_pdu->dci_dl_pdu,&dl_config_dlsch_pdu->dlsch_pdu); LOG_I(PHY,"Frame %d, Slot %d: DCI processing - populated pdcch_vars->dci_alloc[%d] proc:slot_tx:%d idx:%d pdcch_vars->num_dci:%d\n",proc->frame_tx,proc->slot_tx, pdcch_vars->num_dci, proc->slot_tx, idx, pdcch_vars->num_dci); } @@ -92,6 +93,7 @@ void nr_schedule_response(NR_Sched_Rsp_t *Sched_INFO){ uint8_t number_dl_pdu = DL_req->dl_config_request_body.number_pdu; nfapi_nr_dl_config_request_pdu_t *dl_config_pdu; + nfapi_nr_dl_config_request_pdu_t *dl_config_dlsch_pdu; int i; @@ -102,6 +104,8 @@ void nr_schedule_response(NR_Sched_Rsp_t *Sched_INFO){ int do_oai =0; int dont_send =0; + gNB->pdcch_vars.num_dci = 0; + gNB->pdcch_vars.num_pdsch_rnti = 0; gNB->pdcch_vars.num_dci=0; @@ -125,11 +129,14 @@ void nr_schedule_response(NR_Sched_Rsp_t *Sched_INFO){ break; case NFAPI_NR_DL_CONFIG_DCI_DL_PDU_TYPE: + dl_config_dlsch_pdu = &DL_req->dl_config_request_body.dl_config_pdu_list[++i]; handle_nfapi_nr_dci_dl_pdu(gNB, frame, slot, proc, - dl_config_pdu); + dl_config_pdu, + dl_config_dlsch_pdu); gNB->pdcch_vars.num_dci++; + gNB->pdcch_vars.num_pdsch_rnti++; do_oai=1; break; } diff --git a/openair1/SCHED_NR/fapi_nr_l1.h b/openair1/SCHED_NR/fapi_nr_l1.h index 4c6b25e969..66eb0fada7 100644 --- a/openair1/SCHED_NR/fapi_nr_l1.h +++ b/openair1/SCHED_NR/fapi_nr_l1.h @@ -42,4 +42,5 @@ void nr_schedule_response(NR_Sched_Rsp_t *Sched_INFO); void handle_nfapi_nr_dci_dl_pdu(PHY_VARS_gNB *gNB, int frame, int subframe, gNB_L1_rxtx_proc_t *proc, - nfapi_nr_dl_config_request_pdu_t *dl_config_pdu); + nfapi_nr_dl_config_request_pdu_t *dl_config_pdu, + nfapi_nr_dl_config_request_pdu_t *dl_config_dlsch_pdu); diff --git a/openair1/SCHED_NR/nr_ru_procedures.c b/openair1/SCHED_NR/nr_ru_procedures.c index 67a22869b4..65dfb64734 100644 --- a/openair1/SCHED_NR/nr_ru_procedures.c +++ b/openair1/SCHED_NR/nr_ru_procedures.c @@ -130,6 +130,7 @@ void nr_feptx_ofdm_2thread(RU_t *ru) { // this copy should be done in the precoding thread (currently inactive) for (int aa=0;aanb_tx;aa++) memcpy((void*)ru->common.txdataF_BF[aa], + (void*)ru->gNB_list[0]->common_vars.txdataF[aa], fp->samples_per_slot_wCP*sizeof(int32_t)); VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PHY_PROCEDURES_RU_FEPTX_OFDM , 1 ); diff --git a/openair1/SCHED_NR/phy_procedures_nr_gNB.c b/openair1/SCHED_NR/phy_procedures_nr_gNB.c index 050b601c54..81c7cd4693 100644 --- a/openair1/SCHED_NR/phy_procedures_nr_gNB.c +++ b/openair1/SCHED_NR/phy_procedures_nr_gNB.c @@ -23,6 +23,7 @@ #include "PHY/defs_gNB.h" #include "sched_nr.h" #include "PHY/NR_TRANSPORT/nr_transport.h" +#include "PHY/NR_TRANSPORT/nr_dlsch.h" #include "SCHED/sched_eNB.h" #include "SCHED/sched_common_extern.h" #include "nfapi_interface.h" @@ -165,8 +166,9 @@ void phy_procedures_gNB_TX(PHY_VARS_gNB *gNB, { int aa; int frame=proc->frame_tx; + int slot=proc->slot_tx; - uint8_t num_dci=0; + uint8_t num_dci=0,num_pdsch_rnti; NR_DL_FRAME_PARMS *fp=&gNB->frame_parms; nfapi_nr_config_request_t *cfg = &gNB->gNB_config; @@ -190,17 +192,27 @@ void phy_procedures_gNB_TX(PHY_VARS_gNB *gNB, } num_dci = gNB->pdcch_vars.num_dci; + + num_pdsch_rnti = gNB->pdcch_vars.num_pdsch_rnti; if (num_dci) { LOG_I(PHY, "[gNB %d] Frame %d slot %d \ Calling nr_generate_dci_top (number of DCI %d)\n", gNB->Mod_id, frame, slot, num_dci); - - if (nfapi_mode == 0 || nfapi_mode == 1) + if (nfapi_mode == 0 || nfapi_mode == 1){ nr_generate_dci_top(gNB->pdcch_vars, &gNB->nrPolar_params, gNB->nr_gold_pdcch_dmrs[slot], gNB->common_vars.txdataF[0], AMP, *fp, *cfg); + if (num_pdsch_rnti) { + LOG_I(PHY, "PDSCH generation started (%d)\n", num_pdsch_rnti); + nr_generate_pdsch(*gNB->dlsch[0][0], + gNB->pdcch_vars.dci_alloc[0], + gNB->nr_gold_pdsch_dmrs[slot], + gNB->common_vars.txdataF, + AMP, slot, *fp, *cfg); + } + } } VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PHY_PROCEDURES_ENB_TX+offset,0); diff --git a/openair1/SCHED_NR_UE/defs.h b/openair1/SCHED_NR_UE/defs.h index d0ac5d3d6b..5a8dc96fab 100644 --- a/openair1/SCHED_NR_UE/defs.h +++ b/openair1/SCHED_NR_UE/defs.h @@ -112,7 +112,7 @@ int phy_procedures_RN_UE_RX(unsigned char last_slot, unsigned char next_slot, re @param mode calib/normal mode @param r_type indicates the relaying operation: 0: no_relaying, 1: unicast relaying type 1, 2: unicast relaying type 2, 3: multicast relaying */ -void phy_procedures_UE_TX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id,uint8_t abstraction_flag,runmode_t mode,relaying_type_t r_type); +void phy_procedures_nrUE_TX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id,uint8_t abstraction_flag,runmode_t mode,relaying_type_t r_type); /*! \brief Scheduling for UE RX procedures in normal subframes. @param last_slot Index of last slot (0-19) @param phy_vars_ue Pointer to UE variables on which to act @@ -124,7 +124,8 @@ void phy_procedures_UE_TX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB @param phy_vars_rn pointer to RN variables */ int phy_procedures_nrUE_RX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id,uint8_t do_pdcch_flag,runmode_t mode); -int phy_procedures_slot_parallelization_UE_RX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id,uint8_t abstraction_flag,uint8_t do_pdcch_flag,runmode_t mode,relaying_type_t r_type); +int phy_procedures_slot_parallelization_nrUE_RX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id,uint8_t abstraction_flag,uint8_t do_pdcch_flag,runmode_t mode,relaying_type_t r_type); + #ifdef UE_SLOT_PARALLELISATION void *UE_thread_slot1_dl_processing(void *arg); @@ -136,7 +137,7 @@ void *UE_thread_slot1_dl_processing(void *arg); @param abstraction_flag Indicator of PHY abstraction @param r_type indicates the relaying operation: 0: no_relaying, 1: unicast relaying type 1, 2: unicast relaying type 2, 3: multicast relaying */ -void phy_procedures_UE_S_TX(PHY_VARS_NR_UE *ue,uint8_t eNB_id,uint8_t abstraction_flag,relaying_type_t r_type); +//void phy_procedures_UE_S_TX(PHY_VARS_NR_UE *ue,uint8_t eNB_id,uint8_t abstraction_flag,relaying_type_t r_type); /*! \brief Scheduling for UE RX procedures in TDD S-subframes. @param phy_vars_ue Pointer to UE variables on which to act @@ -144,7 +145,7 @@ void phy_procedures_UE_S_TX(PHY_VARS_NR_UE *ue,uint8_t eNB_id,uint8_t abstractio @param abstraction_flag Indicator of PHY abstraction @param r_type indicates the relaying operation: 0: no_relaying, 1: unicast relaying type 1, 2: unicast relaying type 2, 3: multicast relaying */ -void phy_procedures_UE_S_RX(PHY_VARS_NR_UE *ue,uint8_t eNB_id,uint8_t abstraction_flag, relaying_type_t r_type); +//void phy_procedures_UE_S_RX(PHY_VARS_NR_UE *ue,uint8_t eNB_id,uint8_t abstraction_flag, relaying_type_t r_type); /*! \brief Function to compute subframe type as a function of Frame type and TDD Configuration (implements Table 4.2.2 from 36.211, p.11 from version 8.6) and subframe index. @@ -154,7 +155,7 @@ void phy_procedures_UE_S_RX(PHY_VARS_NR_UE *ue,uint8_t eNB_id,uint8_t abstractio */ -nr_subframe_t nr_subframe_select(NR_DL_FRAME_PARMS *frame_parms,uint8_t subframe); +//nr_subframe_t nr_subframe_select(NR_DL_FRAME_PARMS *frame_parms,uint8_t subframe); /*! \brief Function to compute subframe type as a function of Frame type and TDD Configuration (implements Table 4.2.2 from 36.211, p.11 from version 8.6) and subframe index. Same as nr_subframe_select, except that it uses the Mod_id and is provided as a service to the MAC scheduler. @@ -368,6 +369,11 @@ int nr_is_srs_occasion_common(NR_DL_FRAME_PARMS *frame_parms,int frame_tx,int su void nr_compute_srs_pos(lte_frame_type_t frameType,uint16_t isrs,uint16_t *psrsPeriodicity,uint16_t *psrsOffset); +void set_tx_harq_id(NR_UE_ULSCH_t *ulsch, int harq_pid, int slot_tx); +int get_tx_harq_id(NR_UE_ULSCH_t *ulsch, int slot_tx); + + + /*@}*/ diff --git a/openair1/SCHED_NR_UE/fapi_nr_ue_l1.c b/openair1/SCHED_NR_UE/fapi_nr_ue_l1.c index a27af95632..9e4027ab13 100644 --- a/openair1/SCHED_NR_UE/fapi_nr_ue_l1.c +++ b/openair1/SCHED_NR_UE/fapi_nr_ue_l1.c @@ -42,6 +42,7 @@ extern PHY_VARS_NR_UE ***PHY_vars_UE_g; int8_t nr_ue_scheduled_response(nr_scheduled_response_t *scheduled_response){ + /// module id module_id_t module_id = scheduled_response->module_id; /// component carrier id @@ -106,7 +107,8 @@ int8_t nr_ue_scheduled_response(nr_scheduled_response_t *scheduled_response){ //dlsch0->harq_processes[0]->mcs = &dlsch_config_pdu->mcs; - NR_DL_UE_HARQ_t dlsch0_harq = dlsch0->harq_processes[current_harq_pid]; + NR_DL_UE_HARQ_t dlsch0_harq = *(dlsch0->harq_processes[current_harq_pid]); + //dlsch0->harq_processes[current_harq_pid]->nb_rb = dlsch_config_pdu->number_rbs; diff --git a/openair1/SCHED_NR_UE/harq_nr.c b/openair1/SCHED_NR_UE/harq_nr.c index 93a3f998cf..23d828ec93 100644 --- a/openair1/SCHED_NR_UE/harq_nr.c +++ b/openair1/SCHED_NR_UE/harq_nr.c @@ -120,7 +120,7 @@ void get_dci_info_for_harq(PHY_VARS_NR_UE *ue, NR_DCI_INFO_EXTRACTED_t *nr_dci_i dlsch[0]->current_harq_pid = nr_dci_info_extracted->harq_process_number; - NR_DL_UE_HARQ_t *dl_harq = &dlsch[0]->harq_processes[dlsch[0]->current_harq_pid]; + NR_DL_UE_HARQ_t *dl_harq = dlsch[0]->harq_processes[dlsch[0]->current_harq_pid]; dl_harq->harq_ack.vDAI_DL = nr_dci_info_extracted->dai+1; dl_harq->harq_ack.pucch_resource_indicator = nr_dci_info_extracted->pucch_resource_ind; @@ -401,9 +401,9 @@ void config_downlink_harq_process(PHY_VARS_NR_UE *ue, int gNB_id, int TB_id, int memset(&dlsch->harq_processes[harq_pid],0,sizeof(NR_DL_UE_HARQ_t)); - NR_DL_UE_HARQ_t dl_harq = dlsch->harq_processes[harq_pid]; + NR_DL_UE_HARQ_t *dl_harq = dlsch->harq_processes[harq_pid]; - init_downlink_harq_status(&dl_harq); + init_downlink_harq_status(dl_harq); } } diff --git a/openair1/SCHED_NR_UE/phy_procedures_nr_ue.c b/openair1/SCHED_NR_UE/phy_procedures_nr_ue.c index 400e46ba1f..02384f95d2 100644 --- a/openair1/SCHED_NR_UE/phy_procedures_nr_ue.c +++ b/openair1/SCHED_NR_UE/phy_procedures_nr_ue.c @@ -40,21 +40,23 @@ #include "PHY/phy_extern_nr_ue.h" #include "PHY/NR_REFSIG/refsig_defs_ue.h" #include "PHY/MODULATION/modulation_UE.h" -//#include "PHY/NR_UE_TRANSPORT/nr_transport_ue.h" +#include "PHY/NR_UE_TRANSPORT/nr_transport_ue.h" +#include "PHY/NR_UE_TRANSPORT/nr_transport_proto_ue.h" //#include "PHY/extern.h" #include "SCHED_NR_UE/defs.h" #include "SCHED_NR/extern.h" //#include //#include "targets/RT/USER/nr-softmodem.h" +#include "PHY/NR_UE_ESTIMATION/nr_estimation.h" #ifdef EMOS #include "SCHED/phy_procedures_emos.h" #endif -#define DEBUG_PHY_PROC +//#define DEBUG_PHY_PROC #define NR_PDCCH_SCHED -#define NR_PDCCH_SCHED_DEBUG +//#define NR_PDCCH_SCHED_DEBUG //#define NR_PUCCH_SCHED //#define NR_PUCCH_SCHED_DEBUG @@ -91,10 +93,6 @@ char mode_string[4][20] = {"NOT SYNCHED","PRACH","RAR","PUSCH"}; extern double cpuf; -#ifdef LOG_D -#undef LOG_D -#define LOG_D(A,B...) printf(B) -#endif int32_t nr_rx_pdcch(PHY_VARS_NR_UE *ue, uint32_t frame, @@ -142,6 +140,7 @@ int nr_generate_ue_ul_dlsch_params_from_dci(PHY_VARS_NR_UE *ue, uint16_t crc_scrambled_values[TOTAL_NBR_SCRAMBLED_VALUES], NR_DCI_INFO_EXTRACTED_t *nr_dci_info_extracted); + #if defined(EXMIMO) || defined(OAI_USRP) || defined(OAI_BLADERF) || defined(OAI_LMSSDR) || defined(OAI_ADRV9371_ZC706) extern uint32_t downlink_frequency[MAX_NUM_CCs][4]; #endif @@ -1457,143 +1456,7 @@ void ulsch_common_procedures(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, uint8_ VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PHY_PROCEDURES_UE_TX_ULSCH_COMMON,VCD_FUNCTION_OUT); } -#endif //0 -void ue_nr_prach_procedures(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id,uint8_t abstraction_flag,runmode_t mode) { - - int frame_tx = proc->frame_tx; - int nr_tti_tx = proc->nr_tti_tx; - int prach_power; - PRACH_RESOURCES_t prach_resources_local; - - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PHY_PROCEDURES_UE_TX_PRACH, VCD_FUNCTION_IN); - - ue->generate_prach=0; - - if (ue->mac_enabled==0){ - ue->prach_resources[eNB_id] = &prach_resources_local; - prach_resources_local.ra_RNTI = 0xbeef; - prach_resources_local.ra_PreambleIndex = 0; - } - - if (ue->mac_enabled==1){ - // ask L2 for RACH transport - if ((mode != rx_calib_ue) && (mode != rx_calib_ue_med) && (mode != rx_calib_ue_byp) && (mode != no_L2_connect) ) { - LOG_D(PHY,"Getting PRACH resources\n"); - //ue->prach_resources[eNB_id] = mac_xface->ue_get_rach(ue->Mod_id, - // ue->CC_id, - // frame_tx, - // eNB_id, - // nr_tti_tx); - //LOG_D(PHY,"Got prach_resources for eNB %d address %p, RRCCommon %p\n",eNB_id,ue->prach_resources[eNB_id],UE_mac_inst[ue->Mod_id].radioResourceConfigCommon); - LOG_D(PHY,"Prach resources %p\n",ue->prach_resources[eNB_id]); - } - } - - if (ue->prach_resources[eNB_id]!=NULL) { - - ue->generate_prach=1; - ue->prach_cnt=0; -#ifdef SMBV - ue->prach_resources[eNB_id]->ra_PreambleIndex = 19; -#endif -#ifdef OAI_EMU - ue->prach_PreambleIndex=ue->prach_resources[eNB_id]->ra_PreambleIndex; -#endif - - if (abstraction_flag == 0) { - - LOG_I(PHY,"mode %d\n",mode); - - if ((ue->mac_enabled==1) && (mode != calib_prach_tx)) { - ue->tx_power_dBm[nr_tti_tx] = ue->prach_resources[eNB_id]->ra_PREAMBLE_RECEIVED_TARGET_POWER+get_PL(ue->Mod_id,ue->CC_id,eNB_id); - } - else { - ue->tx_power_dBm[nr_tti_tx] = ue->tx_power_max_dBm; - ue->prach_resources[eNB_id]->ra_PreambleIndex = 19; - } - - LOG_I(PHY,"[UE %d][RAPROC] Frame %d, nr_tti_rx %d : Generating PRACH, preamble %d,PL %d, P0_PRACH %d, TARGET_RECEIVED_POWER %d dBm, PRACH TDD Resource index %d, RA-RNTI %d\n", - ue->Mod_id, - frame_tx, - nr_tti_tx, - ue->prach_resources[eNB_id]->ra_PreambleIndex, - get_PL(ue->Mod_id,ue->CC_id,eNB_id), - ue->tx_power_dBm[nr_tti_tx], - ue->prach_resources[eNB_id]->ra_PREAMBLE_RECEIVED_TARGET_POWER, - ue->prach_resources[eNB_id]->ra_TDD_map_index, - ue->prach_resources[eNB_id]->ra_RNTI); - - ue->tx_total_RE[nr_tti_tx] = 96; - -#if defined(EXMIMO) || defined(OAI_USRP) || defined(OAI_BLADERF) || defined(OAI_LMSSDR) || defined(OAI_ADRV9371_ZC706) - ue->prach_vars[eNB_id]->amp = get_tx_amp(ue->tx_power_dBm[nr_tti_tx], - ue->tx_power_max_dBm, - ue->frame_parms.N_RB_UL, - 6); -#else - ue->prach_vars[eNB_id]->amp = AMP; -#endif - if ((mode == calib_prach_tx) && (((proc->frame_tx&0xfffe)%100)==0)) - LOG_D(PHY,"[UE %d][RAPROC] Frame %d, nr_tti_rx %d : PRACH TX power %d dBm, amp %d\n", - ue->Mod_id, - proc->frame_rx, - proc->nr_tti_tx, - ue->tx_power_dBm[nr_tti_tx], - ue->prach_vars[eNB_id]->amp); - - - // start_meas(&ue->tx_prach); - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_GENERATE_PRACH, VCD_FUNCTION_IN); - - //prach_power = generate_prach(ue,eNB_id,nr_tti_tx,frame_tx); - prach_power = generate_nr_prach(ue,eNB_id,nr_tti_tx,frame_tx); - - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_GENERATE_PRACH, VCD_FUNCTION_OUT); - // stop_meas(&ue->tx_prach); - LOG_D(PHY,"[UE %d][RAPROC] PRACH PL %d dB, power %d dBm, digital power %d dB (amp %d)\n", - ue->Mod_id, - get_PL(ue->Mod_id,ue->CC_id,eNB_id), - ue->tx_power_dBm[nr_tti_tx], - dB_fixed(prach_power), - ue->prach_vars[eNB_id]->amp); - } else { - //UE_transport_info[ue->Mod_id][ue->CC_id].cntl.prach_flag=1; - //UE_transport_info[ue->Mod_id][ue->CC_id].cntl.prach_id=ue->prach_resources[eNB_id]->ra_PreambleIndex; - } - - if (ue->mac_enabled==1){ - //mac_xface->Msg1_transmitted(ue->Mod_id, - // ue->CC_id, - // frame_tx, - // eNB_id); - } - - LOG_I(PHY,"[UE %d][RAPROC] Frame %d, nr_tti_rx %d: Generating PRACH (eNB %d) preamble index %d for UL, TX power %d dBm (PL %d dB), l3msg \n", - ue->Mod_id,frame_tx,nr_tti_tx,eNB_id, - ue->prach_resources[eNB_id]->ra_PreambleIndex, - ue->prach_resources[eNB_id]->ra_PREAMBLE_RECEIVED_TARGET_POWER+get_PL(ue->Mod_id,ue->CC_id,eNB_id), - get_PL(ue->Mod_id,ue->CC_id,eNB_id)); - - } - - - // if we're calibrating the PRACH kill the pointer to its resources so that the RA protocol doesn't continue - if (mode == calib_prach_tx) - ue->prach_resources[eNB_id]=NULL; - - LOG_D(PHY,"[UE %d] frame %d nr_tti_rx %d : generate_prach %d, prach_cnt %d\n", - ue->Mod_id,frame_tx,nr_tti_tx,ue->generate_prach,ue->prach_cnt); - - ue->prach_cnt++; - - if (ue->prach_cnt==3) - ue->generate_prach=0; - - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PHY_PROCEDURES_UE_TX_PRACH, VCD_FUNCTION_OUT); -} - -#if 0 void ue_prach_procedures(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id,uint8_t abstraction_flag,runmode_t mode) { int frame_tx = proc->frame_tx; @@ -2625,195 +2488,8 @@ void ue_pucch_procedures(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_ VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PHY_PROCEDURES_UE_TX_PUCCH,VCD_FUNCTION_OUT); } -#endif //0 - -void phy_procedures_nrUE_TX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id,uint8_t abstraction_flag,runmode_t mode,relaying_type_t r_type) { - NR_DL_FRAME_PARMS *frame_parms=&ue->frame_parms; - //int32_t ulsch_start=0; - int nr_tti_tx = proc->nr_tti_tx; - int frame_tx = proc->frame_tx; - unsigned int aa; - uint8_t isSubframeSRS; - - uint8_t next1_thread_id = ue->current_thread_id[proc->nr_tti_rx]== (RX_NB_TH-1) ? 0:(ue->current_thread_id[proc->nr_tti_rx]+1); - uint8_t next2_thread_id = next1_thread_id== (RX_NB_TH-1) ? 0:(next1_thread_id+1); - - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PHY_PROCEDURES_UE_TX,VCD_FUNCTION_IN); - - LOG_D(PHY,"****** start TX-Chain for AbsSubframe %d.%d ******\n", frame_tx, nr_tti_tx); -#if T_TRACER - T(T_UE_PHY_UL_TICK, T_INT(ue->Mod_id), T_INT(frame_tx%1024), T_INT(nr_tti_tx)); -#endif - - ue->generate_ul_signal[eNB_id] = 0; -#if UE_TIMING_TRACE - start_meas(&ue->phy_proc_tx); -#endif - -#ifdef EMOS - //phy_procedures_emos_UE_TX(next_slot); -#endif - - ue->tx_power_dBm[nr_tti_tx]=-127; - - if (abstraction_flag==0) { - for (aa=0; aanb_antennas_tx; aa++) { - memset(&ue->common_vars.txdataF[aa][nr_tti_tx*frame_parms->ofdm_symbol_size*frame_parms->symbols_per_tti], - 0, - frame_parms->ofdm_symbol_size*frame_parms->symbols_per_tti*sizeof(int32_t)); - } - } - - if (ue->UE_mode[eNB_id] != PRACH) { - // check cell srs nr_tti_rx and ue srs nr_tti_rx. This has an impact on pusch encoding - //isSubframeSRS = nr_is_srs_occasion_common(&ue->frame_parms,proc->frame_tx,proc->nr_tti_tx); - - //ue_compute_srs_occasion(ue,proc,eNB_id,isSubframeSRS); - - //ue_ulsch_uespec_procedures(ue,proc,eNB_id,abstraction_flag); - - } - - if (ue->UE_mode[eNB_id] == PUSCH) { - // check if we need to use PUCCH 1a/1b - //ue_pucch_procedures(ue,proc,eNB_id,abstraction_flag); - // check if we need to use SRS - //ue_srs_procedures(ue,proc,eNB_id,abstraction_flag); - } // UE_mode==PUSCH - - -#ifdef CBA - - if ((ue->ulsch[eNB_id]->harq_processes[harq_pid]->subframe_cba_scheduling_flag >= 1) && - (ue->ulsch[eNB_id]->harq_processes[harq_pid]->status == CBA_ACTIVE)) { - ue->ulsch[eNB_id]->harq_processes[harq_pid]->subframe_scheduling_flag=0; //-=1 - // ue->ulsch[eNB_id]->harq_processes[harq_pid]->status= IDLE; - first_rb = ue->ulsch[eNB_id]->harq_processes[harq_pid]->first_rb; - nb_rb = ue->ulsch[eNB_id]->harq_processes[harq_pid]->nb_rb; - //cba_mcs=ue->ulsch[eNB_id]->harq_processes[harq_pid]->mcs; - input_buffer_length = ue->ulsch[eNB_id]->harq_processes[harq_pid]->TBS/8; - access_mode=CBA_ACCESS; - - LOG_D(PHY,"[UE %d] Frame %d, nr_tti_rx %d: CBA num dci %d\n", - Mod_id,frame_tx,nr_tti_tx, - ue->ulsch[eNB_id]->num_cba_dci[nr_tti_tx]); - - /*mac_xface->ue_get_sdu(Mod_id, - CC_id, - frame_tx, - proc->subframe_tx, - nr_tti_tx%(ue->frame_parms.ttis_per_subframe), - eNB_id, - ulsch_input_buffer, - input_buffer_length, - &access_mode);*/ - - ue->ulsch[eNB_id]->num_cba_dci[nr_tti_tx]=0; - - if (access_mode > UNKNOWN_ACCESS) { - - if (abstraction_flag==0) { - if (ulsch_encoding(ulsch_input_buffer, - ue, - harq_pid, - eNB_id, - proc->nr_tti_rx, - ue->transmission_mode[eNB_id],0, - 0)!=0) { // Nbundled, to be updated!!!! - LOG_E(PHY,"ulsch_coding.c: FATAL ERROR: returning\n"); - return; - } - } - -#ifdef PHY_ABSTRACTION - else { - ulsch_encoding_emul(ulsch_input_buffer,ue,eNB_id,proc->nr_tti_rx,harq_pid,0); - } - -#endif - } else { - ue->ulsch[eNB_id]->harq_processes[harq_pid]->status= IDLE; - //reset_cba_uci(ue->ulsch[eNB_id]->o); - LOG_N(PHY,"[UE %d] Frame %d, nr_tti_rx %d: CBA transmission cancelled or postponed\n", - Mod_id, frame_tx,nr_tti_tx); - } - } - -#endif // end CBA - - - if (abstraction_flag == 0) { - //ulsch_common_procedures(ue,proc, (ue->generate_ul_signal[eNB_id] == 0)); - } // mode != PRACH - - - if ((ue->UE_mode[eNB_id] == PRACH) && - (ue->frame_parms.prach_config_common.prach_Config_enabled==1)) { - - // check if we have PRACH opportunity - - if (is_nr_prach_subframe(&ue->frame_parms,frame_tx,nr_tti_tx)) { - - ue_nr_prach_procedures(ue,proc,eNB_id,abstraction_flag,mode); - } - } // mode is PRACH - else { - ue->generate_prach=0; - } - - // reset DL ACK/NACK status - uint8_t N_bundled = 0; - uint8_t current_pid = ue->dlsch[ue->current_thread_id[proc->nr_tti_rx]][eNB_id][0]->current_harq_pid; - if (ue->dlsch[ue->current_thread_id[proc->nr_tti_rx]][eNB_id][0] != NULL) - { - /* - nr_reset_ack(&ue->frame_parms, - ue->dlsch[ue->current_thread_id[proc->nr_tti_rx]][eNB_id][0]->harq_processes[current_pid]->harq_ack, - nr_tti_tx, - proc->nr_tti_rx, - ue->ulsch[eNB_id]->o_ACK, - &N_bundled, - 0); - nr_reset_ack(&ue->frame_parms, - ue->dlsch[next1_thread_id][eNB_id][0]->harq_processes[current_pid]->harq_ack, - nr_tti_tx, - proc->nr_tti_rx, - ue->ulsch[eNB_id]->o_ACK, - &N_bundled, - 0); - nr_reset_ack(&ue->frame_parms, - ue->dlsch[next2_thread_id][eNB_id][0]->harq_processes[current_pid]->harq_ack, - nr_tti_tx, - proc->nr_tti_rx, - ue->ulsch[eNB_id]->o_ACK, - &N_bundled, - 0); - */ - } - - if (ue->dlsch_SI[eNB_id] != NULL) - /* - nr_reset_ack(&ue->frame_parms, - ue->dlsch_SI[eNB_id]->harq_processes[current_pid]->harq_ack, - nr_tti_tx, - proc->nr_tti_rx, - ue->ulsch[eNB_id]->o_ACK, - &N_bundled, - 0); - */ - - LOG_D(PHY,"****** end TX-Chain for AbsSubframe %d.%d ******\n", frame_tx, nr_tti_tx); - - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PHY_PROCEDURES_UE_TX, VCD_FUNCTION_OUT); -#if UE_TIMING_TRACE - stop_meas(&ue->phy_proc_tx); -#endif -} - - -#if 0 void phy_procedures_UE_TX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id,uint8_t abstraction_flag,runmode_t mode,relaying_type_t r_type) { @@ -2994,9 +2670,7 @@ void phy_procedures_UE_TX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB stop_meas(&ue->phy_proc_tx); #endif } -#endif //0 -#if 0 void phy_procedures_UE_S_TX(PHY_VARS_NR_UE *ue,uint8_t eNB_id,uint8_t abstraction_flag,relaying_type_t r_type) { int aa;//i,aa; @@ -3020,19 +2694,24 @@ void phy_procedures_UE_S_TX(PHY_VARS_NR_UE *ue,uint8_t eNB_id,uint8_t abstractio } } -void ue_measurement_procedures( - uint16_t l, // symbol index of each slot [0..6] - PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc, uint8_t eNB_id, - uint16_t slot, // slot index of each radio frame [0..19] - uint8_t abstraction_flag,runmode_t mode) +#endif + +void nr_ue_measurement_procedures( + uint16_t l, // symbol index of each slot [0..6] + PHY_VARS_NR_UE *ue, + UE_nr_rxtx_proc_t *proc, + uint8_t eNB_id, + uint16_t slot, // slot index of each radio frame [0..19] + runmode_t mode) { - //LOG_I(PHY,"ue_measurement_procedures l %d Ncp %d\n",l,ue->frame_parms.Ncp); + LOG_D(PHY,"ue_measurement_procedures l %u Ncp %d\n",l,ue->frame_parms.Ncp); NR_DL_FRAME_PARMS *frame_parms=&ue->frame_parms; int nr_tti_rx = proc->nr_tti_rx; +#if 0 VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_MEASUREMENT_PROCEDURES, VCD_FUNCTION_IN); if (l==0) { @@ -3078,16 +2757,14 @@ void ue_measurement_procedures( abstraction_flag); VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_RRC_MEASUREMENTS, VCD_FUNCTION_OUT); - if (abstraction_flag==1) - ue->sinr_eff = sinr_eff_cqi_calc(ue, 0, nr_tti_rx); } - +#endif // accumulate and filter timing offset estimation every nr_tti_rx (instead of every frame) - if (( (slot%2) == 0) && (l==(4-frame_parms->Ncp))) { + if (( slot == 2) && (l==(1-frame_parms->Ncp))) { // AGC - +/* VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_GAIN_CONTROL, VCD_FUNCTION_IN); #ifndef OAI_USRP @@ -3101,24 +2778,27 @@ void ue_measurement_procedures( #endif VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_GAIN_CONTROL, VCD_FUNCTION_OUT); - +*/ eNB_id = 0; - if (abstraction_flag == 0) { - if (ue->no_timing_correction==0) - lte_adjust_synch(&ue->frame_parms, - ue, - eNB_id, - nr_tti_rx, - 0, - 16384); - } + + LOG_D(PHY,"start adjust sync l = %d slot = %d no timing %d\n",l, slot, ue->no_timing_correction); + if (ue->no_timing_correction==0) + nr_adjust_synch_ue(&ue->frame_parms, + ue, + eNB_id, + nr_tti_rx, + 0, + 16384); + + } VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_MEASUREMENT_PROCEDURES, VCD_FUNCTION_OUT); } + #ifdef EMOS void phy_procedures_emos_UE_RX(PHY_VARS_NR_UE *ue,uint8_t last_slot,uint8_t eNB_id) { @@ -3207,7 +2887,7 @@ void phy_procedures_emos_UE_RX(PHY_VARS_NR_UE *ue,uint8_t last_slot,uint8_t eNB_ } #endif - +#if 0 void restart_phy(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc, uint8_t eNB_id,uint8_t abstraction_flag) { @@ -3271,14 +2951,15 @@ void restart_phy(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc, uint8_t eNB_id,uint //ue->bitrate[eNB_id] = 0; //ue->total_received_bits[eNB_id] = 0; } +#endif //(0) - -void ue_pbch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc, uint8_t abstraction_flag) +void nr_ue_pbch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc, uint8_t abstraction_flag) { // int i; - int pbch_tx_ant=0; - uint8_t pbch_phase; + //int pbch_tx_ant=0; + //uint8_t pbch_phase; + int ret = 0; uint16_t frame_tx; static uint8_t first_run = 1; uint8_t pbch_trials = 0; @@ -3290,151 +2971,34 @@ void ue_pbch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *pro VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_PBCH_PROCEDURES, VCD_FUNCTION_IN); - pbch_phase=(frame_rx%4); - - if (pbch_phase>=4) - pbch_phase=0; - - for (pbch_trials=0; pbch_trials<4; pbch_trials++) { - //for (pbch_phase=0;pbch_phase<4;pbch_phase++) { - //LOG_I(PHY,"[UE %d] Frame %d, Trying PBCH %d (NidCell %d, eNB_id %d)\n",ue->Mod_id,frame_rx,pbch_phase,ue->frame_parms.Nid_cell,eNB_id); - if (abstraction_flag == 0) { - pbch_tx_ant = rx_pbch(ue, proc, - ue->pbch_vars[eNB_id], - &ue->frame_parms, - eNB_id, - ue->frame_parms.mode1_flag==1?SISO:ALAMOUTI, - ue->high_speed_flag, - pbch_phase); - - - - } - -#ifdef PHY_ABSTRACTION - else { - pbch_tx_ant = rx_pbch_emul(ue, - eNB_id, - pbch_phase); - } - -#endif - - if ((pbch_tx_ant>0) && (pbch_tx_ant<=4)) { - break; - } - - pbch_phase++; - - if (pbch_phase>=4) - pbch_phase=0; - } - - - - if ((pbch_tx_ant>0) && (pbch_tx_ant<=4)) { - - if (opt_enabled) { - static uint8_t dummy[3]; - dummy[0] = ue->pbch_vars[eNB_id]->decoded_output[2]; - dummy[1] = ue->pbch_vars[eNB_id]->decoded_output[1]; - dummy[2] = ue->pbch_vars[eNB_id]->decoded_output[0]; - trace_pdu(1, dummy, 3, ue->Mod_id, 0, 0, - frame_rx, nr_tti_rx, 0, 0); - LOG_D(OPT,"[UE %d][PBCH] Frame %d trace pdu for PBCH\n", - ue->Mod_id, nr_tti_rx); - } - - if (pbch_tx_ant>2) { - LOG_W(PHY,"[openair][SCHED][SYNCH] PBCH decoding: pbch_tx_ant>2 not supported\n"); - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_PBCH_PROCEDURES, VCD_FUNCTION_OUT); - return; - } - + //LOG_I(PHY,"[UE %d] Frame %d, Trying PBCH %d (NidCell %d, eNB_id %d)\n",ue->Mod_id,frame_rx,pbch_phase,ue->frame_parms.Nid_cell,eNB_id); + ret = nr_rx_pbch(ue, proc, + ue->pbch_vars[eNB_id], + &ue->frame_parms, + eNB_id, + SISO, + ue->high_speed_flag); + if (ret==0) { ue->pbch_vars[eNB_id]->pdu_errors_conseq = 0; - frame_tx = (((int)(ue->pbch_vars[eNB_id]->decoded_output[2]&0x03))<<8); - frame_tx += ((int)(ue->pbch_vars[eNB_id]->decoded_output[1]&0xfc)); - frame_tx += pbch_phase; - //if (ue->mac_enabled==1) { - //mac_xface->dl_phy_sync_success(ue->Mod_id,frame_rx,eNB_id,first_run); - //} - -#ifdef EMOS - //emos_dump_UE.frame_tx = frame_tx; - //emos_dump_UE.mimo_mode = ue->pbch_vars[eNB_id]->decoded_output[1]; -#endif - - if (first_run) { - first_run = 0; - - proc->frame_rx = (proc->frame_rx & 0xFFFFFC00) | (frame_tx & 0x000003FF); - proc->frame_tx = proc->frame_rx; - for(int th_id=0; th_idproc.proc_rxtx[th_id].frame_rx = proc->frame_rx; - ue->proc.proc_rxtx[th_id].frame_tx = proc->frame_tx; - - printf("[UE %d] frame %d, nr_tti_rx %d: Adjusting frame counter (PBCH ant_tx=%d, frame_tx=%d, phase %d, rx_offset %d) => new frame %d\n", - ue->Mod_id, - ue->proc.proc_rxtx[th_id].frame_rx, - nr_tti_rx, - pbch_tx_ant, - frame_tx, - pbch_phase, - ue->rx_offset, - proc->frame_rx); - } - - - frame_rx = proc->frame_rx; - - } else if (((frame_tx & 0x03FF) != (proc->frame_rx & 0x03FF))) { - //(pbch_tx_ant != ue->frame_parms.nb_antennas_tx)) { - LOG_D(PHY,"[UE %d] frame %d, nr_tti_rx %d: Re-adjusting frame counter (PBCH ant_tx=%d, frame_rx=%d, frame%%1024=%d, phase %d).\n", - ue->Mod_id, - proc->frame_rx, - nr_tti_rx, - pbch_tx_ant, - frame_tx, - frame_rx & 0x03FF, - pbch_phase); - - proc->frame_rx = (proc->frame_rx & 0xFFFFFC00) | (frame_tx & 0x000003FF); - proc->frame_tx = proc->frame_rx; - frame_rx = proc->frame_rx; - - for(int th_id=0; th_idproc.proc_rxtx[th_id].frame_rx = (proc->frame_rx & 0xFFFFFC00) | (frame_tx & 0x000003FF); - ue->proc.proc_rxtx[th_id].frame_tx = proc->frame_rx; - } - - } #ifdef DEBUG_PHY_PROC - LOG_D(PHY,"[UE %d] frame %d, nr_tti_rx %d, Received PBCH (MIB): mode1_flag %d, tx_ant %d, frame_tx %d. N_RB_DL %d, phich_duration %d, phich_resource %d/6!\n", - ue->Mod_id, - frame_rx, - nr_tti_rx, - ue->frame_parms.mode1_flag, - pbch_tx_ant, - frame_tx, - ue->frame_parms.N_RB_DL, - ue->frame_parms.phich_config_common.phich_duration, - ue->frame_parms.phich_config_common.phich_resource); + LOG_D(PHY,"[UE %d] frame %d, nr_tti_rx %d, Received PBCH (MIB): frame_tx %d. N_RB_DL %d\n", + ue->Mod_id, + frame_rx, + nr_tti_rx, + frame_tx, + ue->frame_parms.N_RB_DL); #endif } else { + LOG_E(PHY,"[UE %d] frame %d, nr_tti_rx %d, Error decoding PBCH!\n", + ue->Mod_id,frame_rx, nr_tti_rx); + /* - LOG_E(PHY,"[UE %d] frame %d, nr_tti_rx %d, Error decoding PBCH!\n", - ue->Mod_id,frame_rx, nr_tti_rx); + write_output("rxsig0.m","rxs0", ue->common_vars.rxdata[0],ue->frame_parms.samples_per_subframe,1,1); - LOG_I(PHY,"[UE %d] rx_offset %d\n",ue->Mod_id,ue->rx_offset); - - - write_output("rxsig0.m","rxs0", ue->common_vars.rxdata[0],ue->frame_parms.samples_per_subframe,1,1); write_output("H00.m","h00",&(ue->common_vars.dl_ch_estimates[0][0][0]),((ue->frame_parms.Ncp==0)?7:6)*(ue->frame_parms.ofdm_symbol_size),1,1); write_output("H10.m","h10",&(ue->common_vars.dl_ch_estimates[0][2][0]),((ue->frame_parms.Ncp==0)?7:6)*(ue->frame_parms.ofdm_symbol_size),1,1); @@ -3448,14 +3012,10 @@ void ue_pbch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *pro ue->pbch_vars[eNB_id]->pdu_errors_conseq++; ue->pbch_vars[eNB_id]->pdu_errors++; - if (ue->mac_enabled == 1) { - //mac_xface->out_of_sync_ind(ue->Mod_id,frame_rx,eNB_id); - } - else{ - if (ue->pbch_vars[eNB_id]->pdu_errors_conseq>=100) { - LOG_E(PHY,"More that 100 consecutive PBCH errors! Exiting!\n"); - //mac_xface->macphy_exit("More that 100 consecutive PBCH errors!"); - } + + if (ue->pbch_vars[eNB_id]->pdu_errors_conseq>=100) { + LOG_E(PHY,"More that 100 consecutive PBCH errors! Exiting!\n"); + exit_fun("More that 100 consecutive PBCH errors! Exiting!\n"); } } @@ -3474,7 +3034,6 @@ void ue_pbch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *pro } -#endif //(0) unsigned int get_tx_amp(int power_dBm, int power_max_dBm, int N_RB_UL, int nb_rb) { @@ -3495,7 +3054,7 @@ unsigned int get_tx_amp(int power_dBm, int power_max_dBm, int N_RB_UL, int nb_rb #ifdef NR_PDCCH_SCHED -int nr_ue_pdcch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t abstraction_flag) +int nr_ue_pdcch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc) { // unsigned int dci_cnt=0, i; //removed for nr_ue_pdcch_procedures and added in the loop for nb_coreset_active @@ -3513,47 +3072,41 @@ int nr_ue_pdcch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t * // each element represents the size in bits for each dci field, for each decoded dci -> [dci_cnt-1] // each time a dci is decode at dci_cnt, the values of the table dci_fields_sizes[i][j] will be copied at table dci_fields_sizes_cnt[dci_cnt-1][i][j] // table dci_fields_sizes_cnt[dci_cnt-1][i][j] will then be used in function nr_extract_dci_info - uint8_t dci_fields_sizes_cnt[MAX_NR_DCI_DECODED_SLOT][NBR_NR_DCI_FIELDS][NBR_NR_FORMATS] = {0}; + uint8_t dci_fields_sizes_cnt[MAX_NR_DCI_DECODED_SLOT][NBR_NR_DCI_FIELDS][NBR_NR_FORMATS]; int nb_searchspace_active=0; NR_UE_PDCCH **pdcch_vars = ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]]; NR_UE_PDCCH *pdcch_vars2 = ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]; - - uint16_t c_rnti=pdcch_vars[eNB_id]->crnti; - uint16_t cs_rnti=1,new_rnti=1,tc_rnti=1; - uint16_t p_rnti=P_RNTI; - uint16_t si_rnti=SI_RNTI; - uint16_t ra_rnti=3; - uint16_t sp_csi_rnti=1,sfi_rnti=1,int_rnti=1,tpc_pusch_rnti=1,tpc_pucch_rnti=1,tpc_srs_rnti=1; //FIXME - uint16_t crc_scrambled_values[TOTAL_NBR_SCRAMBLED_VALUES] = - {c_rnti,cs_rnti,new_rnti,tc_rnti,p_rnti,si_rnti,ra_rnti,sp_csi_rnti,sfi_rnti,int_rnti,tpc_pusch_rnti,tpc_pucch_rnti,tpc_srs_rnti}; - // s in TS 38.212 Subclause 10.1, for each active BWP the UE can deal with 10 different search spaces // Higher layers have updated the number of searchSpaces with are active in the current slot and this value is stored in variable nb_searchspace_total int nb_searchspace_total = pdcch_vars2->nb_search_space; -#ifdef NR_PDCCH_SCHED_DEBUG + + uint16_t c_rnti=pdcch_vars[eNB_id]->crnti; + uint16_t cs_rnti,new_rnti,tc_rnti; + uint16_t p_rnti=P_RNTI; + uint16_t si_rnti=SI_RNTI; + uint16_t ra_rnti=99; + uint16_t sp_csi_rnti,sfi_rnti,int_rnti,tpc_pusch_rnti,tpc_pucch_rnti,tpc_srs_rnti; //FIXME + uint16_t crc_scrambled_values[TOTAL_NBR_SCRAMBLED_VALUES] = + {c_rnti,cs_rnti,new_rnti,tc_rnti,p_rnti,si_rnti,ra_rnti,sp_csi_rnti,sfi_rnti,int_rnti,tpc_pusch_rnti,tpc_pucch_rnti,tpc_srs_rnti}; + #ifdef NR_PDCCH_SCHED_DEBUG printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> there is a bug in FAPI to calculate nb_searchspace_total=%d\n",nb_searchspace_total); -#endif + #endif if (nb_searchspace_total>1) nb_searchspace_total=1; // to be removed when fixing bug in FAPI -#ifdef NR_PDCCH_SCHED_DEBUG + #ifdef NR_PDCCH_SCHED_DEBUG printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> there is a bug in FAPI to calculate nb_searchspace_total so we set it to 1...\n"); printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> the number of searchSpaces active in the current slot(%d) is %d) \n", nr_tti_rx,nb_searchspace_total); -#endif + #endif + // p in TS 38.212 Subclause 10.1, for each active BWP the UE can deal with 3 different CORESETs (including coresetId 0 for common search space) - //int nb_coreset_total = NR_NBR_CORESET_ACT_BWP; - uint8_t dci_cnt=0; + int nb_coreset_total = NR_NBR_CORESET_ACT_BWP; + unsigned int dci_cnt=0; // this table contains 56 (NBR_NR_DCI_FIELDS) elements for each dci field and format described in TS 38.212. Each element represents the size in bits for each dci field - //uint8_t dci_fields_sizes[NBR_NR_DCI_FIELDS][NBR_NR_FORMATS] = {0}; + uint8_t dci_fields_sizes[NBR_NR_DCI_FIELDS][NBR_NR_FORMATS] = {{0}}; // this is the UL bandwidth part. FIXME! To be defined where this value comes from // uint16_t n_RB_ULBWP = 106; // this is the DL bandwidth part. FIXME! To be defined where this value comes from - //uint16_t n_RB_DLBWP = 106; - //#ifdef NR_PDCCH_SCHED_DEBUG - // printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> n_RB_ULBWP=%d n_RB_DLBWP=%d\n", - // n_RB_ULBWP, - // n_RB_DLBWP); - // #endif // First we have to identify each searchSpace active at a time and do PDCCH monitoring corresponding to current searchSpace // Up to 10 searchSpaces can be configured to UE (s<=10) @@ -3577,60 +3130,44 @@ int nr_ue_pdcch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t * * To be implemented LATER !!! */ //int _offset,_index,_M; - //int searchSpace_id = pdcch_vars2->searchSpace[nb_searchspace_active].searchSpaceId; - /* - * The following code has been removed as it is handled by higher layers (fapi) - if (searchSpace_id == 0){ // Implementing TS 38.213 subclause 13, UE procedure for monitoring Type0-PDCCH common search space - * - * according to TS 38.213 subclause 13 - * For the SS/PBCH block and control resource set (CORESET) multiplexing pattern 1, - * a UE monitors PDCCH in the Type0-PDCCH common search space over two consecutive slots starting from slot n0 - * - if (frame_rx%2 == 0) { - if ((((_offset*2+((_index*_M)/20))%2) != 0) || ((((_offset*2)+(_index*_M))%20) != nr_tti_rx) || ((((_offset*2)+(_index*_M))%20) != nr_tti_rx-1)){ - do_pdcch_monitoring_current_slot = 0; - } - } - if (frame_rx%2 == 1) { - if ((((_offset*2+((_index*_M)/20))%2) != 1) || ((((_offset*2)+(_index*_M))%20) != nr_tti_rx) || ((((_offset*2)+(_index*_M))%20) != nr_tti_rx-1)){ - do_pdcch_monitoring_current_slot = 0; - } - } - }*/ -#ifdef NR_PDCCH_SCHED_DEBUG - printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> nb_searchspace_active=%d do_pdcch_monitoring_current_slot=%d (to be removed)\n", - nb_searchspace_active, - do_pdcch_monitoring_current_slot); -#endif + int searchSpace_id = pdcch_vars2->searchSpace[nb_searchspace_active].searchSpaceId; - // if (do_pdcch_monitoring_current_slot) { - // the searchSpace indicates that we need to monitor PDCCH in current nr_tti_rx - // get the parameters describing the current SEARCHSPACE - // the CORESET id applicable to the current SearchSpace - //int searchSpace_coreset_id = pdcch_vars2->searchSpace[nb_searchspace_active].controlResourceSetId; - // FIXME this variable is a bit string (14 bits) identifying every OFDM symbol in a slot. - // at the moment we will not take into consideration this variable and we will consider that the OFDM symbol offset is always the first OFDM in a symbol - uint16_t symbol_within_slot_mon = pdcch_vars2->searchSpace[nb_searchspace_active].monitoringSymbolWithinSlot; - // get the remaining parameters describing the current SEARCHSPACE: // FIXME! To be defined where we get this information from - //NR_UE_SEARCHSPACE_nbrCAND_t num_cand_L1 = pdcch_vars2->searchSpace[nb_searchspace_active].nrofCandidates_aggrlevel1; - //NR_UE_SEARCHSPACE_nbrCAND_t num_cand_L2 = pdcch_vars2->searchSpace[nb_searchspace_active].nrofCandidates_aggrlevel2; - //NR_UE_SEARCHSPACE_nbrCAND_t num_cand_L4 = pdcch_vars2->searchSpace[nb_searchspace_active].nrofCandidates_aggrlevel4; - //NR_UE_SEARCHSPACE_nbrCAND_t num_cand_L8 = pdcch_vars2->searchSpace[nb_searchspace_active].nrofCandidates_aggrlevel8; - //NR_UE_SEARCHSPACE_nbrCAND_t num_cand_L16 = pdcch_vars2->searchSpace[nb_searchspace_active].nrofCandidates_aggrlevel16; - // FIXME! A table of five enum elements - // searchSpaceType indicates whether this is a common search space or a UE-specific search space - //int searchSpaceType = pdcch_vars2->searchSpace[nb_searchspace_active].searchSpaceType.type; - NR_SEARCHSPACE_TYPE_t searchSpaceType = common; -#ifdef NR_PDCCH_SCHED_DEBUG - printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> searchSpaceType=%d is hardcoded THIS HAS TO BE FIXED!!!\n", - searchSpaceType); -#endif - /*while ((searchSpace_coreset_id != pdcch_vars2->coreset[nb_coreset_active].controlResourceSetId) && (nb_coreset_active= nb_coreset_total) return 0; // the coreset_id could not be found. There is a problem - }*/ + #ifdef NR_PDCCH_SCHED_DEBUG + printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> nb_searchspace_active=%d do_pdcch_monitoring_current_slot=%d (to be removed)\n", + nb_searchspace_active, + do_pdcch_monitoring_current_slot); + #endif + +// if (do_pdcch_monitoring_current_slot) { + // the searchSpace indicates that we need to monitor PDCCH in current nr_tti_rx + // get the parameters describing the current SEARCHSPACE + // the CORESET id applicable to the current SearchSpace + int searchSpace_coreset_id = pdcch_vars2->searchSpace[nb_searchspace_active].controlResourceSetId; + // FIXME this variable is a bit string (14 bits) identifying every OFDM symbol in a slot. + // at the moment we will not take into consideration this variable and we will consider that the OFDM symbol offset is always the first OFDM in a symbol + uint16_t symbol_within_slot_mon = pdcch_vars2->searchSpace[nb_searchspace_active].monitoringSymbolWithinSlot; + // get the remaining parameters describing the current SEARCHSPACE: // FIXME! To be defined where we get this information from + NR_UE_SEARCHSPACE_nbrCAND_t num_cand_L1 = pdcch_vars2->searchSpace[nb_searchspace_active].nrofCandidates_aggrlevel1; + NR_UE_SEARCHSPACE_nbrCAND_t num_cand_L2 = pdcch_vars2->searchSpace[nb_searchspace_active].nrofCandidates_aggrlevel2; + NR_UE_SEARCHSPACE_nbrCAND_t num_cand_L4 = pdcch_vars2->searchSpace[nb_searchspace_active].nrofCandidates_aggrlevel4; + NR_UE_SEARCHSPACE_nbrCAND_t num_cand_L8 = pdcch_vars2->searchSpace[nb_searchspace_active].nrofCandidates_aggrlevel8; + NR_UE_SEARCHSPACE_nbrCAND_t num_cand_L16 = pdcch_vars2->searchSpace[nb_searchspace_active].nrofCandidates_aggrlevel16; + // FIXME! A table of five enum elements + // searchSpaceType indicates whether this is a common search space or a UE-specific search space + //int searchSpaceType = pdcch_vars2->searchSpace[nb_searchspace_active].searchSpaceType.type; + NR_SEARCHSPACE_TYPE_t searchSpaceType = common; + #ifdef NR_PDCCH_SCHED_DEBUG + printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> searchSpaceType=%d is hardcoded THIS HAS TO BE FIXED!!!\n", + searchSpaceType); + #endif + + /*while ((searchSpace_coreset_id != pdcch_vars2->coreset[nb_coreset_active].controlResourceSetId) && (nb_coreset_active= nb_coreset_total) return 0; // the coreset_id could not be found. There is a problem + }*/ + /* * we do not need these parameters yet @@ -3663,173 +3200,73 @@ int nr_ue_pdcch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t * // or TPC-PUSCH-RNTI, or TPC-PUCCH-RNTI, or TPC-SRS-RNTI, or C-RNTI, or CS-RNTI(s), or SP-CSI-RNTI -#ifdef PHY_ABSTRACTION - int CC_id; - int UE_id; - uint8_t harq_pid; -#endif VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_PDCCH_PROCEDURES, VCD_FUNCTION_IN); #if UE_TIMING_TRACE - start_meas(&ue->dlsch_rx_pdcch_stats); + start_meas(&ue->dlsch_rx_pdcch_stats); #endif - // if (nr_tti_rx != 5) - // return 0; - if (abstraction_flag == 0) { VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_RX_PDCCH, VCD_FUNCTION_IN); - /* rx_pdcch(ue, - proc->frame_rx, - nr_tti_rx, - eNB_id, - (ue->frame_parms.mode1_flag == 1) ? SISO : ALAMOUTI, - ue->high_speed_flag, - ue->is_secondary_ue);*/ //removed for nr_ue_pdcch_procedures #ifdef NR_PDCCH_SCHED_DEBUG printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> Entering function nr_rx_pdcch with eNB_id=%d (nb_coreset_active=%d, (symbol_within_slot_mon&0x3FFF)=%d, searchSpaceType=%d)\n", - eNB_id,nb_coreset_active,(symbol_within_slot_mon&0x3FFF), - searchSpaceType); + eNB_id,nb_coreset_active,(symbol_within_slot_mon&0x3FFF), + searchSpaceType); #endif - nr_rx_pdcch(ue, - proc->frame_rx, - nr_tti_rx, - eNB_id, - //(ue->frame_parms.mode1_flag == 1) ? SISO : ALAMOUTI, - SISO, - ue->high_speed_flag, - ue->is_secondary_ue, - nb_coreset_active, - (symbol_within_slot_mon&0x3FFF), - searchSpaceType); + nr_rx_pdcch(ue, + proc->frame_rx, + nr_tti_rx, + eNB_id, + //(ue->frame_parms.mode1_flag == 1) ? SISO : ALAMOUTI, + SISO, + ue->high_speed_flag, + ue->is_secondary_ue, + nb_coreset_active, + (symbol_within_slot_mon&0x3FFF), + searchSpaceType); #ifdef NR_PDCCH_SCHED_DEBUG - printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> Ending function nr_rx_pdcch(nb_coreset_active=%d, (symbol_within_slot_mon&0x3FFF)=%d, searchSpaceType=%d)\n", - nb_coreset_active,(symbol_within_slot_mon&0x3FFF), - searchSpaceType); + printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> Ending function nr_rx_pdcch(nb_coreset_active=%d, (symbol_within_slot_mon&0x3FFF)=%d, searchSpaceType=%d)\n", + nb_coreset_active,(symbol_within_slot_mon&0x3FFF), + searchSpaceType); #endif - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_RX_PDCCH, VCD_FUNCTION_OUT); - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_DCI_DECODING, VCD_FUNCTION_IN); + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_RX_PDCCH, VCD_FUNCTION_OUT); + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_DCI_DECODING, VCD_FUNCTION_IN); - /* - //printf("Decode SIB frame param agregation + DCI %d %d \n",agregationLevel,dciFormat); - - //agregation level == FF means no configuration on - if(ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->agregationLevel == 0xFF || ue->decode_SIB) - { - // search all possible dcis - dci_cnt = dci_decoding_procedure(ue, - dci_alloc_rx, - (ue->UE_mode[eNB_id] < PUSCH)? 1 : 0, // if we're in PUSCH don't listen to common search space, - // later when we need paging or RA during connection, update this ... - eNB_id,nr_tti_rx); - } - else - { - // search only preconfigured dcis - // search C RNTI dci - dci_cnt = dci_CRNTI_decoding_procedure(ue, - dci_alloc_rx, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->dciFormat, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->agregationLevel, - eNB_id, - nr_tti_rx); - } - */ //removed for nr_ue_pdcch_procedures - crc_scrambled_t crc_scrambled; - format_found_t format_found=255; + crc_scrambled_t crc_scrambled; + format_found_t format_found=255; + #ifdef NR_PDCCH_SCHED_DEBUG - if (searchSpaceType == common) { // search all possible dci's for COMMON SEARCH SPACES according to the current SEARCHSPACE configuration - printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> Entering function nr_dci_decoding_procedure with eNB_id=%d (n_RB_BWP=%d, searchSpaceType=%d, nb_searchspace_active=%d, nb_coreset_active=%d) -> dci_cnt=%d\n", - eNB_id,pdcch_vars2->n_RB_BWP[nb_searchspace_active],searchSpaceType, + printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> Entering function nr_dci_decoding_procedure with eNB_id=%d (n_RB_ULBWP=%d, n_RB_DLBWP=%d, searchSpaceType=%d, nb_searchspace_active=%d, nb_coreset_active=%d) -> dci_cnt=%d\n", + eNB_id,n_RB_ULBWP,n_RB_DLBWP,searchSpaceType, nb_searchspace_active, nb_coreset_active, dci_cnt); - } - if (searchSpaceType == ue_specific){// search all possible dci's for UE-SPECIFIC SEARCH SPACES according to the current SEARCHSPACE configuration - printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> Entering function nr_dci_decoding_procedure with eNB_id=%d (n_RB_BWP=%d, searchSpaceType=%d, nb_searchspace_active=%d, nb_coreset_active=%d) -> dci_cnt=%d\n", - eNB_id,pdcch_vars2->n_RB_BWP[nb_searchspace_active],searchSpaceType, - nb_searchspace_active, - nb_coreset_active, - dci_cnt); - } + #endif + - - dci_cnt += nr_dci_decoding_procedure(nb_searchspace_active, - nb_coreset_active, - ue, - &dci_alloc_rx[dci_cnt], - searchSpaceType, - eNB_id, - nr_tti_rx, - dci_fields_sizes_cnt, - pdcch_vars2->n_RB_BWP[nb_searchspace_active], - pdcch_vars2->n_RB_BWP[nb_searchspace_active], - &crc_scrambled, - &format_found, - crc_scrambled_values); + dci_cnt += nr_dci_decoding_procedure(nb_searchspace_active, + nb_coreset_active, + ue, + &dci_alloc_rx[dci_cnt], + searchSpaceType, // if we're in PUSCH don't listen to common search space, + // later when we need paging or RA during connection, update this ... + eNB_id, + nr_tti_rx, + dci_fields_sizes_cnt, + pdcch_vars2->n_RB_BWP[nb_searchspace_active], + pdcch_vars2->n_RB_BWP[nb_searchspace_active], + &crc_scrambled, + &format_found, + crc_scrambled_values); #ifdef NR_PDCCH_SCHED_DEBUG - printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> Ending function nr_dci_decoding_procedure() -> dci_cnt=%d\n",dci_cnt); + printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> Ending function nr_dci_decoding_procedure() -> dci_cnt=%d\n",dci_cnt); #endif - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_DCI_DECODING, VCD_FUNCTION_OUT); - //LOG_D(PHY,"[UE %d][PUSCH] Frame %d nr_tti_rx %d PHICH RX\n",ue->Mod_id,frame_rx,nr_tti_rx); + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_DCI_DECODING, VCD_FUNCTION_OUT); + //LOG_D(PHY,"[UE %d][PUSCH] Frame %d nr_tti_rx %d PHICH RX\n",ue->Mod_id,frame_rx,nr_tti_rx); - /* if (is_phich_subframe(&ue->frame_parms,nr_tti_rx)) { - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_RX_PHICH, VCD_FUNCTION_IN); - rx_phich(ue,proc,nr_tti_rx,eNB_id); - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_RX_PHICH, VCD_FUNCTION_OUT); - }*/ //removed for nr_ue_pdcch_procedures - } - -#ifdef PHY_ABSTRACTION - else { - for (i=0; iframe_parms.Nid_cell == ue->frame_parms.Nid_cell) - break; - - if (CC_id < MAX_NUM_CCs) - break; - } - - if (i==NB_eNB_INST) { - LOG_E(PHY,"[UE %d] phy_procedures_lte_ue.c: FATAL : Could not find attached eNB for DCI emulation (Nid_cell %d)!!!!\n",ue->Mod_id,ue->frame_parms.Nid_cell); - mac_xface->macphy_exit("Could not find attached eNB for DCI emulation"); - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_PDCCH_PROCEDURES, VCD_FUNCTION_OUT); - return(-1); - } - - LOG_D(PHY,"Calling dci_decoding_proc_emul ...\n"); - dci_cnt = dci_decoding_procedure_emul(ue->pdcch_vars[nr_tti_rx&1], - PHY_vars_eNB_g[i][CC_id]->num_ue_spec_dci[nr_tti_rx&1], - PHY_vars_eNB_g[i][CC_id]->num_common_dci[nr_tti_rx&1], - PHY_vars_eNB_g[i][CC_id]->dci_alloc[nr_tti_rx&1], - dci_alloc_rx, - eNB_id); - // printf("DCI: dci_cnt %d\n",dci_cnt); - UE_id = (uint32_t)find_ue((int16_t)ue->pdcch_vars[nr_tti_rx&1][eNB_id]->crnti,PHY_vars_eNB_g[i][CC_id]); - - if (UE_id>=0) { - // printf("Checking PHICH for UE %d (eNB %d)\n",UE_id,i); - if (is_phich_subframe(&ue->frame_parms,nr_tti_rx)) { - harq_pid = phich_subframe_to_harq_pid(&ue->frame_parms,frame_rx,nr_tti_rx); - - if (ue->ulsch[eNB_id]->harq_processes[harq_pid]->status == ACTIVE) { - // ue->ulsch[eNB_id]->harq_processes[harq_pid]->phich_ACK=1; - ue->ulsch[eNB_id]->harq_processes[harq_pid]->subframe_scheduling_flag =0; - ue->ulsch[eNB_id]->harq_processes[harq_pid]->status = IDLE; - ue->ulsch_Msg3_active[eNB_id] = 0; - ue->ulsch[eNB_id]->harq_processes[harq_pid]->round = 0; - LOG_D(PHY,"Msg3 inactive\n"); - - } // harq_pid is ACTIVE - } // This is a PHICH nr_tti_rx - } // UE_id exists - } - -#endif uint8_t *nCCE_current = &ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->nCCE[nr_tti_rx]; uint8_t *nCCE_dest = &ue->pdcch_vars[next1_thread_id][eNB_id]->nCCE[nr_tti_rx]; @@ -3864,217 +3301,76 @@ int nr_ue_pdcch_procedures(uint8_t eNB_id,PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t * (dci_alloc_rx[i].rnti != crc_scrambled_values[_P_RNTI_]) && (dci_alloc_rx[i].rnti != crc_scrambled_values[_SI_RNTI_]) && (dci_alloc_rx[i].rnti != crc_scrambled_values[_RA_RNTI_])) ue->dlsch_received[eNB_id]++; - if ((dci_alloc_rx[i].rnti == crc_scrambled_values[_SI_RNTI_]) && (dci_alloc_rx[i].format == format1_0)) ue->dlsch_SI_received[eNB_id]++; - if ((dci_alloc_rx[i].rnti == crc_scrambled_values[_P_RNTI_]) && (dci_alloc_rx[i].format == format1_0)) ue->dlsch_p_received[eNB_id]++; - if ((dci_alloc_rx[i].rnti == crc_scrambled_values[_RA_RNTI_]) && (dci_alloc_rx[i].format == format1_0)) ue->dlsch_ra_received[eNB_id]++; - if ((dci_alloc_rx[i].format == format2_0)){ - } - if ((dci_alloc_rx[i].format == format2_1)){ - } - if ((dci_alloc_rx[i].format == format2_2)){ - } - if ((dci_alloc_rx[i].format == format2_3)){ - } - if ((dci_alloc_rx[i].format == format0_1)){ // This format not implemented at a first time. FIXME - } - if ((dci_alloc_rx[i].format == format1_1)){ // This format not implemented at a first time. FIXME - } - NR_DCI_INFO_EXTRACTED_t nr_dci_info_extracted; - NR_DCI_INFO_EXTRACTED_t *ptr_nr_dci_info_extracted = &nr_dci_info_extracted; -#ifdef NR_PDCCH_SCHED_DEBUG - printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> Entering function nr_generate_ue_ul_dlsch_params_from_dci with eNB_id=%d dci_cnt=%d format=%d, rnti=%d\n", - eNB_id, - dci_cnt, - dci_alloc_rx[i].format, - dci_alloc_rx[i].rnti); -#endif - - nr_generate_ue_ul_dlsch_params_from_dci(ue, - eNB_id, - frame_rx, - nr_tti_rx, - dci_alloc_rx[i].dci_pdu, - dci_alloc_rx[i].rnti, - dci_alloc_rx[i].dci_length, - dci_alloc_rx[i].format, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id], - ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id], - ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id], - ue->ulsch[eNB_id], - &ue->frame_parms, - ue->pdsch_config_dedicated, - ue->transmission_mode[eNB_id]<7?0:ue->transmission_mode[eNB_id], - dci_fields_sizes_cnt[i], - pdcch_vars2->n_RB_BWP[nb_searchspace_active], - pdcch_vars2->n_RB_BWP[nb_searchspace_active], - crc_scrambled_values, - ptr_nr_dci_info_extracted);//&nr_dci_info_extracted); - - ue->dci_ind.dci_list[i].rnti = dci_alloc_rx[i].rnti; - ue->dci_ind.dci_list[i].dci_format = dci_alloc_rx[i].format; - ue->dci_ind.dci_list[i].n_CCE = dci_alloc_rx[i].firstCCE; - ue->dci_ind.dci_list[i].N_CCE = (int)dci_alloc_rx[i].L; - memcpy(&ue->dci_ind.dci_list[i].dci, &nr_dci_info_extracted, sizeof(fapi_nr_dci_pdu_rel15_t) ); - - //printf(">>> example mcs=%d\n",nr_dci_info_extracted.mcs); - //printf(">>> calling MAC with dl_indication DCI_IND\n"); - // TODO: check where should we send up this message. - ue->dl_indication.dci_ind=&ue->dci_ind; - ue->if_inst->dl_indication(&ue->dl_indication); + + if ((dci_alloc_rx[i].rnti == crc_scrambled_values[_SI_RNTI_]) && (dci_alloc_rx[i].format == format1_0)) ue->dlsch_SI_received[eNB_id]++; + if ((dci_alloc_rx[i].rnti == crc_scrambled_values[_P_RNTI_]) && (dci_alloc_rx[i].format == format1_0)) ue->dlsch_p_received[eNB_id]++; + if ((dci_alloc_rx[i].rnti == crc_scrambled_values[_RA_RNTI_]) && (dci_alloc_rx[i].format == format1_0)) ue->dlsch_ra_received[eNB_id]++; + if ((dci_alloc_rx[i].format == format2_0)){ + } + if ((dci_alloc_rx[i].format == format2_1)){ + } + if ((dci_alloc_rx[i].format == format2_2)){ + } + if ((dci_alloc_rx[i].format == format2_3)){ + } + if ((dci_alloc_rx[i].format == format0_1)){ // This format not implemented at a first time. FIXME + } + if ((dci_alloc_rx[i].format == format1_1)){ // This format not implemented at a first time. FIXME + } + NR_DCI_INFO_EXTRACTED_t nr_dci_info_extracted; + #ifdef NR_PDCCH_SCHED_DEBUG + printf("<-NR_PDCCH_PHY_PROCEDURES_LTE_UE (nr_ue_pdcch_procedures)-> Entering function nr_generate_ue_ul_dlsch_params_from_dci with eNB_id=%d dci_cnt=%d\n", + eNB_id, + dci_cnt); + #endif + nr_generate_ue_ul_dlsch_params_from_dci(ue, + eNB_id, + frame_rx, + nr_tti_rx, + dci_alloc_rx[i].dci_pdu, + dci_alloc_rx[i].rnti, + dci_alloc_rx[i].dci_length, + dci_alloc_rx[i].format, + ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id], + ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id], + ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id], + ue->ulsch[eNB_id], + &ue->frame_parms, + ue->pdsch_config_dedicated, + ue->transmission_mode[eNB_id]<7?0:ue->transmission_mode[eNB_id], + dci_fields_sizes_cnt[i], + pdcch_vars2->n_RB_BWP[nb_searchspace_active], + pdcch_vars2->n_RB_BWP[nb_searchspace_active], + crc_scrambled_values, + &nr_dci_info_extracted); + + ue->dci_ind.dci_list[i].rnti = dci_alloc_rx[i].rnti; + ue->dci_ind.dci_list[i].dci_format = dci_alloc_rx[i].format; + ue->dci_ind.dci_list[i].n_CCE = dci_alloc_rx[i].firstCCE; + ue->dci_ind.dci_list[i].N_CCE = (int)dci_alloc_rx[i].L; + ue->dci_ind.number_of_dcis = ue->dci_ind.number_of_dcis + 1; + memcpy(&ue->dci_ind.dci_list[i].dci, &nr_dci_info_extracted, sizeof(fapi_nr_dci_pdu_rel15_t) ); - /* - * This is the LTE part to be removed + // TODO: check where should we send up this message. + //ue->if_inst->dl_indication(&ue->dl_indication); - if ((ue->UE_mode[eNB_id]>PRACH) && (dci_alloc_rx[i].rnti == ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->crnti) && (dci_alloc_rx[i].format != format0)) { + } // end for loop dci_cnt - LOG_D(PHY,"[UE %d][DCI][PDSCH %x] AbsSubframe %d.%d: format %d, num_pdcch_symbols %d, nCCE %d, total CCEs %d\n", - ue->Mod_id,dci_alloc_rx[i].rnti, - frame_rx%1024,nr_tti_rx, - dci_alloc_rx[i].format, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->nCCE[nr_tti_rx], - get_nCCE(3,&ue->frame_parms,get_mi(&ue->frame_parms,0))); - - //dump_dci(&ue->frame_parms, &dci_alloc_rx[i]); - - if ((ue->UE_mode[eNB_id] > PRACH) && - (generate_ue_dlsch_params_from_dci(frame_rx, - nr_tti_rx, - (void *)&dci_alloc_rx[i].dci_pdu, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->crnti, - dci_alloc_rx[i].format, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id], - ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id], - ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id], - &ue->frame_parms, - ue->pdsch_config_dedicated, - SI_RNTI, - 0, - P_RNTI, - ue->transmission_mode[eNB_id]<7?0:ue->transmission_mode[eNB_id], - ue->pdcch_vars[0%RX_NB_TH][eNB_id]->crnti_is_temporary? ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->crnti: 0)==0)) { - - // update TPC for PUCCH - if((dci_alloc_rx[i].format == format1) || - (dci_alloc_rx[i].format == format1A) || - (dci_alloc_rx[i].format == format1B) || - (dci_alloc_rx[i].format == format2) || - (dci_alloc_rx[i].format == format2A) || - (dci_alloc_rx[i].format == format2B)) - { - //ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->g_pucch += ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->harq_processes[ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->current_harq_pid]->delta_PUCCH; - int32_t delta_pucch = ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->harq_processes[ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->current_harq_pid]->delta_PUCCH; - for(int th_id=0; th_iddlsch[th_id][eNB_id][0]->g_pucch += delta_pucch; - } - LOG_D(PHY,"update TPC for PUCCH %d.%d / pid %d delta_PUCCH %d g_pucch %d %d \n",frame_rx, nr_tti_rx,ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->current_harq_pid, - delta_pucch, - ue->dlsch[0][eNB_id][0]->g_pucch, - ue->dlsch[1][eNB_id][0]->g_pucch - //ue->dlsch[2][eNB_id][0]->g_pucch - ); - } - - ue->dlsch_received[eNB_id]++; - - #ifdef DEBUG_PHY_PROC - LOG_D(PHY,"[UE %d] Generated UE DLSCH C_RNTI format %d\n",ue->Mod_id,dci_alloc_rx[i].format); - dump_dci(&ue->frame_parms, &dci_alloc_rx[i]); - LOG_D(PHY,"[UE %d] *********** dlsch->active in nr_tti_rx %d=> %d\n",ue->Mod_id,nr_tti_rx,ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active); - #endif - - // we received a CRNTI, so we're in PUSCH - if (ue->UE_mode[eNB_id] != PUSCH) { - #ifdef DEBUG_PHY_PROC - LOG_I(PHY,"[UE %d] Frame %d, nr_tti_rx %d: Received DCI with CRNTI %x => Mode PUSCH\n",ue->Mod_id,frame_rx,nr_tti_rx,ue->pdcch_vars[nr_tti_rx&1][eNB_id]->crnti); - #endif - //dump_dci(&ue->frame_parms, &dci_alloc_rx[i]); - ue->UE_mode[eNB_id] = PUSCH; - //mac_xface->macphy_exit("Connected. Exiting\n"); - } - } else { - LOG_E(PHY,"[UE %d] Frame %d, nr_tti_rx %d: Problem in DCI!\n",ue->Mod_id,frame_rx,nr_tti_rx); - dump_dci(&ue->frame_parms, &dci_alloc_rx[i]); - } - } - - else if ((dci_alloc_rx[i].rnti == SI_RNTI) && - ((dci_alloc_rx[i].format == format1A) || (dci_alloc_rx[i].format == format1C))) { - - #ifdef DEBUG_PHY_PROC - LOG_I(PHY,"[UE %d] nr_tti_rx %d: Found rnti %x, format 1%s, dci_cnt %d\n",ue->Mod_id,nr_tti_rx,dci_alloc_rx[i].rnti,dci_alloc_rx[i].format==format1A?"A":"C",i); - #endif +#if UE_TIMING_TRACE + stop_meas(&ue->dlsch_rx_pdcch_stats); +#endif + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_PDCCH_PROCEDURES, VCD_FUNCTION_OUT); - if (generate_ue_dlsch_params_from_dci(frame_rx, - nr_tti_rx, - (void *)&dci_alloc_rx[i].dci_pdu, - SI_RNTI, - dci_alloc_rx[i].format, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id], - ue->pdsch_vars_SI[eNB_id], - &ue->dlsch_SI[eNB_id], - &ue->frame_parms, - ue->pdsch_config_dedicated, - SI_RNTI, - 0, - P_RNTI, - ue->transmission_mode[eNB_id]<7?0:ue->transmission_mode[eNB_id], - 0)==0) { - - ue->dlsch_SI_received[eNB_id]++; + } // end for loop nb_searchspace_active + return(0); +} +#endif // NR_PDCCH_SCHED - LOG_I(PHY,"[UE %d] Frame %d, nr_tti_rx %d : Generate UE DLSCH SI_RNTI format 1%s\n",ue->Mod_id,frame_rx,nr_tti_rx,dci_alloc_rx[i].format==format1A?"A":"C"); - //dump_dci(&ue->frame_parms, &dci_alloc_rx[i]); - - } - } - - else if ((dci_alloc_rx[i].rnti == P_RNTI) && - ((dci_alloc_rx[i].format == format1A) || (dci_alloc_rx[i].format == format1C))) { - - #ifdef DEBUG_PHY_PROC - LOG_I(PHY,"[UE %d] nr_tti_rx %d: Found rnti %x, format 1%s, dci_cnt %d\n",ue->Mod_id,nr_tti_rx,dci_alloc_rx[i].rnti,dci_alloc_rx[i].format==format1A?"A":"C",i); - #endif - if (generate_ue_dlsch_params_from_dci(frame_rx, - nr_tti_rx, - (void *)&dci_alloc_rx[i].dci_pdu, - P_RNTI, - dci_alloc_rx[i].format, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id], - ue->pdsch_vars_p[eNB_id], - &ue->dlsch_SI[eNB_id], - &ue->frame_parms, - ue->pdsch_config_dedicated, - SI_RNTI, - 0, - P_RNTI, - ue->transmission_mode[eNB_id]<7?0:ue->transmission_mode[eNB_id], - 0)==0) { - - ue->dlsch_p_received[eNB_id]++; - LOG_D(PHY,"[UE %d] Frame %d, nr_tti_rx %d : Generate UE DLSCH P_RNTI format 1%s\n",ue->Mod_id,frame_rx,nr_tti_rx,dci_alloc_rx[i].format==format1A?"A":"C"); - //dump_dci(&ue->frame_parms, &dci_alloc_rx[i]); - - } - } - - else if ((ue->prach_resources[eNB_id]) && - (dci_alloc_rx[i].rnti == ue->prach_resources[eNB_id]->ra_RNTI) && - (dci_alloc_rx[i].format == format1A)) { - - #ifdef DEBUG_PHY_PROC - LOG_D(PHY,"[UE %d][RAPROC] nr_tti_rx %d: Found RA rnti %x, format 1A, dci_cnt %d\n",ue->Mod_id,nr_tti_rx,dci_alloc_rx[i].rnti,i); - - //if (((frame_rx%100) == 0) || (frame_rx < 20)) - //dump_dci(&ue->frame_parms, &dci_alloc_rx[i]); - //mac_xface->macphy_exit("so far so good...\n"); - #endif - +#if 0 if (generate_ue_dlsch_params_from_dci(frame_rx, nr_tti_rx, @@ -4367,15 +3663,13 @@ void ue_pmch_procedures(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc,int eNB_id,i void copy_harq_proc_struct(NR_DL_UE_HARQ_t *harq_processes_dest, NR_DL_UE_HARQ_t *current_harq_processes) { - harq_processes_dest->B = current_harq_processes->B ; - harq_processes_dest->C = current_harq_processes->C ; - harq_processes_dest->Cminus = current_harq_processes->Cminus ; - harq_processes_dest->Cplus = current_harq_processes->Cplus ; - harq_processes_dest->DCINdi = current_harq_processes->DCINdi ; - harq_processes_dest->F = current_harq_processes->F ; - harq_processes_dest->G = current_harq_processes->G ; - harq_processes_dest->Kminus = current_harq_processes->Kminus ; - harq_processes_dest->Kplus = current_harq_processes->Kplus ; + + harq_processes_dest->B = current_harq_processes->B ; + harq_processes_dest->C = current_harq_processes->C ; + harq_processes_dest->DCINdi = current_harq_processes->DCINdi ; + harq_processes_dest->F = current_harq_processes->F ; + harq_processes_dest->G = current_harq_processes->G ; + harq_processes_dest->K = current_harq_processes->K ; harq_processes_dest->Nl = current_harq_processes->Nl ; harq_processes_dest->Qm = current_harq_processes->Qm ; harq_processes_dest->TBS = current_harq_processes->TBS ; @@ -4400,6 +3694,7 @@ void copy_harq_proc_struct(NR_DL_UE_HARQ_t *harq_processes_dest, NR_DL_UE_HARQ_t harq_processes_dest->rvidx = current_harq_processes->rvidx ; harq_processes_dest->status = current_harq_processes->status ; harq_processes_dest->vrb_type = current_harq_processes->vrb_type ; + } /*void copy_ack_struct(nr_harq_status_t *harq_ack_dest, nr_harq_status_t *current_harq_ack) @@ -4407,8 +3702,7 @@ void copy_harq_proc_struct(NR_DL_UE_HARQ_t *harq_processes_dest, NR_DL_UE_HARQ_t memcpy(harq_ack_dest, current_harq_ack, sizeof(nr_harq_status_t)); }*/ -#if 0 -void ue_pdsch_procedures(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, int eNB_id, PDSCH_t pdsch, NR_UE_DLSCH_t *dlsch0, NR_UE_DLSCH_t *dlsch1, int s0, int s1, int abstraction_flag) { +void nr_ue_pdsch_procedures(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, int eNB_id, PDSCH_t pdsch, NR_UE_DLSCH_t *dlsch0, NR_UE_DLSCH_t *dlsch1, int s0, int s1) { int nr_tti_rx = proc->nr_tti_rx; int m; @@ -4423,15 +3717,15 @@ void ue_pdsch_procedures(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, int eNB_id if (dlsch0 && (!dlsch1)) { harq_pid = dlsch0->current_harq_pid; - LOG_D(PHY,"[UE %d] PDSCH active in nr_tti_rx %d, harq_pid %d Symbol %d\n",ue->Mod_id,nr_tti_rx,harq_pid,m); + LOG_D(PHY,"[UE %d] PDSCH type %d active in nr_tti_rx %d, harq_pid %d Symbol %d\n",ue->Mod_id,pdsch,nr_tti_rx,harq_pid,m); if ((pdsch==PDSCH) && (ue->transmission_mode[eNB_id] == 5) && - (dlsch0->harq_processes[harq_pid].dl_power_off==0) && + (dlsch0->harq_processes[harq_pid]->dl_power_off==0) && (ue->use_ia_receiver ==1)) { dual_stream_UE = 1; eNB_id_i = ue->n_connected_eNB; - i_mod = dlsch0->harq_processes[harq_pid].Qm; + i_mod = dlsch0->harq_processes[harq_pid]->Qm; } else if((pdsch==PDSCH) && (ue->transmission_mode[eNB_id]==3)) @@ -4457,7 +3751,7 @@ void ue_pdsch_procedures(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, int eNB_id } } - if ((m==s0) && (m<4)) + if ((m==s0) && (m<3)) first_symbol_flag = 1; else first_symbol_flag = 0; @@ -4468,7 +3762,7 @@ void ue_pdsch_procedures(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, int eNB_id start_meas(&ue->dlsch_llr_stats_parallelization[ue->current_thread_id[nr_tti_rx]][slot]); #endif // process DLSCH received in first slot - rx_pdsch(ue, + nr_rx_pdsch(ue, pdsch, eNB_id, eNB_id_i, @@ -4496,8 +3790,6 @@ void ue_pdsch_procedures(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, int eNB_id } // CRNTI active } } -#endif - #if 0 void process_rar(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, int eNB_id, runmode_t mode, int abstraction_flag) { @@ -4590,16 +3882,16 @@ void process_rar(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, int eNB_id, runmod } } +#endif -void ue_dlsch_procedures(PHY_VARS_NR_UE *ue, - UE_nr_rxtx_proc_t *proc, - int eNB_id, - PDSCH_t pdsch, - NR_UE_DLSCH_t *dlsch0, - NR_UE_DLSCH_t *dlsch1, - int *dlsch_errors, - runmode_t mode, - int abstraction_flag) { +void nr_ue_dlsch_procedures(PHY_VARS_NR_UE *ue, + UE_nr_rxtx_proc_t *proc, + int eNB_id, + PDSCH_t pdsch, + NR_UE_DLSCH_t *dlsch0, + NR_UE_DLSCH_t *dlsch1, + int *dlsch_errors, + runmode_t mode) { int harq_pid; int frame_rx = proc->frame_rx; @@ -4609,6 +3901,10 @@ void ue_dlsch_procedures(PHY_VARS_NR_UE *ue, NR_UE_PDSCH *pdsch_vars; uint8_t is_cw0_active = 0; uint8_t is_cw1_active = 0; + // to be updated by higher layer + uint8_t nb_re_dmrs = 6; + uint16_t length_dmrs = 1; + uint16_t nb_symb_sch = 8; if (dlsch0==NULL) AssertFatal(0,"dlsch0 should be defined at this level \n"); @@ -4619,8 +3915,8 @@ void ue_dlsch_procedures(PHY_VARS_NR_UE *ue, if(dlsch1) is_cw1_active = dlsch1->harq_processes[harq_pid]->status; - LOG_D(PHY,"AbsSubframe %d.%d Start Turbo Decoder for CW0 [harq_pid %d] ? %d \n", frame_rx%1024, nr_tti_rx, harq_pid, is_cw0_active); - LOG_D(PHY,"AbsSubframe %d.%d Start Turbo Decoder for CW1 [harq_pid %d] ? %d \n", frame_rx%1024, nr_tti_rx, harq_pid, is_cw1_active); + LOG_D(PHY,"AbsSubframe %d.%d Start LDPC Decoder for CW0 [harq_pid %d] ? %d \n", frame_rx%1024, nr_tti_rx, harq_pid, is_cw0_active); + LOG_D(PHY,"AbsSubframe %d.%d Start LDPC Decoder for CW1 [harq_pid %d] ? %d \n", frame_rx%1024, nr_tti_rx, harq_pid, is_cw1_active); if(is_cw0_active && is_cw1_active) { @@ -4670,41 +3966,37 @@ void ue_dlsch_procedures(PHY_VARS_NR_UE *ue, } } - if (abstraction_flag == 0) { - // start turbo decode for CW 0 - dlsch0->harq_processes[harq_pid]->G = get_G(&ue->frame_parms, - dlsch0->harq_processes[harq_pid]->nb_rb, - dlsch0->harq_processes[harq_pid]->rb_alloc_even, - dlsch0->harq_processes[harq_pid]->Qm, - dlsch0->harq_processes[harq_pid]->Nl, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, - frame_rx, - nr_tti_rx, - ue->transmission_mode[eNB_id]<7?0:ue->transmission_mode[eNB_id]); + // start ldpc decode for CW 0 + dlsch0->harq_processes[harq_pid]->G = nr_get_G(dlsch0->harq_processes[harq_pid]->nb_rb, + nb_symb_sch, + nb_re_dmrs, + length_dmrs, + dlsch0->harq_processes[harq_pid]->Qm, + dlsch0->harq_processes[harq_pid]->Nl); #if UE_TIMING_TRACE start_meas(&ue->dlsch_unscrambling_stats); #endif - dlsch_unscrambling(&ue->frame_parms, - 0, - dlsch0, - dlsch0->harq_processes[harq_pid]->G, - pdsch_vars->llr[0], - 0, - nr_tti_rx<<1); + nr_dlsch_unscrambling(pdsch_vars->llr[0], + dlsch0->harq_processes[harq_pid]->G, + 0, + ue->frame_parms.Nid_cell, + dlsch0->rnti); + + #if UE_TIMING_TRACE stop_meas(&ue->dlsch_unscrambling_stats); #endif #if 0 - LOG_I(PHY," ------ start turbo decoder for AbsSubframe %d.%d / %d ------ \n", frame_rx, nr_tti_rx, harq_pid); - LOG_I(PHY,"start turbo decode for CW 0 for AbsSubframe %d.%d / %d --> nb_rb %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->harq_processes[harq_pid]->nb_rb); - LOG_I(PHY,"start turbo decode for CW 0 for AbsSubframe %d.%d / %d --> rb_alloc_even %x \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->harq_processes[harq_pid]->rb_alloc_even); - LOG_I(PHY,"start turbo decode for CW 0 for AbsSubframe %d.%d / %d --> Qm %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->harq_processes[harq_pid]->Qm); - LOG_I(PHY,"start turbo decode for CW 0 for AbsSubframe %d.%d / %d --> Nl %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->harq_processes[harq_pid]->Nl); - LOG_I(PHY,"start turbo decode for CW 0 for AbsSubframe %d.%d / %d --> G %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->harq_processes[harq_pid]->G); - LOG_I(PHY,"start turbo decode for CW 0 for AbsSubframe %d.%d / %d --> Kmimo %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->Kmimo); - LOG_I(PHY,"start turbo decode for CW 0 for AbsSubframe %d.%d / %d --> Pdcch Sym %d \n", frame_rx, nr_tti_rx, harq_pid, ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols); + LOG_I(PHY," ------ start ldpc decoder for AbsSubframe %d.%d / %d ------ \n", frame_rx, nr_tti_rx, harq_pid); + LOG_I(PHY,"start ldpc decode for CW 0 for AbsSubframe %d.%d / %d --> nb_rb %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->harq_processes[harq_pid]->nb_rb); + LOG_I(PHY,"start ldpc decode for CW 0 for AbsSubframe %d.%d / %d --> rb_alloc_even %x \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->harq_processes[harq_pid]->rb_alloc_even); + LOG_I(PHY,"start ldpc decode for CW 0 for AbsSubframe %d.%d / %d --> Qm %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->harq_processes[harq_pid]->Qm); + LOG_I(PHY,"start ldpc decode for CW 0 for AbsSubframe %d.%d / %d --> Nl %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->harq_processes[harq_pid]->Nl); + LOG_I(PHY,"start ldpc decode for CW 0 for AbsSubframe %d.%d / %d --> G %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->harq_processes[harq_pid]->G); + LOG_I(PHY,"start ldpc decode for CW 0 for AbsSubframe %d.%d / %d --> Kmimo %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch0->Kmimo); + LOG_I(PHY,"start ldpc decode for CW 0 for AbsSubframe %d.%d / %d --> Pdcch Sym %d \n", frame_rx, nr_tti_rx, harq_pid, ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols); #endif #if UE_TIMING_TRACE @@ -4712,23 +4004,25 @@ void ue_dlsch_procedures(PHY_VARS_NR_UE *ue, #endif #ifdef UE_DLSCH_PARALLELISATION - ret = dlsch_decoding_mthread(ue,proc,eNB_id, - pdsch_vars->llr[0], - &ue->frame_parms, - dlsch0, - dlsch0->harq_processes[harq_pid], - frame_rx, - nr_tti_rx, - harq_pid, - pdsch==PDSCH?1:0, - dlsch0->harq_processes[harq_pid]->TBS>256?1:0); -#else - ret = dlsch_decoding(ue, + ret = nr_dlsch_decoding_mthread(ue,proc,eNB_id, pdsch_vars->llr[0], &ue->frame_parms, dlsch0, dlsch0->harq_processes[harq_pid], frame_rx, + nb_symb_sch, + nr_tti_rx, + harq_pid, + pdsch==PDSCH?1:0, + dlsch0->harq_processes[harq_pid]->TBS>256?1:0); +#else + ret = nr_dlsch_decoding(ue, + pdsch_vars->llr[0], + &ue->frame_parms, + dlsch0, + dlsch0->harq_processes[harq_pid], + frame_rx, + nb_symb_sch, nr_tti_rx, harq_pid, pdsch==PDSCH?1:0,//proc->decoder_switch, @@ -4740,51 +4034,46 @@ void ue_dlsch_procedures(PHY_VARS_NR_UE *ue, stop_meas(&ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]]); #if DISABLE_LOG_X printf(" --> Unscrambling for CW0 %5.3f\n", - (ue->dlsch_unscrambling_stats.p_time)/(cpuf*1000.0)); - printf("AbsSubframe %d.%d --> Turbo Decoding for CW0 %5.3f\n", - frame_rx%1024, nr_tti_rx,(ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); + (ue->dlsch_unscrambling_stats.p_time)/(cpuf*1000.0)); + printf("AbsSubframe %d.%d --> LDPC Decoding for CW0 %5.3f\n", + frame_rx%1024, nr_tti_rx,(ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); #else LOG_I(PHY, " --> Unscrambling for CW0 %5.3f\n", - (ue->dlsch_unscrambling_stats.p_time)/(cpuf*1000.0)); - LOG_I(PHY, "AbsSubframe %d.%d --> Turbo Decoding for CW0 %5.3f\n", - frame_rx%1024, nr_tti_rx,(ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); + (ue->dlsch_unscrambling_stats.p_time)/(cpuf*1000.0)); + LOG_I(PHY, "AbsSubframe %d.%d --> LDPC Decoding for CW0 %5.3f\n", + frame_rx%1024, nr_tti_rx,(ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); #endif #endif if(is_cw1_active) - { - // start turbo decode for CW 1 - dlsch1->harq_processes[harq_pid]->G = get_G(&ue->frame_parms, - dlsch1->harq_processes[harq_pid]->nb_rb, - dlsch1->harq_processes[harq_pid]->rb_alloc_even, - dlsch1->harq_processes[harq_pid]->Qm, - dlsch1->harq_processes[harq_pid]->Nl, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, - frame_rx, - nr_tti_rx, - ue->transmission_mode[eNB_id]<7?0:ue->transmission_mode[eNB_id]); + { + // start ldpc decode for CW 1 + dlsch1->harq_processes[harq_pid]->G = nr_get_G(dlsch1->harq_processes[harq_pid]->nb_rb, + nb_symb_sch, + nb_re_dmrs, + length_dmrs, + dlsch1->harq_processes[harq_pid]->Qm, + dlsch1->harq_processes[harq_pid]->Nl); #if UE_TIMING_TRACE start_meas(&ue->dlsch_unscrambling_stats); #endif - dlsch_unscrambling(&ue->frame_parms, - 0, - dlsch1, - dlsch1->harq_processes[harq_pid]->G, - pdsch_vars->llr[1], - 1, - nr_tti_rx<<1); + nr_dlsch_unscrambling(pdsch_vars->llr[1], + dlsch1->harq_processes[harq_pid]->G, + 0, + ue->frame_parms.Nid_cell, + dlsch1->rnti); #if UE_TIMING_TRACE stop_meas(&ue->dlsch_unscrambling_stats); #endif #if 0 - LOG_I(PHY,"start turbo decode for CW 1 for AbsSubframe %d.%d / %d --> nb_rb %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->harq_processes[harq_pid]->nb_rb); - LOG_I(PHY,"start turbo decode for CW 1 for AbsSubframe %d.%d / %d --> rb_alloc_even %x \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->harq_processes[harq_pid]->rb_alloc_even); - LOG_I(PHY,"start turbo decode for CW 1 for AbsSubframe %d.%d / %d --> Qm %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->harq_processes[harq_pid]->Qm); - LOG_I(PHY,"start turbo decode for CW 1 for AbsSubframe %d.%d / %d --> Nl %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->harq_processes[harq_pid]->Nl); - LOG_I(PHY,"start turbo decode for CW 1 for AbsSubframe %d.%d / %d --> G %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->harq_processes[harq_pid]->G); - LOG_I(PHY,"start turbo decode for CW 1 for AbsSubframe %d.%d / %d --> Kmimo %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->Kmimo); - LOG_I(PHY,"start turbo decode for CW 1 for AbsSubframe %d.%d / %d --> Pdcch Sym %d \n", frame_rx, nr_tti_rx, harq_pid, ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols); + LOG_I(PHY,"start ldpc decode for CW 1 for AbsSubframe %d.%d / %d --> nb_rb %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->harq_processes[harq_pid]->nb_rb); + LOG_I(PHY,"start ldpc decode for CW 1 for AbsSubframe %d.%d / %d --> rb_alloc_even %x \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->harq_processes[harq_pid]->rb_alloc_even); + LOG_I(PHY,"start ldpc decode for CW 1 for AbsSubframe %d.%d / %d --> Qm %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->harq_processes[harq_pid]->Qm); + LOG_I(PHY,"start ldpc decode for CW 1 for AbsSubframe %d.%d / %d --> Nl %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->harq_processes[harq_pid]->Nl); + LOG_I(PHY,"start ldpc decode for CW 1 for AbsSubframe %d.%d / %d --> G %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->harq_processes[harq_pid]->G); + LOG_I(PHY,"start ldpc decode for CW 1 for AbsSubframe %d.%d / %d --> Kmimo %d \n", frame_rx, nr_tti_rx, harq_pid, dlsch1->Kmimo); + LOG_I(PHY,"start ldpc decode for CW 1 for AbsSubframe %d.%d / %d --> Pdcch Sym %d \n", frame_rx, nr_tti_rx, harq_pid, ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols); #endif #if UE_TIMING_TRACE @@ -4792,28 +4081,30 @@ void ue_dlsch_procedures(PHY_VARS_NR_UE *ue, #endif #ifdef UE_DLSCH_PARALLELISATION - ret1 = dlsch_decoding_mthread(ue,proc, eNB_id, - pdsch_vars->llr[1], - &ue->frame_parms, - dlsch1, - dlsch1->harq_processes[harq_pid], - frame_rx, - nr_tti_rx, - harq_pid, - pdsch==PDSCH?1:0, - dlsch1->harq_processes[harq_pid]->TBS>256?1:0); + ret1 = nr_dlsch_decoding_mthread(ue,proc, eNB_id, + pdsch_vars->llr[1], + &ue->frame_parms, + dlsch1, + dlsch1->harq_processes[harq_pid], + frame_rx, + nb_symb_sch, + nr_tti_rx, + harq_pid, + pdsch==PDSCH?1:0, + dlsch1->harq_processes[harq_pid]->TBS>256?1:0); #else - ret1 = dlsch_decoding(ue, - pdsch_vars->llr[1], - &ue->frame_parms, - dlsch1, - dlsch1->harq_processes[harq_pid], - frame_rx, - nr_tti_rx, - harq_pid, - pdsch==PDSCH?1:0,//proc->decoder_switch, - dlsch1->harq_processes[harq_pid]->TBS>256?1:0); + ret1 = nr_dlsch_decoding(ue, + pdsch_vars->llr[1], + &ue->frame_parms, + dlsch1, + dlsch1->harq_processes[harq_pid], + frame_rx, + nb_symb_sch, + nr_tti_rx, + harq_pid, + pdsch==PDSCH?1:0,//proc->decoder_switch, + dlsch1->harq_processes[harq_pid]->TBS>256?1:0); printf("start cw1 dlsch decoding\n"); #endif @@ -4821,190 +4112,29 @@ void ue_dlsch_procedures(PHY_VARS_NR_UE *ue, stop_meas(&ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]]); #if DISABLE_LOG_X printf(" --> Unscrambling for CW1 %5.3f\n", - (ue->dlsch_unscrambling_stats.p_time)/(cpuf*1000.0)); - printf("AbsSubframe %d.%d --> Turbo Decoding for CW1 %5.3f\n", - frame_rx%1024, nr_tti_rx,(ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); + (ue->dlsch_unscrambling_stats.p_time)/(cpuf*1000.0)); + printf("AbsSubframe %d.%d --> ldpc Decoding for CW1 %5.3f\n", + frame_rx%1024, nr_tti_rx,(ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); #else LOG_D(PHY, " --> Unscrambling for CW1 %5.3f\n", - (ue->dlsch_unscrambling_stats.p_time)/(cpuf*1000.0)); - LOG_D(PHY, "AbsSubframe %d.%d --> Turbo Decoding for CW1 %5.3f\n", - frame_rx%1024, nr_tti_rx,(ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); + (ue->dlsch_unscrambling_stats.p_time)/(cpuf*1000.0)); + LOG_D(PHY, "AbsSubframe %d.%d --> ldpc Decoding for CW1 %5.3f\n", + frame_rx%1024, nr_tti_rx,(ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); #endif #endif - LOG_I(PHY,"AbsSubframe %d.%d --> Turbo Decoding for CW1 %5.3f\n", - frame_rx%1024, nr_tti_rx,(ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); - } - - LOG_D(PHY," ------ end turbo decoder for AbsSubframe %d.%d ------ \n", frame_rx, nr_tti_rx); - } - - else { - LOG_D(PHY,"Calling dlsch_decoding_emul ...\n"); -#ifdef PHY_ABSTRACTION - ret = dlsch_decoding_emul(ue, - nr_tti_rx, - pdsch, - eNB_id); -#endif - } - - // Check CRC for CW 0 - if (ret == (1+dlsch0->max_turbo_iterations)) { - *dlsch_errors=*dlsch_errors+1; - - if(dlsch0->rnti != 0xffff) - { - LOG_D(PHY,"[UE %d][PDSCH %x/%d] AbsSubframe %d.%d : DLSCH CW0 in error (rv %d,round %d, mcs %d,TBS %d)\n", - ue->Mod_id,dlsch0->rnti, - harq_pid,frame_rx,nr_tti_rx, - dlsch0->harq_processes[harq_pid]->rvidx, - dlsch0->harq_processes[harq_pid]->round, - dlsch0->harq_processes[harq_pid]->mcs, - dlsch0->harq_processes[harq_pid]->TBS); - } - - - } else { - if(dlsch0->rnti != 0xffff) - { - LOG_D(PHY,"[UE %d][PDSCH %x/%d] AbsSubframe %d.%d : Received DLSCH CW0 (rv %d,round %d, mcs %d,TBS %d)\n", - ue->Mod_id,dlsch0->rnti, - harq_pid,frame_rx,nr_tti_rx, - dlsch0->harq_processes[harq_pid]->rvidx, - dlsch0->harq_processes[harq_pid]->round, - dlsch0->harq_processes[harq_pid]->mcs, - dlsch0->harq_processes[harq_pid]->TBS); - } - -#ifdef DEBUG_DLSCH - int j; - LOG_D(PHY,"dlsch harq_pid %d (rx): \n",dlsch0->current_harq_pid); - - for (j=0; jharq_processes[dlsch0->current_harq_pid]->TBS>>3; j++) - LOG_T(PHY,"%x.",dlsch0->harq_processes[dlsch0->current_harq_pid]->b[j]); - - LOG_T(PHY,"\n"); -#endif - - - switch (pdsch) { - case PDSCH: - /* if (ue->mac_enabled == 1) - mac_xface->ue_send_sdu(ue->Mod_id, - CC_id, - frame_rx, - nr_tti_rx, - dlsch0->harq_processes[dlsch0->current_harq_pid]->b, - dlsch0->harq_processes[dlsch0->current_harq_pid]->TBS>>3, - eNB_id);*/ - break; - case SI_PDSCH: - /*mac_xface->ue_decode_si(ue->Mod_id, - CC_id, - frame_rx, - eNB_id, - ue->dlsch_SI[eNB_id]->harq_processes[0]->b, - ue->dlsch_SI[eNB_id]->harq_processes[0]->TBS>>3);*/ - break; - case P_PDSCH: - /* if (ue->mac_enabled == 1) - mac_xface->ue_decode_p(ue->Mod_id, - CC_id, - frame_rx, - eNB_id, - ue->dlsch_SI[eNB_id]->harq_processes[0]->b, - ue->dlsch_SI[eNB_id]->harq_processes[0]->TBS>>3);*/ - break; - case RA_PDSCH: - if (ue->mac_enabled == 1) - process_rar(ue,proc,eNB_id,mode,abstraction_flag); - break; - case PDSCH1: - LOG_E(PHY,"Shouldn't have PDSCH1 yet, come back later\n"); - AssertFatal(1==0,"exiting"); - break; - case PMCH: - LOG_E(PHY,"Shouldn't have PMCH here\n"); - AssertFatal(1==0,"exiting"); - break; - - } - ue->total_TBS[eNB_id] = ue->total_TBS[eNB_id] + - dlsch0->harq_processes[dlsch0->current_harq_pid]->TBS; - ue->total_received_bits[eNB_id] = ue->total_TBS[eNB_id] + - dlsch0->harq_processes[dlsch0->current_harq_pid]->TBS; - } - // Check CRC for CW 1 - if(is_cw1_active) - { - if (ret1 == (1+dlsch0->max_turbo_iterations)) { - LOG_I(PHY,"[UE %d][PDSCH %x/%d] Frame %d nr_tti_rx %d DLSCH CW1 in error (rv %d,mcs %d,TBS %d)\n", - ue->Mod_id,dlsch0->rnti, - harq_pid,frame_rx,nr_tti_rx, - dlsch0->harq_processes[harq_pid]->rvidx, - dlsch0->harq_processes[harq_pid]->mcs, - dlsch0->harq_processes[harq_pid]->TBS); - - } else { - LOG_I(PHY,"[UE %d][PDSCH %x/%d] Frame %d nr_tti_rx %d: Received DLSCH CW1 (rv %d,mcs %d,TBS %d)\n", - ue->Mod_id,dlsch0->rnti, - harq_pid,frame_rx,nr_tti_rx, - dlsch0->harq_processes[harq_pid]->rvidx, - dlsch0->harq_processes[harq_pid]->mcs, - dlsch0->harq_processes[harq_pid]->TBS); - - - if (ue->mac_enabled == 1) { - switch (pdsch) { - case PDSCH: - if(is_cw1_active) - /*mac_xface->ue_send_sdu(ue->Mod_id, - CC_id, - frame_rx, - nr_tti_rx, - dlsch1->harq_processes[dlsch1->current_harq_pid]->b, - dlsch1->harq_processes[dlsch1->current_harq_pid]->TBS>>3, - eNB_id);*/ - break; - case SI_PDSCH: - case P_PDSCH: - case RA_PDSCH: - case PDSCH1: - case PMCH: - AssertFatal(0,"exiting"); - break; - } - } - } + LOG_I(PHY,"AbsSubframe %d.%d --> ldpc Decoding for CW1 %5.3f\n", + frame_rx%1024, nr_tti_rx,(ue->dlsch_decoding_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); } -#ifdef DEBUG_PHY_PROC - LOG_D(PHY,"[UE %d][PDSCH %x/%d] Frame %d nr_tti_rx %d: PDSCH/DLSCH decoding iter %d (mcs %d, rv %d, TBS %d)\n", - ue->Mod_id, - dlsch0->rnti,harq_pid, - frame_rx,nr_tti_rx,ret, - dlsch0->harq_processes[harq_pid]->mcs, - dlsch0->harq_processes[harq_pid]->rvidx, - dlsch0->harq_processes[harq_pid]->TBS); - - if (frame_rx%100==0) { - LOG_D(PHY,"[UE %d][PDSCH %x] Frame %d nr_tti_rx %d dlsch_errors %d, dlsch_received %d, dlsch_fer %d, current_dlsch_cqi %d\n", - ue->Mod_id,dlsch0->rnti, - frame_rx,nr_tti_rx, - ue->dlsch_errors[eNB_id], - ue->dlsch_received[eNB_id], - ue->dlsch_fer[eNB_id], - ue->measurements.wideband_cqi_tot[eNB_id]); + LOG_D(PHY," ------ end ldpc decoder for AbsSubframe %d.%d ------ \n", frame_rx, nr_tti_rx); } -#endif - } } -#endif + /*! * \brief This is the UE synchronize thread. @@ -5210,7 +4340,7 @@ void *UE_thread_slot1_dl_processing(void *arg) { // start slave thread for Pdsch Procedure (slot1) // do procedures for C-RNTI uint8_t eNB_id = 0; - uint8_t abstraction_flag = 0; + if (ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active == 1) { //wait until first ofdm symbol is processed //wait = 0; @@ -5303,9 +4433,9 @@ void *UE_thread_slot1_dl_processing(void *arg) { #endif #ifdef UE_SLOT_PARALLELISATION -int phy_procedures_slot_parallelization_UE_RX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id, - uint8_t abstraction_flag,uint8_t do_pdcch_flag,runmode_t mode, - relaying_type_t r_type) { +int phy_procedures_slot_parallelization_nrUE_RX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eNB_id, + uint8_t abstraction_flag,uint8_t do_pdcch_flag,runmode_t mode, + relaying_type_t r_type) { int l,l2; int pmch_flag=0; @@ -5829,15 +4959,21 @@ int phy_procedures_nrUE_RX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eN uint8_t do_pdcch_flag,runmode_t mode) { - //int l,l2; - //int pilot1; + + int l,l2; + int pilot1; + int frame_rx = proc->frame_rx; int nr_tti_rx = proc->nr_tti_rx; - proc->decoder_switch = 0; + uint16_t nb_symb_sch = 8; // to be updated by higher layer + uint8_t nb_symb_pdcch =2; + //proc->decoder_switch = 0; //int counter_decoder = 0; + + LOG_D(PHY," ****** start RX-Chain for AbsSubframe %d.%d ****** \n", frame_rx%1024, nr_tti_rx); - //uint8_t next1_thread_id = ue->current_thread_id[nr_tti_rx]== (RX_NB_TH-1) ? 0:(ue->current_thread_id[nr_tti_rx]+1); - //uint8_t next2_thread_id = next1_thread_id== (RX_NB_TH-1) ? 0:(next1_thread_id+1); + uint8_t next1_thread_id = ue->current_thread_id[nr_tti_rx]== (RX_NB_TH-1) ? 0:(ue->current_thread_id[nr_tti_rx]+1); + uint8_t next2_thread_id = next1_thread_id== (RX_NB_TH-1) ? 0:(next1_thread_id+1); #if 0 VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PHY_PROCEDURES_UE_RX, VCD_FUNCTION_IN); @@ -5860,12 +4996,6 @@ int phy_procedures_nrUE_RX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eN start_meas(&ue->generic_stat); #endif - - - - - - if (do_pdcch_flag) { // deactivate reception until we scan pdcch if (ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]) @@ -5894,20 +5024,22 @@ int phy_procedures_nrUE_RX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eN pilot1 = 3; } - + /* if (nr_subframe_select(&ue->frame_parms,nr_tti_rx) == SF_S) { // S-subframe, do first 5 symbols only l2 = 5; - } else { // normal nr_tti_rx, last symbol to be processed is the first of the second slot + } else */ + { // normal nr_tti_rx, last symbol to be processed is the first of the second slot l2 = (ue->frame_parms.symbols_per_tti/2)-1; } - int prev_nr_tti_rx = (nr_tti_rx - 1)<0? 9: (nr_tti_rx - 1); + int prev_nr_tti_rx = (nr_tti_rx - 1)<0? 9: (nr_tti_rx - 1);/* if (nr_subframe_select(&ue->frame_parms,prev_nr_tti_rx) != SF_DL) { //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ // RX processing of symbols l=0...l2 //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ l=0; - } else { + } else */ + { //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ // RX processing of symbols l=1...l2 (l=0 is done in last scheduling epoch) //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ @@ -5918,13 +5050,15 @@ int phy_procedures_nrUE_RX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eN LOG_D(PHY," ------ --> FFT/ChannelEst/PDCCH slot 0: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); #endif +#ifdef NR_PDCCH_SCHED //nr_gold_pdcch(ue,0, 2); - - for (int l=0; l<2; l++) { + + if (nr_tti_rx==1){ + for (uint16_t l=0; lofdm_demod_stats); #endif - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_SLOT_FEP, VCD_FUNCTION_IN); nr_slot_fep(ue, l, @@ -5937,180 +5071,128 @@ int phy_procedures_nrUE_RX(PHY_VARS_NR_UE *ue,UE_nr_rxtx_proc_t *proc,uint8_t eN #if UE_TIMING_TRACE stop_meas(&ue->ofdm_demod_stats); #endif + + //printf("phy procedure pdcch start measurement l =%d\n",l); + nr_ue_measurement_procedures(l,ue,proc,eNB_id,(nr_tti_rx<<1),mode); + } - - write_output("rxdataF.m","rxF",ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx>>1]].rxdataF[0],ue->frame_parms.ofdm_symbol_size*2,1,1); - - //ue_measurement_procedures(l-1,ue,proc,eNB_id,(nr_tti_rx<<1),abstraction_flag,mode); - - //if (do_pdcch_flag) { - // if ((l==pilot1) || ((pmch_flag==1)&(l==l2))) { - // LOG_D(PHY,"[UE %d] Frame %d: Calling pdcch procedures (eNB %d)\n",ue->Mod_id,frame_rx,eNB_id); - - //start_meas(&ue->rx_pdcch_stats[ue->current_thread_id[nr_tti_rx]]); - - //printf(">>> at phy_procedures_nrUE_RX, nr_ue_pdcch_procedures init, dlsch->active=%d\n", - // ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active); - - if (nr_ue_pdcch_procedures(eNB_id,ue,proc,0) == -1) { + if (nr_ue_pdcch_procedures(eNB_id,ue,proc) == -1) { LOG_E(PHY,"[UE %d] Frame %d, nr_tti_rx %d: Error in pdcch procedures\n",ue->Mod_id,frame_rx,nr_tti_rx); return(-1); } - //for (int m=0;m<1000;m++) printf("%d",m); - //printf("\n>>> at phy_procedures_nrUE_RX, nr_ue_pdcch_procedures end, dlsch->active=%d\n", - // ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active); - - // preparing for PDSCH procedures - // if (ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active == 1) { - // ue_pdsch_procedures(ue, - // proc, - // eNB_id, - // PDSCH, - // ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0], - // NULL, - // ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, - // ue->frame_parms.symbols_per_tti>>1, - // abstraction_flag); - // } -#if 0 - //stop_meas(&ue->rx_pdcch_stats[ue->current_thread_id[nr_tti_rx]]); - //printf("nr_tti_rx %d n_pdcch_sym %d pdcch procedures %5.3f \n", - // nr_tti_rx, ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, - // (ue->rx_pdcch_stats[ue->current_thread_id[nr_tti_rx]].p_time)/(cpuf*1000.0)); - LOG_D(PHY,"num_pdcch_symbols %d\n",ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols); - // } - //} - - // } // for l=1..l2 - - - ue_measurement_procedures(l-1,ue,proc,eNB_id,(nr_tti_rx<<1),abstraction_flag,mode); - - LOG_D(PHY," ------ end FFT/ChannelEst/PDCCH slot 0: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); - - return 0; -} - -nr_slot_fep(ue, - 0, - 1+(nr_tti_rx<<1), - 0, - 0, - 0, - NR_PDSCH_EST); -#endif - -// first slot has been processed (FFTs + Channel Estimation, PCFICH/PHICH/PDCCH) -#if UE_TIMING_TRACE -stop_meas(&ue->generic_stat); -#if DISABLE_LOG_X -printf("[SFN %d] Slot0: FFT + Channel Estimate + PCFICH/PHICH/PDCCH %5.2f \n",nr_tti_rx,ue->generic_stat.p_time/(cpuf*1000.0)); -#else -LOG_D(PHY, "[SFN %d] Slot0: FFT + Channel Estimate + PCFICH/PHICH/PDCCH %5.2f \n",nr_tti_rx,ue->generic_stat.p_time/(cpuf*1000.0)); -#endif - -#endif -#if 0 -LOG_D(PHY," ------ --> PDSCH ChannelComp/LLR slot 0: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); -//to update from pdsch config -nr_gold_pdsch(ue,0,0, 1); - -nr_slot_fep(ue, - 2, //to be updated from higher layer - (nr_tti_rx<<1), - 0, - 0, - 0, - NR_PDSCH_EST); - -#if UE_TIMING_TRACE -start_meas(&ue->generic_stat); -#endif -// do procedures for C-RNTI -if (ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active == 1) { - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC, VCD_FUNCTION_IN); - ue_pdsch_procedures(ue, - proc, - eNB_id, - PDSCH, - ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0], - NULL, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, - ue->frame_parms.symbols_per_tti>>1, - 0); - - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC, VCD_FUNCTION_OUT); } +#endif //NR_PDCCH_SCHED -LOG_D(PHY," ------ end PDSCH ChannelComp/LLR slot 0: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); -// do procedures for SI-RNTI -if ((ue->dlsch_SI[eNB_id]) && (ue->dlsch_SI[eNB_id]->active == 1)) { - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_SI, VCD_FUNCTION_IN); - ue_pdsch_procedures(ue, - proc, - eNB_id, - SI_PDSCH, - ue->dlsch_SI[eNB_id], - NULL, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, - ue->frame_parms.symbols_per_tti>>1, - 0); - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_SI, VCD_FUNCTION_OUT); - } - -// do procedures for SI-RNTI -if ((ue->dlsch_p[eNB_id]) && (ue->dlsch_p[eNB_id]->active == 1)) { - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_P, VCD_FUNCTION_IN); - ue_pdsch_procedures(ue, - proc, - eNB_id, - P_PDSCH, - ue->dlsch_p[eNB_id], - NULL, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, - ue->frame_parms.symbols_per_tti>>1, - 0); - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_P, VCD_FUNCTION_OUT); - } - -// do procedures for RA-RNTI -if ((ue->dlsch_ra[eNB_id]) && (ue->dlsch_ra[eNB_id]->active == 1)) { - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_RA, VCD_FUNCTION_IN); - ue_pdsch_procedures(ue, - proc, - eNB_id, - RA_PDSCH, - ue->dlsch_ra[eNB_id], - NULL, - ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, - ue->frame_parms.symbols_per_tti>>1, - 0); - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_RA, VCD_FUNCTION_OUT); - } - -LOG_D(PHY," ------ slot 1 Processing: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); -LOG_D(PHY," ------ --> FFT/ChannelEst/PDCCH slot 1: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); - -if (nr_subframe_select(&ue->frame_parms,nr_tti_rx) != SF_S) { // do front-end processing for second slot, and first symbol of next nr_tti_rx - for (l=1; lframe_parms.symbols_per_tti>>1; l++) { - if (abstraction_flag == 0) { -#if UE_TIMING_TRACE - start_meas(&ue->ofdm_demod_stats); -#endif - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_SLOT_FEP, VCD_FUNCTION_IN); + LOG_D(PHY," ------ --> PDSCH ChannelComp/LLR slot 0: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); + + if (nr_tti_rx==1){ + //to update from pdsch config + nr_gold_pdsch(ue,nb_symb_pdcch,0, 1); + + for (uint16_t m=nb_symb_pdcch;m<=(nb_symb_sch+nb_symb_pdcch-1) ; m++){ nr_slot_fep(ue, - l, - 1+(nr_tti_rx<<1), - 0, + m, //to be updated from higher layer + nr_tti_rx<<1, 0, 0, + 1, NR_PDSCH_EST); - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_SLOT_FEP, VCD_FUNCTION_OUT); + + //printf("phy procedure pdsch start measurement\n"); + nr_ue_measurement_procedures(m,ue,proc,eNB_id,(nr_tti_rx<<1),mode); + + } + //set active for testing, to be removed + ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active = 1; + } + else + ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active = 0; + +#if UE_TIMING_TRACE + start_meas(&ue->generic_stat); +#endif + // do procedures for C-RNTI + if (ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active == 1) { + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC, VCD_FUNCTION_IN); + nr_ue_pdsch_procedures(ue, + proc, + eNB_id, + PDSCH, + ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0], + NULL, + nb_symb_pdcch, //ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, + (nb_symb_sch+nb_symb_pdcch-1)); //ue->frame_parms.symbols_per_tti>>1, + + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC, VCD_FUNCTION_OUT); + } + + LOG_D(PHY," ------ end PDSCH ChannelComp/LLR slot 0: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); + // do procedures for SI-RNTI + if ((ue->dlsch_SI[eNB_id]) && (ue->dlsch_SI[eNB_id]->active == 1)) { + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_SI, VCD_FUNCTION_IN); + nr_ue_pdsch_procedures(ue, + proc, + eNB_id, + SI_PDSCH, + ue->dlsch_SI[eNB_id], + NULL, + ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, + ue->frame_parms.symbols_per_tti>>1); + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_SI, VCD_FUNCTION_OUT); + } + + // do procedures for SI-RNTI + if ((ue->dlsch_p[eNB_id]) && (ue->dlsch_p[eNB_id]->active == 1)) { + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_P, VCD_FUNCTION_IN); + nr_ue_pdsch_procedures(ue, + proc, + eNB_id, + P_PDSCH, + ue->dlsch_p[eNB_id], + NULL, + ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, + ue->frame_parms.symbols_per_tti>>1); + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_P, VCD_FUNCTION_OUT); + } + + // do procedures for RA-RNTI + if ((ue->dlsch_ra[eNB_id]) && (ue->dlsch_ra[eNB_id]->active == 1)) { + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_RA, VCD_FUNCTION_IN); + nr_ue_pdsch_procedures(ue, + proc, + eNB_id, + RA_PDSCH, + ue->dlsch_ra[eNB_id], + NULL, + ue->pdcch_vars[ue->current_thread_id[nr_tti_rx]][eNB_id]->num_pdcch_symbols, + ue->frame_parms.symbols_per_tti>>1); + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC_RA, VCD_FUNCTION_OUT); + } +//#if 0 + LOG_D(PHY," ------ slot 1 Processing: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); + LOG_D(PHY," ------ --> FFT/ChannelEst/PDCCH slot 1: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); + + /*if (nr_subframe_select(&ue->frame_parms,nr_tti_rx) != SF_S)*/ + { // do front-end processing for second slot, and first symbol of next nr_tti_rx + for (l=1; lframe_parms.symbols_per_tti>>1; l++) { +#if UE_TIMING_TRACE + start_meas(&ue->ofdm_demod_stats); +#endif + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_SLOT_FEP, VCD_FUNCTION_IN); + /*nr_slot_fep(ue, + l, + 1+(nr_tti_rx<<1), + 0, + 0, + 0, + NR_PDSCH_EST);*/ + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_UE_SLOT_FEP, VCD_FUNCTION_OUT); + #if UE_TIMING_TRACE stop_meas(&ue->ofdm_demod_stats); #endif - } + //ue_measurement_procedures(l-1,ue,proc,eNB_id,1+(nr_tti_rx<<1),abstraction_flag,mode); @@ -6118,66 +5200,80 @@ if (nr_subframe_select(&ue->frame_parms,nr_tti_rx) != SF_S) { // do front-end p // do first symbol of next downlink nr_tti_rx for channel estimation int next_nr_tti_rx = (1+nr_tti_rx)%10; - if (nr_subframe_select(&ue->frame_parms,next_nr_tti_rx) != SF_UL) + /* if (nr_subframe_select(&ue->frame_parms,next_nr_tti_rx) != SF_UL)*/ { - nr_slot_fep(ue, - 0, - (next_nr_tti_rx<<1), - 0, - 0, - 0, - NR_PDSCH_EST); + /*nr_slot_fep(ue, + 0, + (next_nr_tti_rx<<1), + 0, + 0, + 0, + NR_PDSCH_EST);*/ } - } // not an S-subframe + } // not an S-subframe #if UE_TIMING_TRACE -stop_meas(&ue->generic_stat); + stop_meas(&ue->generic_stat); #if DISABLE_LOG_X -printf("[SFN %d] Slot1: FFT + Channel Estimate + Pdsch Proc Slot0 %5.2f \n",nr_tti_rx,ue->generic_stat.p_time/(cpuf*1000.0)); + printf("[SFN %d] Slot1: FFT + Channel Estimate + Pdsch Proc Slot0 %5.2f \n",nr_tti_rx,ue->generic_stat.p_time/(cpuf*1000.0)); #else -LOG_D(PHY, "[SFN %d] Slot1: FFT + Channel Estimate + Pdsch Proc Slot0 %5.2f \n",nr_tti_rx,ue->generic_stat.p_time/(cpuf*1000.0)); + LOG_D(PHY, "[SFN %d] Slot1: FFT + Channel Estimate + Pdsch Proc Slot0 %5.2f \n",nr_tti_rx,ue->generic_stat.p_time/(cpuf*1000.0)); +#endif + #endif -#endif + //LOG_D(PHY," ------ end FFT/ChannelEst/PDCCH slot 1: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); -LOG_D(PHY," ------ end FFT/ChannelEst/PDCCH slot 1: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); - -if ( (nr_tti_rx == 0) && (ue->decode_MIB == 1)) - { - ue_pbch_procedures(eNB_id,ue,proc,0); - } - -// do procedures for C-RNTI -LOG_D(PHY," ------ --> PDSCH ChannelComp/LLR slot 0: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); -if (ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active == 1) { - - VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC, VCD_FUNCTION_IN); + if ( (nr_tti_rx == 0) && (ue->decode_MIB == 1)) + { + for (int i=0; i<3; i++) + nr_slot_fep(ue, + (5+i), //mu=1 case B + nr_tti_rx, + 0, + 0, + 1, + NR_PBCH_EST); + + nr_ue_pbch_procedures(eNB_id,ue,proc,0); + } + + // do procedures for C-RNTI + LOG_D(PHY," ------ --> PDSCH ChannelComp/LLR slot 0: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); + if (ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->active == 1) { + + VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_PDSCH_PROC, VCD_FUNCTION_IN); #if UE_TIMING_TRACE - start_meas(&ue->pdsch_procedures_stat[ue->current_thread_id[nr_tti_rx]]); + + start_meas(&ue->pdsch_procedures_stat[ue->current_thread_id[nr_tti_rx]]); #endif - ue_pdsch_procedures(ue, - proc, - eNB_id, - PDSCH, - ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0], - NULL, - 1+(ue->frame_parms.symbols_per_tti>>1), - ue->frame_parms.symbols_per_tti-1, - 0); - LOG_D(PHY," ------ end PDSCH ChannelComp/LLR slot 0: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); - LOG_D(PHY," ------ --> PDSCH Turbo Decoder slot 0/1: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); + /*nr_ue_pdsch_procedures(ue, + proc, + eNB_id, + PDSCH, + ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0], + NULL, + 1+(ue->frame_parms.symbols_per_tti>>1), + ue->frame_parms.symbols_per_tti-1, + abstraction_flag); + LOG_D(PHY," ------ end PDSCH ChannelComp/LLR slot 0: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx); + LOG_D(PHY," ------ --> PDSCH Turbo Decoder slot 0/1: AbsSubframe %d.%d ------ \n", frame_rx%1024, nr_tti_rx);*/ #if UE_TIMING_TRACE - stop_meas(&ue->pdsch_procedures_stat[ue->current_thread_id[nr_tti_rx]]); - start_meas(&ue->dlsch_procedures_stat[ue->current_thread_id[nr_tti_rx]]); + stop_meas(&ue->pdsch_procedures_stat[ue->current_thread_id[nr_tti_rx]]); + start_meas(&ue->dlsch_procedures_stat[ue->current_thread_id[nr_tti_rx]]); #endif - /*ue_dlsch_procedures(ue, - proc, - eNB_id, - PDSCH, - ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0], - ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][1], - &ue->dlsch_errors[eNB_id], - mode, - 0);*/ + +//#endif //slot 1 + + nr_ue_dlsch_procedures(ue, + proc, + eNB_id, + PDSCH, + ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0], + ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][1], + &ue->dlsch_errors[eNB_id], + mode); + + #if UE_TIMING_TRACE stop_meas(&ue->dlsch_procedures_stat[ue->current_thread_id[nr_tti_rx]]); #if DISABLE_LOG_X @@ -6198,103 +5294,103 @@ start_meas(&ue->generic_stat); #endif #if 0 -if(nr_tti_rx==5 && ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->harq_processes[ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->current_harq_pid]->nb_rb > 20){ - //write_output("decoder_llr.m","decllr",dlsch_llr,G,1,0); - //write_output("llr.m","llr", &ue->pdsch_vars[eNB_id]->llr[0][0],(14*nb_rb*12*dlsch1_harq->Qm) - 4*(nb_rb*4*dlsch1_harq->Qm),1,0); - write_output("rxdataF0_current.m" , "rxdataF0", &ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF[0][0],14*ue->frame_parms.ofdm_symbol_size,1,1); - //write_output("rxdataF0_previous.m" , "rxdataF0_prev_sss", &ue->common_vars.common_vars_rx_data_per_thread[next_thread_id].rxdataF[0][0],14*ue->frame_parms.ofdm_symbol_size,1,1); + if(nr_tti_rx==5 && ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->harq_processes[ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->current_harq_pid]->nb_rb > 20){ + //write_output("decoder_llr.m","decllr",dlsch_llr,G,1,0); + //write_output("llr.m","llr", &ue->pdsch_vars[eNB_id]->llr[0][0],(14*nb_rb*12*dlsch1_harq->Qm) - 4*(nb_rb*4*dlsch1_harq->Qm),1,0); - //write_output("rxdataF0_previous.m" , "rxdataF0_prev", &ue->common_vars.common_vars_rx_data_per_thread[next_thread_id].rxdataF[0][0],14*ue->frame_parms.ofdm_symbol_size,1,1); + write_output("rxdataF0_current.m" , "rxdataF0", &ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].rxdataF[0][0],14*ue->frame_parms.ofdm_symbol_size,1,1); + //write_output("rxdataF0_previous.m" , "rxdataF0_prev_sss", &ue->common_vars.common_vars_rx_data_per_thread[next_thread_id].rxdataF[0][0],14*ue->frame_parms.ofdm_symbol_size,1,1); - write_output("dl_ch_estimates.m", "dl_ch_estimates_sfn5", &ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[0][0][0],14*ue->frame_parms.ofdm_symbol_size,1,1); - write_output("dl_ch_estimates_ext.m", "dl_ch_estimatesExt_sfn5", &ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][0]->dl_ch_estimates_ext[0][0],14*ue->frame_parms.N_RB_DL*12,1,1); - write_output("rxdataF_comp00.m","rxdataF_comp00", &ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][0]->rxdataF_comp0[0][0],14*ue->frame_parms.N_RB_DL*12,1,1); - //write_output("magDLFirst.m", "magDLFirst", &phy_vars_ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][0]->dl_ch_mag0[0][0],14*frame_parms->N_RB_DL*12,1,1); - //write_output("magDLSecond.m", "magDLSecond", &phy_vars_ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][0]->dl_ch_magb0[0][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("rxdataF0_previous.m" , "rxdataF0_prev", &ue->common_vars.common_vars_rx_data_per_thread[next_thread_id].rxdataF[0][0],14*ue->frame_parms.ofdm_symbol_size,1,1); - AssertFatal (0,""); - } + write_output("dl_ch_estimates.m", "dl_ch_estimates_sfn5", &ue->common_vars.common_vars_rx_data_per_thread[ue->current_thread_id[nr_tti_rx]].dl_ch_estimates[0][0][0],14*ue->frame_parms.ofdm_symbol_size,1,1); + write_output("dl_ch_estimates_ext.m", "dl_ch_estimatesExt_sfn5", &ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][0]->dl_ch_estimates_ext[0][0],14*ue->frame_parms.N_RB_DL*12,1,1); + write_output("rxdataF_comp00.m","rxdataF_comp00", &ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][0]->rxdataF_comp0[0][0],14*ue->frame_parms.N_RB_DL*12,1,1); + //write_output("magDLFirst.m", "magDLFirst", &phy_vars_ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][0]->dl_ch_mag0[0][0],14*frame_parms->N_RB_DL*12,1,1); + //write_output("magDLSecond.m", "magDLSecond", &phy_vars_ue->pdsch_vars[ue->current_thread_id[nr_tti_rx]][0]->dl_ch_magb0[0][0],14*frame_parms->N_RB_DL*12,1,1); + + AssertFatal (0,""); + } #endif -// do procedures for SI-RNTI -if ((ue->dlsch_SI[eNB_id]) && (ue->dlsch_SI[eNB_id]->active == 1)) { - ue_pdsch_procedures(ue, - proc, - eNB_id, - SI_PDSCH, - ue->dlsch_SI[eNB_id], - NULL, - 1+(ue->frame_parms.symbols_per_tti>>1), - ue->frame_parms.symbols_per_tti-1, - 0); + // do procedures for SI-RNTI + if ((ue->dlsch_SI[eNB_id]) && (ue->dlsch_SI[eNB_id]->active == 1)) { + nr_ue_pdsch_procedures(ue, + proc, + eNB_id, + SI_PDSCH, + ue->dlsch_SI[eNB_id], + NULL, + 1+(ue->frame_parms.symbols_per_tti>>1), + ue->frame_parms.symbols_per_tti-1); - /*ue_dlsch_procedures(ue, - proc, - eNB_id, - SI_PDSCH, - ue->dlsch_SI[eNB_id], - NULL, - &ue->dlsch_SI_errors[eNB_id], - mode, - 0); + /*ue_dlsch_procedures(ue, + proc, + eNB_id, + SI_PDSCH, + ue->dlsch_SI[eNB_id], + NULL, + &ue->dlsch_SI_errors[eNB_id], + mode, + abstraction_flag); ue->dlsch_SI[eNB_id]->active = 0;*/ - } + } -// do procedures for P-RNTI -if ((ue->dlsch_p[eNB_id]) && (ue->dlsch_p[eNB_id]->active == 1)) { - ue_pdsch_procedures(ue, - proc, - eNB_id, - P_PDSCH, - ue->dlsch_p[eNB_id], - NULL, - 1+(ue->frame_parms.symbols_per_tti>>1), - ue->frame_parms.symbols_per_tti-1, - 0); + // do procedures for P-RNTI + if ((ue->dlsch_p[eNB_id]) && (ue->dlsch_p[eNB_id]->active == 1)) { + nr_ue_pdsch_procedures(ue, + proc, + eNB_id, + P_PDSCH, + ue->dlsch_p[eNB_id], + NULL, + 1+(ue->frame_parms.symbols_per_tti>>1), + ue->frame_parms.symbols_per_tti-1); - /*ue_dlsch_procedures(ue, - proc, - eNB_id, - P_PDSCH, - ue->dlsch_p[eNB_id], - NULL, - &ue->dlsch_p_errors[eNB_id], - mode, - 0);*/ - ue->dlsch_p[eNB_id]->active = 0; - } -// do procedures for RA-RNTI -if ((ue->dlsch_ra[eNB_id]) && (ue->dlsch_ra[eNB_id]->active == 1)) { - ue_pdsch_procedures(ue, - proc, - eNB_id, - RA_PDSCH, - ue->dlsch_ra[eNB_id], - NULL, - 1+(ue->frame_parms.symbols_per_tti>>1), - ue->frame_parms.symbols_per_tti-1, - 0); - /*ue_dlsch_procedures(ue, - proc, - eNB_id, - RA_PDSCH, - ue->dlsch_ra[eNB_id], - NULL, - &ue->dlsch_ra_errors[eNB_id], - mode, - 0);*/ - ue->dlsch_ra[eNB_id]->active = 0; - } -// duplicate harq structure + /*ue_dlsch_procedures(ue, + proc, + eNB_id, + P_PDSCH, + ue->dlsch_p[eNB_id], + NULL, + &ue->dlsch_p_errors[eNB_id], + mode, + abstraction_flag);*/ + ue->dlsch_p[eNB_id]->active = 0; + } + // do procedures for RA-RNTI + if ((ue->dlsch_ra[eNB_id]) && (ue->dlsch_ra[eNB_id]->active == 1)) { + nr_ue_pdsch_procedures(ue, + proc, + eNB_id, + RA_PDSCH, + ue->dlsch_ra[eNB_id], + NULL, + 1+(ue->frame_parms.symbols_per_tti>>1), + ue->frame_parms.symbols_per_tti-1); -uint8_t current_harq_pid = ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->current_harq_pid; -NR_DL_UE_HARQ_t *current_harq_processes = ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->harq_processes[current_harq_pid]; -NR_DL_UE_HARQ_t *harq_processes_dest = ue->dlsch[next1_thread_id][eNB_id][0]->harq_processes[current_harq_pid]; -NR_DL_UE_HARQ_t *harq_processes_dest1 = ue->dlsch[next2_thread_id][eNB_id][0]->harq_processes[current_harq_pid]; + /*ue_dlsch_procedures(ue, + proc, + eNB_id, + RA_PDSCH, + ue->dlsch_ra[eNB_id], + NULL, + &ue->dlsch_ra_errors[eNB_id], + mode, + abstraction_flag);*/ + ue->dlsch_ra[eNB_id]->active = 0; + } -/*nr_harq_status_t *current_harq_ack = &ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->harq_ack[nr_tti_rx]; + // duplicate harq structure + + uint8_t current_harq_pid = ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->current_harq_pid; + NR_DL_UE_HARQ_t *current_harq_processes = ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->harq_processes[current_harq_pid]; + NR_DL_UE_HARQ_t *harq_processes_dest = ue->dlsch[next1_thread_id][eNB_id][0]->harq_processes[current_harq_pid]; + NR_DL_UE_HARQ_t *harq_processes_dest1 = ue->dlsch[next2_thread_id][eNB_id][0]->harq_processes[current_harq_pid]; + + /*nr_harq_status_t *current_harq_ack = &ue->dlsch[ue->current_thread_id[nr_tti_rx]][eNB_id][0]->harq_ack[nr_tti_rx]; nr_harq_status_t *harq_ack_dest = &ue->dlsch[next1_thread_id][eNB_id][0]->harq_ack[nr_tti_rx]; nr_harq_status_t *harq_ack_dest1 = &ue->dlsch[next2_thread_id][eNB_id][0]->harq_ack[nr_tti_rx]; */ @@ -6350,7 +5446,7 @@ LOG_D(PHY, "------FULL RX PROC [SFN %d]: %5.2f ------\n",nr_tti_rx,ue->phy_proc_ #endif #endif -#endif //pdsch +//#endif //pdsch LOG_D(PHY," ****** end RX-Chain for AbsSubframe %d.%d ****** \n", frame_rx%1024, nr_tti_rx); return (0); diff --git a/openair1/SCHED_NR_UE/pucch_uci_ue_nr.c b/openair1/SCHED_NR_UE/pucch_uci_ue_nr.c index bc23c5a974..fb38c80eb8 100644 --- a/openair1/SCHED_NR_UE/pucch_uci_ue_nr.c +++ b/openair1/SCHED_NR_UE/pucch_uci_ue_nr.c @@ -193,7 +193,7 @@ bool pucch_procedures_ue_nr(PHY_VARS_NR_UE *ue, uint8_t gNB_id, UE_nr_rxtx_proc_ if (dl_harq_pid < ue->dlsch[ue->current_thread_id[proc->nr_tti_rx]][gNB_id][0]->number_harq_processes_for_pdsch) { /* pucch indicator can be reseted in function get_downlink_ack so it should be get now */ - pucch_resource_indicator = ue->dlsch[ue->current_thread_id[proc->nr_tti_rx]][gNB_id][0]->harq_processes[dl_harq_pid].harq_ack.pucch_resource_indicator; + pucch_resource_indicator = ue->dlsch[ue->current_thread_id[proc->nr_tti_rx]][gNB_id][0]->harq_processes[dl_harq_pid]->harq_ack.pucch_resource_indicator; } /* Part - I @@ -292,7 +292,7 @@ bool pucch_procedures_ue_nr(PHY_VARS_NR_UE *ue, uint8_t gNB_id, UE_nr_rxtx_proc_ int occ_length = 0; int occ_Index = 0; - NR_UE_HARQ_STATUS_t *harq_status = &ue->dlsch[ue->current_thread_id[proc->nr_tti_rx]][gNB_id][0]->harq_processes[dl_harq_pid].harq_ack; + NR_UE_HARQ_STATUS_t *harq_status = &ue->dlsch[ue->current_thread_id[proc->nr_tti_rx]][gNB_id][0]->harq_processes[dl_harq_pid]->harq_ack; if (select_pucch_resource(ue, gNB_id, N_UCI, pucch_resource_indicator, &initial_pucch_id, &pucch_resource_set, &pucch_resource_id, harq_status) == TRUE) { @@ -691,7 +691,7 @@ uint8_t get_downlink_ack(PHY_VARS_NR_UE *ue, uint8_t gNB_id, UE_nr_rxtx_proc_t for (int dl_harq_pid = 0; dl_harq_pid < number_pid_dl; dl_harq_pid++) { - harq_status = &ue->dlsch[ue->current_thread_id[proc->nr_tti_rx]][gNB_id][code_word]->harq_processes[dl_harq_pid].harq_ack; + harq_status = &ue->dlsch[ue->current_thread_id[proc->nr_tti_rx]][gNB_id][code_word]->harq_processes[dl_harq_pid]->harq_ack; /* check if current tx slot should transmit downlink acknowlegment */ if (harq_status->slot_for_feedback_ack == proc->nr_tti_tx) { diff --git a/openair1/SIMULATION/NR_PHY/dlschsim.c b/openair1/SIMULATION/NR_PHY/dlschsim.c new file mode 100644 index 0000000000..565776d584 --- /dev/null +++ b/openair1/SIMULATION/NR_PHY/dlschsim.c @@ -0,0 +1,677 @@ +/* + * Licensed to the OpenAirInterface (OAI) Software Alliance under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The OpenAirInterface Software Alliance licenses this file to You under + * the OAI Public License, Version 1.1 (the "License"); you may not use this file + * except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.openairinterface.org/?page_id=698 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------------------- + * For more information about the OpenAirInterface (OAI) Software Alliance: + * contact@openairinterface.org + */ + +#include +#include +#include +#include +#include +#include + +#include "common/config/config_userapi.h" +#include "common/utils/LOG/log.h" +#include "common/ran_context.h" + +#include "SIMULATION/TOOLS/sim.h" +#include "SIMULATION/RF/rf.h" +#include "PHY/types.h" +#include "PHY/defs_nr_common.h" +#include "PHY/defs_nr_UE.h" +#include "PHY/defs_gNB.h" +#include "PHY/NR_REFSIG/refsig_defs_ue.h" +#include "PHY/NR_REFSIG/nr_mod_table.h" +#include "PHY/MODULATION/modulation_eNB.h" +#include "PHY/MODULATION/modulation_UE.h" +#include "PHY/INIT/phy_init.h" +#include "PHY/NR_TRANSPORT/nr_transport.h" +#include "PHY/NR_UE_TRANSPORT/nr_transport_proto_ue.h" +#include "PHY/NR_TRANSPORT/nr_dlsch.h" + +#include "SCHED_NR/sched_nr.h" + +//#include "PHY/MODULATION/modulation_common.h" +//#include "common/config/config_load_configmodule.h" +//#include "UTIL/LISTS/list.h" +//#include "common/ran_context.h" + +PHY_VARS_gNB *gNB; +PHY_VARS_NR_UE *UE; +RAN_CONTEXT_t RC; + + +double cpuf; + +// dummy functions +int nfapi_mode=0; +int oai_nfapi_hi_dci0_req(nfapi_hi_dci0_request_t *hi_dci0_req) { + return(0); +} +int oai_nfapi_tx_req(nfapi_tx_request_t *tx_req) { + return(0); +} + +int oai_nfapi_dl_config_req(nfapi_dl_config_request_t *dl_config_req) { + return(0); +} + +int oai_nfapi_ul_config_req(nfapi_ul_config_request_t *ul_config_req) { + return(0); +} + +int oai_nfapi_nr_dl_config_req(nfapi_nr_dl_config_request_t *dl_config_req) { + return(0); +} + +uint32_t from_earfcn(int eutra_bandP,uint32_t dl_earfcn) { + return(0); +} +int32_t get_uldl_offset(int eutra_bandP) { + return(0); +} + +NR_IF_Module_t *NR_IF_Module_init(int Mod_id) { + return(NULL); +} + +void exit_function(const char *file, const char *function, const int line,const char *s) { + const char *msg= s==NULL ? "no comment": s; + printf("Exiting at: %s:%d %s(), %s\n", file, line, function, msg); + exit(-1); +} + +// needed for some functions +PHY_VARS_NR_UE *PHY_vars_UE_g[1][1]= {{NULL}}; +uint16_t n_rnti=0x1234; + +char quantize(double D,double x,unsigned char B) { + double qxd; + short maxlev; + qxd = floor(x/D); + maxlev = 1<<(B-1);//(char)(pow(2,B-1)); + + //printf("x=%f,qxd=%f,maxlev=%d\n",x,qxd, maxlev); + + if (qxd <= -maxlev) + qxd = -maxlev; + else if (qxd >= maxlev) + qxd = maxlev-1; + + return((char)qxd); +} + +int main(int argc, char **argv) { + char c; + int i; //,j,l,aa; + double SNR,SNR_lin,snr0=-2.0,snr1=2.0; + double snr_step = 0.1; + uint8_t snr1set=0; + int **txdata; + double **s_re,**s_im,**r_re,**r_im; + // int sync_pos, sync_pos_slot; + // FILE *rx_frame_file; + FILE *output_fd = NULL; + uint8_t write_output_file=0; + // int subframe_offset; + // char fname[40], vname[40]; + int trial,n_trials=1,n_errors=0,n_false_positive=0; + uint8_t transmission_mode = 1,n_tx=1,n_rx=1; + uint16_t Nid_cell=0; + channel_desc_t *gNB2UE; + uint8_t extended_prefix_flag=0; + int8_t interf1=-21,interf2=-21; + FILE *input_fd=NULL,*pbch_file_fd=NULL; + char input_val_str[50],input_val_str2[50]; + //uint16_t NB_RB=25; + SCM_t channel_model=AWGN;//Rayleigh1_anticorr; + uint8_t N_RB_DL=106,mu=1; + unsigned char frame_type = 0; + unsigned char pbch_phase = 0; + int frame=0,subframe=0; + int frame_length_complex_samples; + //int frame_length_complex_samples_no_prefix; + NR_DL_FRAME_PARMS *frame_parms; + //nfapi_nr_config_request_t *gNB_config; + uint8_t Kmimo; + uint32_t Nsoft; + double sigma; + unsigned char qbits=8; + int ret; + int run_initial_sync=0; + int loglvl=OAILOG_WARNING; + float target_error_rate = 0.01; + + uint16_t nb_symb_sch =12; + uint16_t nb_rb = 50; + uint8_t Imcs=9; + + cpuf = get_cpu_freq_GHz(); + + if ( load_configmodule(argc,argv) == 0) { + exit_fun("[SOFTMODEM] Error, configuration module init failed\n"); + } + + //logInit(); + randominit(0); + + while ((c = getopt (argc, argv, "df:hpg:i:j:n:l:m:r:s:S:y:z:N:F:R:P:")) != -1) { + switch (c) { + case 'f': + write_output_file=1; + output_fd = fopen(optarg,"w"); + + if (output_fd==NULL) { + printf("Error opening %s\n",optarg); + exit(-1); + } + + break; + + case 'd': + frame_type = 1; + break; + + case 'g': + switch((char)*optarg) { + case 'A': + channel_model=SCM_A; + break; + + case 'B': + channel_model=SCM_B; + break; + + case 'C': + channel_model=SCM_C; + break; + + case 'D': + channel_model=SCM_D; + break; + + case 'E': + channel_model=EPA; + break; + + case 'F': + channel_model=EVA; + break; + + case 'G': + channel_model=ETU; + break; + + default: + msg("Unsupported channel model!\n"); + exit(-1); + } + + break; + + case 'i': + interf1=atoi(optarg); + break; + + case 'j': + interf2=atoi(optarg); + break; + + case 'n': + n_trials = atoi(optarg); + break; + + case 's': + snr0 = atof(optarg); + msg("Setting SNR0 to %f\n",snr0); + break; + + case 'S': + snr1 = atof(optarg); + snr1set=1; + msg("Setting SNR1 to %f\n",snr1); + break; + + case 'p': + extended_prefix_flag=1; + break; + + /* + case 'r': + ricean_factor = pow(10,-.1*atof(optarg)); + if (ricean_factor>1) { + printf("Ricean factor must be between 0 and 1\n"); + exit(-1); + } + break; + */ + + case 'y': + n_tx=atoi(optarg); + + if ((n_tx==0) || (n_tx>2)) { + msg("Unsupported number of tx antennas %d\n",n_tx); + exit(-1); + } + + break; + + case 'z': + n_rx=atoi(optarg); + + if ((n_rx==0) || (n_rx>2)) { + msg("Unsupported number of rx antennas %d\n",n_rx); + exit(-1); + } + + break; + + case 'N': + Nid_cell = atoi(optarg); + break; + + case 'R': + N_RB_DL = atoi(optarg); + break; + + case 'F': + input_fd = fopen(optarg,"r"); + + if (input_fd==NULL) { + printf("Problem with filename %s\n",optarg); + exit(-1); + } + + break; + + case 'P': + pbch_phase = atoi(optarg); + + if (pbch_phase>3) + printf("Illegal PBCH phase (0-3) got %d\n",pbch_phase); + + break; + + case 'm': + Imcs = atoi(optarg); + break; + + case 'l': + nb_symb_sch = atoi(optarg); + break; + + case 'r': + nb_rb = atoi(optarg); + break; + + default: + case 'h': + printf("%s -h(elp) -p(extended_prefix) -N cell_id -f output_filename -F input_filename -g channel_model -n n_frames -t Delayspread -s snr0 -S snr1 -x transmission_mode -y TXant -z RXant -i Intefrence0 -j Interference1 -A interpolation_file -C(alibration offset dB) -N CellId\n", + argv[0]); + printf("-h This message\n"); + printf("-p Use extended prefix mode\n"); + printf("-d Use TDD\n"); + printf("-n Number of frames to simulate\n"); + printf("-s Starting SNR, runs from SNR0 to SNR0 + 5 dB. If n_frames is 1 then just SNR is simulated\n"); + printf("-S Ending SNR, runs from SNR0 to SNR1\n"); + printf("-t Delay spread for multipath channel\n"); + printf("-g [A,B,C,D,E,F,G] Use 3GPP SCM (A,B,C,D) or 36-101 (E-EPA,F-EVA,G-ETU) models (ignores delay spread and Ricean factor)\n"); + printf("-x Transmission mode (1,2,6 for the moment)\n"); + printf("-y Number of TX antennas used in eNB\n"); + printf("-z Number of RX antennas used in UE\n"); + printf("-i Relative strength of first intefering eNB (in dB) - cell_id mod 3 = 1\n"); + printf("-j Relative strength of second intefering eNB (in dB) - cell_id mod 3 = 2\n"); + printf("-N Nid_cell\n"); + printf("-R N_RB_DL\n"); + printf("-O oversampling factor (1,2,4,8,16)\n"); + printf("-A Interpolation_filname Run with Abstraction to generate Scatter plot using interpolation polynomial in file\n"); + // printf("-C Generate Calibration information for Abstraction (effective SNR adjustment to remove Pe bias w.r.t. AWGN)\n"); + printf("-f Output filename (.txt format) for Pe/SNR results\n"); + printf("-F Input filename (.txt format) for RX conformance testing\n"); + exit (-1); + break; + } + } + + logInit(); + set_glog(loglvl); + T_stdout = 1; + + if (snr1set==0) + snr1 = snr0+10; + + gNB2UE = new_channel_desc_scm(n_tx, + n_rx, + channel_model, + 61.44e6, //N_RB2sampling_rate(N_RB_DL), + 40e6, //N_RB2channel_bandwidth(N_RB_DL), + 0, + 0, + 0); + + if (gNB2UE==NULL) { + msg("Problem generating channel model. Exiting.\n"); + exit(-1); + } + + RC.gNB = (PHY_VARS_gNB ** *) malloc(sizeof(PHY_VARS_gNB **)); + RC.gNB[0] = (PHY_VARS_gNB **) malloc(sizeof(PHY_VARS_gNB *)); + RC.gNB[0][0] = malloc(sizeof(PHY_VARS_gNB)); + gNB = RC.gNB[0][0]; + //gNB_config = &gNB->gNB_config; + frame_parms = &gNB->frame_parms; //to be initialized I suppose (maybe not necessary for PBCH) + frame_parms->nb_antennas_tx = n_tx; + frame_parms->nb_antennas_rx = n_rx; + frame_parms->N_RB_DL = N_RB_DL; + crcTableInit(); + nr_phy_config_request_sim(gNB,N_RB_DL,N_RB_DL,mu); + phy_init_nr_gNB(gNB,0,0); + //init_eNB_afterRU(); + frame_length_complex_samples = frame_parms->samples_per_subframe; + //frame_length_complex_samples_no_prefix = frame_parms->samples_per_subframe_wCP; + s_re = malloc(2*sizeof(double *)); + s_im = malloc(2*sizeof(double *)); + r_re = malloc(2*sizeof(double *)); + r_im = malloc(2*sizeof(double *)); + txdata = malloc(2*sizeof(int *)); + + for (i=0; i<2; i++) { + s_re[i] = malloc(frame_length_complex_samples*sizeof(double)); + bzero(s_re[i],frame_length_complex_samples*sizeof(double)); + s_im[i] = malloc(frame_length_complex_samples*sizeof(double)); + bzero(s_im[i],frame_length_complex_samples*sizeof(double)); + r_re[i] = malloc(frame_length_complex_samples*sizeof(double)); + bzero(r_re[i],frame_length_complex_samples*sizeof(double)); + r_im[i] = malloc(frame_length_complex_samples*sizeof(double)); + bzero(r_im[i],frame_length_complex_samples*sizeof(double)); + txdata[i] = malloc(frame_length_complex_samples*sizeof(int)); + bzero(r_re[i],frame_length_complex_samples*sizeof(int)); + } + + if (pbch_file_fd!=NULL) { + load_pbch_desc(pbch_file_fd); + } + + /* for (int k=0; k<2; k++) { + // Create transport channel structures for 2 transport blocks (MIMO) + for (i=0; i<2; i++) { + gNB->dlsch[k][i] = new_gNB_dlsch(Kmimo,8,Nsoft,0,frame_parms,gNB_config); + + if (!gNB->dlsch[k][i]) { + printf("Can't get eNB dlsch structures\n"); + exit(-1); + } + gNB->dlsch[k][i]->Nsoft = 10; + gNB->dlsch[k][i]->rnti = n_rnti+k; + } + }*/ + //configure UE + UE = malloc(sizeof(PHY_VARS_NR_UE)); + memcpy(&UE->frame_parms,frame_parms,sizeof(NR_DL_FRAME_PARMS)); + + //phy_init_nr_top(frame_parms); + if (init_nr_ue_signal(UE, 1, 0) != 0) { + printf("Error at UE NR initialisation\n"); + exit(-1); + } + + //nr_init_frame_parms_ue(&UE->frame_parms); + //init_nr_ue_transport(UE, 0); + for (int sf = 0; sf < 2; sf++) { + for (i=0; i<2; i++) { + UE->dlsch[sf][0][i] = new_nr_ue_dlsch(Kmimo,8,Nsoft,5,N_RB_DL,0); + + if (!UE->dlsch[sf][0][i]) { + printf("Can't get ue dlsch structures\n"); + exit(-1); + } + + UE->dlsch[sf][0][i]->rnti = n_rnti; + } + } + + UE->dlsch_SI[0] = new_nr_ue_dlsch(1,1,Nsoft,5,N_RB_DL,0); + UE->dlsch_ra[0] = new_nr_ue_dlsch(1,1,Nsoft,5,N_RB_DL,0); + unsigned char harq_pid = 0; //dlsch->harq_ids[subframe]; + NR_gNB_DLSCH_t *dlsch = gNB->dlsch[0][0]; + nfapi_nr_dl_config_dlsch_pdu_rel15_t *rel15 = &dlsch->harq_processes[harq_pid]->dlsch_pdu.dlsch_pdu_rel15; + time_stats_t *rm_stats; + time_stats_t *te_stats; + time_stats_t *i_stats; + uint8_t is_crnti; + uint8_t llr8_flag; + unsigned int TBS = 8424; + unsigned int available_bits; + uint8_t nb_re_dmrs = 6; + uint16_t length_dmrs = 1; + unsigned char mod_order; + uint8_t Nl=1; + uint8_t rvidx = 0; + dlsch->rnti =1; + /*dlsch->harq_processes[0]->mcs = Imcs; + dlsch->harq_processes[0]->rvidx = rvidx;*/ + //printf("dlschsim harqid %d nb_rb %d, mscs %d\n",dlsch->harq_ids[subframe], + // dlsch->harq_processes[0]->nb_rb,dlsch->harq_processes[0]->mcs,dlsch->harq_processes[0]->Nl); + mod_order = nr_get_Qm(Imcs,1); + available_bits = nr_get_G(nb_rb, nb_symb_sch, nb_re_dmrs, length_dmrs,mod_order,1); + TBS= nr_compute_tbs(Imcs,nb_rb,nb_symb_sch,nb_re_dmrs,length_dmrs,Nl); + printf("available bits %d TBS %d mod_order %d\n",available_bits, TBS, mod_order); + //dlsch->harq_ids[subframe]= 0; + rel15->n_prb = nb_rb; + rel15->nb_symbols = nb_symb_sch; + rel15->modulation_order = mod_order; + rel15->nb_layers = Nl; + rel15->nb_re_dmrs = nb_re_dmrs; + rel15->transport_block_size = TBS; + double *modulated_input = malloc16(sizeof(double) * 16*68*384); + short *channel_output_fixed = malloc16(sizeof( short) *16* 68*384); + short *channel_output_uncoded = malloc16(sizeof(unsigned short) *16* 68*384); + double errors_bit_uncoded=0; + unsigned char *estimated_output; + unsigned char *estimated_output_bit; + unsigned char *test_input_bit; + unsigned int errors_bit =0; + test_input_bit = (unsigned char *) malloc16(sizeof(unsigned char) * 16*68*384); + estimated_output = (unsigned char *) malloc16(sizeof(unsigned char) * 16*68*384); + estimated_output_bit = (unsigned char *) malloc16(sizeof(unsigned char) * 16*68*384); + NR_UE_DLSCH_t *dlsch0_ue = UE->dlsch[UE->current_thread_id[subframe]][0][0]; + NR_DL_UE_HARQ_t *harq_process = dlsch0_ue->harq_processes[harq_pid]; + harq_process->mcs = Imcs; + harq_process->Nl = Nl; + harq_process->nb_rb = nb_rb; + harq_process->Qm = mod_order; + harq_process->rvidx = rvidx; + printf("harq process ue mcs = %d Qm = %d, symb %d\n", harq_process->mcs, harq_process->Qm,nb_symb_sch); + unsigned char *test_input; + test_input=(unsigned char *)malloc16(sizeof(unsigned char) * TBS/8); + + for (i=0; ib; + + /*for (int i=0; i0x%08x\n",crc24c(test_input, 32)); + + // generate signal + if (input_fd==NULL) { + nr_dlsch_encoding(test_input, + subframe, + dlsch, + frame_parms); + } + + for (SNR=snr0; SNRharq_processes[0]->f[i]); + if (dlsch->harq_processes[0]->f[i]==0) + modulated_input[i]=1.0;///sqrt(2); //QPSK + else + modulated_input[i]=-1.0;///sqrt(2); + + //if (i<16) printf("modulated_input[%d] = %d\n",i,modulated_input[i]); + //SNR =10; + SNR_lin = pow(10,SNR/10.0); + sigma = 1.0/sqrt(2*SNR_lin); + channel_output_fixed[i] = (short)quantize(sigma/4.0/4.0,modulated_input[i] + sigma*gaussdouble(0.0,1.0),qbits); + //channel_output_fixed[i] = (char)quantize8bit(sigma/4.0,(2.0*modulated_input[i]) - 1.0 + sigma*gaussdouble(0.0,1.0)); + //printf("llr[%d]=%d\n",i,channel_output_fixed[i]); + //printf("channel_output_fixed[%d]: %d\n",i,channel_output_fixed[i]); + + //channel_output_fixed[i] = (char)quantize(1,channel_output_fixed[i],qbits); + + //if (i<16) printf("channel_output_fixed[%d] = %d\n",i,channel_output_fixed[i]); + + //Uncoded BER + if (channel_output_fixed[i]<0) + channel_output_uncoded[i]=1; //QPSK demod + else + channel_output_uncoded[i]=0; + + if (channel_output_uncoded[i] != dlsch->harq_processes[harq_pid]->f[i]) + errors_bit_uncoded = errors_bit_uncoded + 1; + } + + //if (errors_bit_uncoded>10) + //printf("errors bits uncoded %f\n", errors_bit_uncoded); +#ifdef DEBUG_CODER + printf("\n"); + exit(-1); +#endif + ret = nr_dlsch_decoding(UE, + channel_output_fixed, + &UE->frame_parms, + dlsch0_ue, + dlsch0_ue->harq_processes[0], + frame, + nb_symb_sch, + subframe, + harq_pid, + is_crnti, + llr8_flag); + + if (ret>dlsch0_ue->max_ldpc_iterations) + n_errors++; + + //count errors + errors_bit = 0; + + for (i=0; i>(i&7); + test_input_bit[i] = (test_input[i/8]&(1<<(i&7)))>>(i&7); // Further correct for multiple segments + + if (estimated_output_bit[i] != test_input_bit[i]) { + errors_bit++; + //printf("estimated bits error occurs @%d ",i); + } + } + + if (errors_bit>0) { + n_false_positive++; + if (n_trials == 1) + printf("\n errors_bit %d (trial %d)\n", errors_bit,trial); + } + } + + printf("SNR %f, BLER %f (false positive %f)\n",SNR,(float)n_errors/(float)n_trials,(float)n_false_positive/(float)n_trials); + + if ((float)n_errors/(float)n_trials < target_error_rate) + break; + } + + /*LOG_M("txsigF0.m","txsF0", gNB->common_vars.txdataF[0],frame_length_complex_samples_no_prefix,1,1); + if (gNB->frame_parms.nb_antennas_tx>1) + LOG_M("txsigF1.m","txsF1", gNB->common_vars.txdataF[1],frame_length_complex_samples_no_prefix,1,1);*/ + + //TODO: loop over slots + /*for (aa=0; aaframe_parms.nb_antennas_tx; aa++) { + if (gNB_config->subframe_config.dl_cyclic_prefix_type.value == 1) { + PHY_ofdm_mod(gNB->common_vars.txdataF[aa], + txdata[aa], + frame_parms->ofdm_symbol_size, + 12, + frame_parms->nb_prefix_samples, + CYCLIC_PREFIX); + } else { + nr_normal_prefix_mod(gNB->common_vars.txdataF[aa], + txdata[aa], + 14, + frame_parms); + } + } + + LOG_M("txsig0.m","txs0", txdata[0],frame_length_complex_samples,1,1); + if (gNB->frame_parms.nb_antennas_tx>1) + LOG_M("txsig1.m","txs1", txdata[1],frame_length_complex_samples,1,1); + + + for (i=0; inb_antennas_tx; aa++) { + r_re[aa][i] = ((double)(((short *)txdata[aa]))[(i<<1)]); + r_im[aa][i] = ((double)(((short *)txdata[aa]))[(i<<1)+1]); + } + }*/ + + for (i=0; i<2; i++) { + printf("gNB %d\n",i); + free_gNB_dlsch(gNB->dlsch[0][i]); + printf("UE %d\n",i); + free_nr_ue_dlsch(UE->dlsch[UE->current_thread_id[subframe]][0][i]); + } + + for (i=0; i<2; i++) { + free(s_re[i]); + free(s_im[i]); + free(r_re[i]); + free(r_im[i]); + free(txdata[i]); + } + + free(s_re); + free(s_im); + free(r_re); + free(r_im); + free(txdata); + + if (output_fd) + fclose(output_fd); + + if (input_fd) + fclose(input_fd); + + return(n_errors); +} + + + diff --git a/openair1/SIMULATION/NR_PHY/dlsim.c b/openair1/SIMULATION/NR_PHY/dlsim.c index 943a239fce..3f717884dc 100644 --- a/openair1/SIMULATION/NR_PHY/dlsim.c +++ b/openair1/SIMULATION/NR_PHY/dlsim.c @@ -37,6 +37,7 @@ #include "PHY/defs_nr_UE.h" #include "PHY/defs_gNB.h" #include "PHY/NR_REFSIG/refsig_defs_ue.h" +#include "PHY/NR_REFSIG/nr_mod_table.h" #include "PHY/MODULATION/modulation_eNB.h" #include "PHY/MODULATION/modulation_UE.h" #include "PHY/INIT/phy_init.h" @@ -494,6 +495,7 @@ int main(int argc, char **argv) gNB_mac = RC.nrmac[0]; config_common(0,0,78,(uint64_t)3640000000L,N_RB_DL); + config_nr_mib(0,0,1,kHz30,0,0,0,0); nr_l2_init_ue(); UE_mac = get_mac_inst(0); @@ -682,10 +684,9 @@ int main(int argc, char **argv) phy_procedures_nrUE_RX(UE, &UE_proc, 0, - 0, do_pdcch_flag, - normal_txrx, - no_relay); + normal_txrx); + if (UE->dci_ind.number_of_dcis==0) n_errors++; } diff --git a/openair1/SIMULATION/NR_PHY/pbchsim.c b/openair1/SIMULATION/NR_PHY/pbchsim.c index 1869c8b3b3..9a9f882531 100644 --- a/openair1/SIMULATION/NR_PHY/pbchsim.c +++ b/openair1/SIMULATION/NR_PHY/pbchsim.c @@ -37,6 +37,7 @@ #include "PHY/defs_nr_UE.h" #include "PHY/defs_gNB.h" #include "PHY/NR_REFSIG/refsig_defs_ue.h" +#include "PHY/NR_REFSIG/nr_mod_table.h" #include "PHY/MODULATION/modulation_eNB.h" #include "PHY/MODULATION/modulation_UE.h" #include "PHY/INIT/phy_init.h" @@ -426,7 +427,7 @@ int main(int argc, char **argv) //configure UE UE = malloc(sizeof(PHY_VARS_NR_UE)); memcpy(&UE->frame_parms,frame_parms,sizeof(NR_DL_FRAME_PARMS)); - phy_init_nr_top(UE); + //phy_init_nr_top(UE); //called from init_nr_ue_signal if (run_initial_sync==1) UE->is_synchronized = 0; else UE->is_synchronized = 1; @@ -524,6 +525,7 @@ int main(int argc, char **argv) if (gNB->frame_parms.nb_antennas_tx>1) LOG_M("rxsig1.m","rxs1", UE->common_vars.rxdata[1],frame_parms->samples_per_subframe,1,1); } + if (UE->is_synchronized == 0) { ret = nr_initial_sync(UE, normal_txrx); printf("nr_initial_sync1 returns %d\n",ret); diff --git a/openair2/GNB_APP/gnb_config.c b/openair2/GNB_APP/gnb_config.c index 82f92b0a4c..4484cc18a8 100644 --- a/openair2/GNB_APP/gnb_config.c +++ b/openair2/GNB_APP/gnb_config.c @@ -629,7 +629,7 @@ void RCconfig_NRRRC(MessageDef *msg_p, uint32_t i, gNB_RRC_INST *rrc) { paramdef_t GNBParams[] = GNBPARAMS_DESC; paramlist_def_t GNBParamList = {GNB_CONFIG_STRING_GNB_LIST,NULL,0}; ////////// Physical parameters - //checkedparam_t config_check_CCparams[] = NRCCPARAMS_CHECK; + checkedparam_t config_check_CCparams[] = NRCCPARAMS_CHECK; paramdef_t CCsParams[] = NRCCPARAMS_DESC; paramlist_def_t CCsParamList = {GNB_CONFIG_STRING_COMPONENT_CARRIERS,NULL,0}; diff --git a/openair2/LAYER2/NR_MAC_gNB/gNB_scheduler.c b/openair2/LAYER2/NR_MAC_gNB/gNB_scheduler.c index 7645da92c7..d29a9fb908 100644 --- a/openair2/LAYER2/NR_MAC_gNB/gNB_scheduler.c +++ b/openair2/LAYER2/NR_MAC_gNB/gNB_scheduler.c @@ -423,7 +423,9 @@ void gNB_dlsch_ulsch_scheduler(module_id_t module_idP, } // Phytest scheduling/ option not activated because of pending bug - nr_schedule_css_dlsch_phytest(module_idP, frameP, slotP); + + if (slotP==2) + nr_schedule_css_dlsch_phytest(module_idP, frameP, slotP); /* diff --git a/openair2/LAYER2/NR_MAC_gNB/gNB_scheduler_phytest.c b/openair2/LAYER2/NR_MAC_gNB/gNB_scheduler_phytest.c index b8e5385316..131e64a93b 100644 --- a/openair2/LAYER2/NR_MAC_gNB/gNB_scheduler_phytest.c +++ b/openair2/LAYER2/NR_MAC_gNB/gNB_scheduler_phytest.c @@ -31,11 +31,13 @@ #include "nr_mac_gNB.h" #include "SCHED_NR/sched_nr.h" #include "mac_proto.h" +#include "PHY/NR_TRANSPORT/nr_dlsch.h" +#include "PHY/NR_TRANSPORT/nr_dci.h" extern RAN_CONTEXT_t RC; /*Scheduling of DLSCH with associated DCI in common search space - * current version has only a DCI for type 1 PDCCH for RA-RNTI*/ + * current version has only a DCI for type 1 PDCCH for C_RNTI*/ void nr_schedule_css_dlsch_phytest(module_id_t module_idP, frame_t frameP, sub_frame_t slotP) @@ -45,9 +47,12 @@ void nr_schedule_css_dlsch_phytest(module_id_t module_idP, gNB_MAC_INST *nr_mac = RC.nrmac[module_idP]; //NR_COMMON_channels_t *cc = nr_mac->common_channels; nfapi_nr_dl_config_request_body_t *dl_req; - nfapi_nr_dl_config_request_pdu_t *dl_config_pdu; + nfapi_nr_dl_config_request_pdu_t *dl_config_dci_pdu; + nfapi_nr_dl_config_request_pdu_t *dl_config_dlsch_pdu; nfapi_tx_request_pdu_t *TX_req; + nfapi_nr_config_request_t *cfg = &nr_mac->config[0]; + uint16_t rnti = 0x1234; uint16_t sfn_sf = frameP << 7 | slotP; int dl_carrier_bandwidth = cfg->rf_config.dl_carrier_bandwidth.value; @@ -60,13 +65,30 @@ void nr_schedule_css_dlsch_phytest(module_id_t module_idP, dl_req = &nr_mac->DL_req[CC_id].dl_config_request_body; - dl_config_pdu = &dl_req->dl_config_pdu_list[dl_req->number_pdu]; - memset((void*)dl_config_pdu,0,sizeof(nfapi_nr_dl_config_request_pdu_t)); - dl_config_pdu->pdu_type = NFAPI_NR_DL_CONFIG_DCI_DL_PDU_TYPE; - dl_config_pdu->pdu_size = (uint8_t)(2+sizeof(nfapi_nr_dl_config_dci_dl_pdu)); + dl_config_dci_pdu = &dl_req->dl_config_pdu_list[dl_req->number_pdu]; + memset((void*)dl_config_dci_pdu,0,sizeof(nfapi_nr_dl_config_request_pdu_t)); + dl_config_dci_pdu->pdu_type = NFAPI_NR_DL_CONFIG_DCI_DL_PDU_TYPE; + dl_config_dci_pdu->pdu_size = (uint8_t)(2+sizeof(nfapi_nr_dl_config_dci_dl_pdu)); - nfapi_nr_dl_config_dci_dl_pdu_rel15_t *pdu_rel15 = &dl_config_pdu->dci_dl_pdu.dci_dl_pdu_rel15; - nfapi_nr_dl_config_pdcch_parameters_rel15_t *params_rel15 = &dl_config_pdu->dci_dl_pdu.pdcch_params_rel15; + dl_config_dlsch_pdu = &dl_req->dl_config_pdu_list[dl_req->number_pdu+1]; + memset((void*)dl_config_dlsch_pdu,0,sizeof(nfapi_nr_dl_config_request_pdu_t)); + dl_config_dlsch_pdu->pdu_type = NFAPI_NR_DL_CONFIG_DLSCH_PDU_TYPE; + dl_config_dlsch_pdu->pdu_size = (uint8_t)(2+sizeof(nfapi_nr_dl_config_dlsch_pdu)); + + nfapi_nr_dl_config_dci_dl_pdu_rel15_t *pdu_rel15 = &dl_config_dci_pdu->dci_dl_pdu.dci_dl_pdu_rel15; + nfapi_nr_dl_config_pdcch_parameters_rel15_t *params_rel15 = &dl_config_dci_pdu->dci_dl_pdu.pdcch_params_rel15; + nfapi_nr_dl_config_dlsch_pdu_rel15_t *dlsch_pdu_rel15 = &dl_config_dlsch_pdu->dlsch_pdu.dlsch_pdu_rel15; + + dlsch_pdu_rel15->start_prb = 0; + dlsch_pdu_rel15->n_prb = 50; + dlsch_pdu_rel15->start_symbol = 2; + dlsch_pdu_rel15->nb_symbols = 8; + dlsch_pdu_rel15->rnti = rnti; + dlsch_pdu_rel15->nb_layers =1; + dlsch_pdu_rel15->nb_codewords = 1; + dlsch_pdu_rel15->mcs_idx = 9; + dlsch_pdu_rel15->ndi = 1; + dlsch_pdu_rel15->redundancy_version = 0; nr_configure_css_dci_initial(params_rel15, @@ -76,20 +98,33 @@ void nr_schedule_css_dlsch_phytest(module_id_t module_idP, params_rel15->first_slot = 0; - pdu_rel15->frequency_domain_assignment = 5; - pdu_rel15->time_domain_assignment = 2; - pdu_rel15->vrb_to_prb_mapping = 0; - pdu_rel15->mcs = 12; + pdu_rel15->frequency_domain_assignment = get_RIV(dlsch_pdu_rel15->start_prb, dlsch_pdu_rel15->n_prb, cfg->rf_config.dl_carrier_bandwidth.value); + pdu_rel15->time_domain_assignment = get_SLIV(dlsch_pdu_rel15->start_symbol, dlsch_pdu_rel15->nb_symbols); + pdu_rel15->vrb_to_prb_mapping = 1; + pdu_rel15->mcs = 9; pdu_rel15->tb_scaling = 1; - LOG_D(MAC, "[gNB scheduler phytest] DCI type 1 payload: freq_alloc %d, time_alloc %d, vrb to prb %d, mcs %d tb_scaling %d\n", + + pdu_rel15->ra_preamble_index = 25; + pdu_rel15->format_indicator = 1; + pdu_rel15->ndi = 1; + pdu_rel15->rv = 0; + pdu_rel15->harq_pid = 0; + pdu_rel15->dai = 2; + pdu_rel15->tpc = 2; + pdu_rel15->pucch_resource_indicator = 7; + pdu_rel15->pdsch_to_harq_feedback_timing_indicator = 7; + + LOG_I(MAC, "[gNB scheduler phytest] DCI type 1 payload: freq_alloc %d, time_alloc %d, vrb to prb %d, mcs %d tb_scaling %d ndi %d rv %d\n", pdu_rel15->frequency_domain_assignment, pdu_rel15->time_domain_assignment, pdu_rel15->vrb_to_prb_mapping, pdu_rel15->mcs, - pdu_rel15->tb_scaling); + pdu_rel15->tb_scaling, + pdu_rel15->ndi, + pdu_rel15->rv); - params_rel15->rnti = 0x03; - params_rel15->rnti_type = NFAPI_NR_RNTI_RA; + params_rel15->rnti = rnti; + params_rel15->rnti_type = NFAPI_NR_RNTI_C; params_rel15->dci_format = NFAPI_NR_DL_DCI_FORMAT_1_0; //params_rel15->aggregation_level = 1; LOG_D(MAC, "DCI type 1 params: rmsi_pdcch_config %d, rnti %d, rnti_type %d, dci_format %d\n \ @@ -108,15 +143,35 @@ void nr_schedule_css_dlsch_phytest(module_id_t module_idP, params_rel15->nb_slots, params_rel15->sfn_mod2, params_rel15->first_slot); + nr_get_tbs(&dl_config_dlsch_pdu->dlsch_pdu, dl_config_dci_pdu->dci_dl_pdu, *cfg); + LOG_I(MAC, "DLSCH PDU: start PRB %d n_PRB %d start symbol %d nb_symbols %d nb_layers %d nb_codewords %d mcs %d\n", + dlsch_pdu_rel15->start_prb, + dlsch_pdu_rel15->n_prb, + dlsch_pdu_rel15->start_symbol, + dlsch_pdu_rel15->nb_symbols, + dlsch_pdu_rel15->nb_layers, + dlsch_pdu_rel15->nb_codewords, + dlsch_pdu_rel15->mcs_idx); + dl_req->number_dci++; - dl_req->number_pdu++; + dl_req->number_pdsch_rnti++; + dl_req->number_pdu+=2; TX_req = &nr_mac->TX_req[CC_id].tx_request_body.tx_pdu_list[nr_mac->TX_req[CC_id].tx_request_body.number_of_pdus]; TX_req->pdu_length = 6; TX_req->pdu_index = nr_mac->pdu_index[CC_id]++; TX_req->num_segments = 1; TX_req->segments[0].segment_length = 8; - //TX_req->segments[0].segment_data = (uint8_t*)pdu_rel15; + nr_mac->TX_req[CC_id].tx_request_body.number_of_pdus++; + nr_mac->TX_req[CC_id].sfn_sf = sfn_sf; + nr_mac->TX_req[CC_id].tx_request_body.tl.tag = NFAPI_TX_REQUEST_BODY_TAG; + nr_mac->TX_req[CC_id].header.message_id = NFAPI_TX_REQUEST; + + TX_req = &nr_mac->TX_req[CC_id].tx_request_body.tx_pdu_list[nr_mac->TX_req[CC_id].tx_request_body.number_of_pdus+1]; + TX_req->pdu_length = dlsch_pdu_rel15->transport_block_size; + TX_req->pdu_index = nr_mac->pdu_index[CC_id]++; + TX_req->num_segments = 1; + TX_req->segments[0].segment_length = 8; nr_mac->TX_req[CC_id].tx_request_body.number_of_pdus++; nr_mac->TX_req[CC_id].sfn_sf = sfn_sf; nr_mac->TX_req[CC_id].tx_request_body.tl.tag = NFAPI_TX_REQUEST_BODY_TAG; diff --git a/targets/ARCH/COMMON/common_lib.h b/targets/ARCH/COMMON/common_lib.h index ae5f4d91d9..685a628b26 100644 --- a/targets/ARCH/COMMON/common_lib.h +++ b/targets/ARCH/COMMON/common_lib.h @@ -204,6 +204,7 @@ typedef struct { //! clock source clock_source_t clock_source; //! Manual SDR IP address +//#if defined(EXMIMO) || defined(OAI_USRP) || defined(OAI_BLADERF) || defined(OAI_LMSSDR) char *sdr_addrs; //! Auto calibration flag int autocal[4]; diff --git a/targets/PROJECTS/GENERIC-LTE-EPC/CONF/gnb.band78.tm1.106PRB.usrpx300.conf b/targets/PROJECTS/GENERIC-LTE-EPC/CONF/gnb.band78.tm1.106PRB.usrpx300.conf index 721c307f71..d62b06aaa5 100644 --- a/targets/PROJECTS/GENERIC-LTE-EPC/CONF/gnb.band78.tm1.106PRB.usrpx300.conf +++ b/targets/PROJECTS/GENERIC-LTE-EPC/CONF/gnb.band78.tm1.106PRB.usrpx300.conf @@ -262,7 +262,7 @@ RUs = ( THREAD_STRUCT = ( { #three config for level of parallelism "PARALLEL_SINGLE_THREAD", "PARALLEL_RU_L1_SPLIT", or "PARALLEL_RU_L1_TRX_SPLIT" - parallel_config = "PARALLEL_RU_L1_TRX_SPLIT"; + parallel_config = "PARALLEL_SINGLE_THREAD"; #two option for worker "WORKER_DISABLE" or "WORKER_ENABLE" worker_config = "WORKER_DISABLE"; } diff --git a/targets/RT/USER/gNB_usrp.gtkw b/targets/RT/USER/gNB_usrp.gtkw index f7ca2c291a..b40f946bc1 100644 --- a/targets/RT/USER/gNB_usrp.gtkw +++ b/targets/RT/USER/gNB_usrp.gtkw @@ -24,9 +24,9 @@ variables.trx_tst[63:0] functions.eNB_thread_rxtx0 @24 variables.frame_number_RX0_RU[63:0] -variables.subframe_number_RX0_RU[63:0] +variables.tti_number_RX0_RU[63:0] variables.frame_number_TX0_RU[63:0] -variables.subframe_number_TX0_RU[63:0] +variables.tti_number_TX0_RU[63:0] @28 functions.mac_schedule_dlsch functions.macxface_eNB_dlsch_ulsch_scheduler @@ -34,16 +34,16 @@ functions.macxface_ue_scheduler functions.phy_eNB_ofdm_mod_l @24 variables.frame_number_RX0_gNB[63:0] -variables.subframe_number_RX0_gNB[63:0] +variables.slot_number_RX0_gNB[63:0] variables.frame_number_TX0_gNB[63:0] -variables.subframe_number_TX0_gNB[63:0] +variables.slot_number_TX0_gNB[63:0] @28 functions.gNB_thread_rxtx0 @24 variables.frame_number_RX1_gNB[63:0] -variables.subframe_number_RX1_gNB[63:0] +variables.slot_number_RX1_gNB[63:0] variables.frame_number_TX1_gNB[63:0] -variables.subframe_number_TX1_gNB[63:0] +variables.slot_number_TX1_gNB[63:0] @28 functions.gNB_thread_rxtx1 functions.phy_procedures_ru_feprx0 diff --git a/targets/RT/USER/nr-gnb.c b/targets/RT/USER/nr-gnb.c index 9d20ff9f02..5d8c98d645 100644 --- a/targets/RT/USER/nr-gnb.c +++ b/targets/RT/USER/nr-gnb.c @@ -904,87 +904,7 @@ void print_opp_meas(void) { print_meas(&softmodem_stats_rx_sf,"[gNB][total_phy_proc_rx]",NULL,NULL); } } -/* -void free_transport(PHY_VARS_gNB *gNB) -{ - int i; - int j; - for (i=0; idlsch[i][j]); - - LOG_I(PHY, "Freeing Transport Channel Buffer for ULSCH, UE %d\n",i); - free_gNB_ulsch(gNB->ulsch[1+i]); - } - free_gNB_ulsch(gNB->ulsch[0]); -}*/ - -/* -void init_transport(PHY_VARS_gNB *gNB) { - - int i; - int j; - NR_DL_FRAME_PARMS *fp = &gNB->frame_parms; - - LOG_I(PHY, "Initialise transport\n"); - - for (i=0; idlsch[i][j] = new_gNB_dlsch(1,8,NSOFT,fp->N_RB_DL,0,fp); - if (!gNB->dlsch[i][j]) { - LOG_E(PHY,"Can't get gNB dlsch structures for UE %d \n", i); - exit(-1); - } else { - gNB->dlsch[i][j]->rnti=0; - LOG_D(PHY,"dlsch[%d][%d] => %p rnti:%d\n",i,j,gNB->dlsch[i][j], gNB->dlsch[i][j]->rnti); - } - } - - LOG_I(PHY,"Allocating Transport Channel Buffer for ULSCH, UE %d\n",i); - gNB->ulsch[1+i] = new_gNB_ulsch(MAX_TURBO_ITERATIONS,fp->N_RB_UL, 0); - - if (!gNB->ulsch[1+i]) { - LOG_E(PHY,"Can't get gNB ulsch structures\n"); - exit(-1); - } - - // this is the transmission mode for the signalling channels - // this will be overwritten with the real transmission mode by the RRC once the UE is connected - gNB->transmission_mode[i] = fp->nb_antenna_ports_gNB==1 ? 1 : 2; - } - // ULSCH for RA - gNB->ulsch[0] = new_gNB_ulsch(MAX_TURBO_ITERATIONS, fp->N_RB_UL, 0); - - if (!gNB->ulsch[0]) { - LOG_E(PHY,"Can't get gNB ulsch structures\n"); - exit(-1); - } - gNB->dlsch_SI = new_gNB_dlsch(1,8,NSOFT,fp->N_RB_DL, 0, fp); - LOG_D(PHY,"gNB %d.%d : SI %p\n",gNB->Mod_id,gNB->CC_id,gNB->dlsch_SI); - gNB->dlsch_ra = new_gNB_dlsch(1,8,NSOFT,fp->N_RB_DL, 0, fp); - LOG_D(PHY,"gNB %d.%d : RA %p\n",gNB->Mod_id,gNB->CC_id,gNB->dlsch_ra); - gNB->dlsch_MCH = new_gNB_dlsch(1,8,NSOFT,fp->N_RB_DL, 0, fp); - LOG_D(PHY,"gNB %d.%d : MCH %p\n",gNB->Mod_id,gNB->CC_id,gNB->dlsch_MCH); - - - gNB->rx_total_gain_dB=130; - - for(i=0; imu_mimo_mode[i].dl_pow_off = 2; - - gNB->check_for_total_transmissions = 0; - - gNB->check_for_MUMIMO_transmissions = 0; - - gNB->FULL_MUMIMO_transmissions = 0; - - gNB->check_for_SUMIMO_transmissions = 0; - - fp->pucch_config_common.deltaPUCCH_Shift = 1; - -} */ /// eNB kept in function name for nffapi calls, TO FIX void init_eNB_afterRU(void) { @@ -1035,8 +955,7 @@ void init_eNB_afterRU(void) { * In monolithic mode, we come here with nb_antennas_rx == 0 * (not tested in other modes). */ -/// Transport init necessary for NR synchro - //init_transport(gNB); + //init_precoding_weights(RC.gNB[inst][CC_id]); } init_gNB_proc(inst); diff --git a/targets/RT/USER/nr-ru.c b/targets/RT/USER/nr-ru.c index 388cba579c..73d7c5dae1 100644 --- a/targets/RT/USER/nr-ru.c +++ b/targets/RT/USER/nr-ru.c @@ -764,6 +764,7 @@ void tx_rf(RU_t *ru) { (SF_type == SF_S)) { int siglen=fp->samples_per_slot,flags=1; + /* if (SF_type == SF_S) { siglen = fp->dl_symbols_in_S_subframe*(fp->ofdm_symbol_size+fp->nb_prefix_samples0); @@ -788,13 +789,9 @@ void tx_rf(RU_t *ru) { VCD_SIGNAL_DUMPER_DUMP_VARIABLE_BY_NAME( VCD_SIGNAL_DUMPER_VARIABLES_TTI_NUMBER_TX0_RU, proc->tti_tx ); for (i=0; inb_tx; i++) + txp[i] = (void*)&ru->common.txdata[i][(proc->tti_tx*fp->samples_per_slot)-sf_extension]; -/* - if (proc->tti_tx == 0){ - write_output("txdataF_frame.m","txdataF_frame",&ru->common.txdataF_BF[i],fp->samples_per_subframe_wCP, 1, 1); - write_output("txdata_frame.m","txdata_frame",&ru->common.txdata[i],fp->samples_per_subframe, 1, 1); - } -*/ + VCD_SIGNAL_DUMPER_DUMP_VARIABLE_BY_NAME( VCD_SIGNAL_DUMPER_VARIABLES_TRX_TST, (proc->timestamp_tx-ru->openair0_cfg.tx_sample_advance)&0xffffffff ); VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME( VCD_SIGNAL_DUMPER_FUNCTIONS_TRX_WRITE, 1 ); // prepare tx buffer pointers diff --git a/targets/RT/USER/nr-softmodem.c b/targets/RT/USER/nr-softmodem.c index d6950fe24b..0c29c90f62 100644 --- a/targets/RT/USER/nr-softmodem.c +++ b/targets/RT/USER/nr-softmodem.c @@ -103,6 +103,9 @@ static pthread_t forms_thread; //xforms #endif //XFORMS +short nr_mod_table[NR_MOD_TABLE_SIZE_SHORT] = {0,0,16384,16384,-16384,-16384,16384,16384,16384,-16384,-16384,16384,-16384,-16384,7327,7327,7327,21981,21981,7327,21981,21981,7327,-7327,7327,-21981,21981,-7327,21981,-21981,-7327,7327,-7327,21981,-21981,7327,-21981,21981,-7327,-7327,-7327,-21981,-21981,-7327,-21981,-21981,10726,10726,10726,3576,3576,10726,3576,3576,10726,17876,10726,25027,3576,17876,3576,25027,17876,10726,17876,3576,25027,10726,25027,3576,17876,17876,17876,25027,25027,17876,25027,25027,10726,-10726,10726,-3576,3576,-10726,3576,-3576,10726,-17876,10726,-25027,3576,-17876,3576,-25027,17876,-10726,17876,-3576,25027,-10726,25027,-3576,17876,-17876,17876,-25027,25027,-17876,25027,-25027,-10726,10726,-10726,3576,-3576,10726,-3576,3576,-10726,17876,-10726,25027,-3576,17876,-3576,25027,-17876,10726,-17876,3576,-25027,10726,-25027,3576,-17876,17876,-17876,25027,-25027,17876,-25027,25027,-10726,-10726,-10726,-3576,-3576,-10726,-3576,-3576,-10726,-17876,-10726,-25027,-3576,-17876,-3576,-25027,-17876,-10726,-17876,-3576,-25027,-10726,-25027,-3576,-17876,-17876,-17876,-25027,-25027,-17876,-25027,-25027,8886,8886,8886,12439,12439,8886,12439,12439,8886,5332,8886,1778,12439,5332,12439,1778,5332,8886,5332,12439,1778,8886,1778,12439,5332,5332,5332,1778,1778,5332,1778,1778,8886,19547,8886,15993,12439,19547,12439,15993,8886,23101,8886,26655,12439,23101,12439,26655,5332,19547,5332,15993,1778,19547,1778,15993,5332,23101,5332,26655,1778,23101,1778,26655,19547,8886,19547,12439,15993,8886,15993,12439,19547,5332,19547,1778,15993,5332,15993,1778,23101,8886,23101,12439,26655,8886,26655,12439,23101,5332,23101,1778,26655,5332,26655,1778,19547,19547,19547,15993,15993,19547,15993,15993,19547,23101,19547,26655,15993,23101,15993,26655,23101,19547,23101,15993,26655,19547,26655,15993,23101,23101,23101,26655,26655,23101,26655,26655,8886,-8886,8886,-12439,12439,-8886,12439,-12439,8886,-5332,8886,-1778,12439,-5332,12439,-1778,5332,-8886,5332,-12439,1778,-8886,1778,-12439,5332,-5332,5332,-1778,1778,-5332,1778,-1778,8886,-19547,8886,-15993,12439,-19547,12439,-15993,8886,-23101,8886,-26655,12439,-23101,12439,-26655,5332,-19547,5332,-15993,1778,-19547,1778,-15993,5332,-23101,5332,-26655,1778,-23101,1778,-26655,19547,-8886,19547,-12439,15993,-8886,15993,-12439,19547,-5332,19547,-1778,15993,-5332,15993,-1778,23101,-8886,23101,-12439,26655,-8886,26655,-12439,23101,-5332,23101,-1778,26655,-5332,26655,-1778,19547,-19547,19547,-15993,15993,-19547,15993,-15993,19547,-23101,19547,-26655,15993,-23101,15993,-26655,23101,-19547,23101,-15993,26655,-19547,26655,-15993,23101,-23101,23101,-26655,26655,-23101,26655,-26655,-8886,8886,-8886,12439,-12439,8886,-12439,12439,-8886,5332,-8886,1778,-12439,5332,-12439,1778,-5332,8886,-5332,12439,-1778,8886,-1778,12439,-5332,5332,-5332,1778,-1778,5332,-1778,1778,-8886,19547,-8886,15993,-12439,19547,-12439,15993,-8886,23101,-8886,26655,-12439,23101,-12439,26655,-5332,19547,-5332,15993,-1778,19547,-1778,15993,-5332,23101,-5332,26655,-1778,23101,-1778,26655,-19547,8886,-19547,12439,-15993,8886,-15993,12439,-19547,5332,-19547,1778,-15993,5332,-15993,1778,-23101,8886,-23101,12439,-26655,8886,-26655,12439,-23101,5332,-23101,1778,-26655,5332,-26655,1778,-19547,19547,-19547,15993,-15993,19547,-15993,15993,-19547,23101,-19547,26655,-15993,23101,-15993,26655,-23101,19547,-23101,15993,-26655,19547,-26655,15993,-23101,23101,-23101,26655,-26655,23101,-26655,26655,-8886,-8886,-8886,-12439,-12439,-8886,-12439,-12439,-8886,-5332,-8886,-1778,-12439,-5332,-12439,-1778,-5332,-8886,-5332,-12439,-1778,-8886,-1778,-12439,-5332,-5332,-5332,-1778,-1778,-5332,-1778,-1778,-8886,-19547,-8886,-15993,-12439,-19547,-12439,-15993,-8886,-23101,-8886,-26655,-12439,-23101,-12439,-26655,-5332,-19547,-5332,-15993,-1778,-19547,-1778,-15993,-5332,-23101,-5332,-26655,-1778,-23101,-1778,-26655,-19547,-8886,-19547,-12439,-15993,-8886,-15993,-12439,-19547,-5332,-19547,-1778,-15993,-5332,-15993,-1778,-23101,-8886,-23101,-12439,-26655,-8886,-26655,-12439,-23101,-5332,-23101,-1778,-26655,-5332,-26655,-1778,-19547,-19547,-19547,-15993,-15993,-19547,-15993,-15993,-19547,-23101,-19547,-26655,-15993,-23101,-15993,-26655,-23101,-19547,-23101,-15993,-26655,-19547,-26655,-15993,-23101,-23101,-23101,-26655,-26655,-23101,-26655,-26655}; + + pthread_cond_t nfapi_sync_cond; pthread_mutex_t nfapi_sync_mutex; int nfapi_sync_var=-1; //!< protected by mutex \ref nfapi_sync_mutex @@ -384,11 +387,11 @@ void reset_stats(FL_OBJECT *button, long arg) for (i=0; idlsch[i][0]->Mlimit; j++) { +/* for (j=0; jdlsch[i][0]->Mlimit; j++) { phy_vars_gNB->UE_stats[i].dlsch_NAK[k][j]=0; phy_vars_gNB->UE_stats[i].dlsch_ACK[k][j]=0; phy_vars_gNB->UE_stats[i].dlsch_trials[k][j]=0; - } + }*/ phy_vars_gNB->UE_stats[i].dlsch_l2_errors[k]=0; phy_vars_gNB->UE_stats[i].ulsch_errors[k]=0; diff --git a/targets/RT/USER/nr-ue.c b/targets/RT/USER/nr-ue.c index f264d8ff3a..dabb5e140a 100644 --- a/targets/RT/USER/nr-ue.c +++ b/targets/RT/USER/nr-ue.c @@ -226,7 +226,7 @@ PHY_VARS_NR_UE* init_nr_ue_vars(NR_DL_FRAME_PARMS *frame_parms, // initialize all signal buffers init_nr_ue_signal(ue,1,abstraction_flag); // intialize transport - //init_nr_ue_transport(ue,abstraction_flag); + init_nr_ue_transport(ue,abstraction_flag); return(ue); } @@ -442,6 +442,8 @@ static void *UE_thread_synch(void *arg) { #endif if (nr_initial_sync( UE, UE->mode ) == 0) { + //write_output("txdata_sym.m", "txdata_sym", UE->common_vars.rxdata[0], (10*UE->frame_parms.samples_per_subframe), 1, 1); + hw_slot_offset = (UE->rx_offset<<1) / UE->frame_parms.samples_per_subframe; printf("Got synch: hw_slot_offset %d, carrier off %d Hz, rxgain %d (DL %u, UL %u), UE_scan_carrier %d\n", hw_slot_offset, @@ -594,7 +596,7 @@ static void *UE_thread_synch(void *arg) { break; } -#ifdef XFORMS +#if 0 //defined XFORMS if (do_forms) { extern FD_lte_phy_scope_ue *form_ue[NUMBER_OF_UE_MAX]; @@ -603,7 +605,7 @@ static void *UE_thread_synch(void *arg) { 0,0,7); } #endif - + AssertFatal ( 0== pthread_mutex_lock(&UE->proc.mutex_synch), ""); // indicate readiness UE->proc.instance_cnt_synch--; @@ -688,10 +690,10 @@ static void *UE_thread_rxn_txnp4(void *arg) { UE->if_inst->dcireq(&UE->dcireq); #ifdef UE_SLOT_PARALLELISATION - phy_procedures_slot_parallelization_UE_RX( UE, proc, 0, 0, 1, UE->mode, no_relay, NULL ); + phy_procedures_slot_parallelization_nrUE_RX( UE, proc, 0, 0, 1, UE->mode, no_relay, NULL ); #else phy_procedures_nrUE_RX( UE, proc, 0, 1, UE->mode); - printf(">>> nr_ue_pdcch_procedures ended\n"); + // printf(">>> nr_ue_pdcch_procedures ended\n"); #endif } @@ -699,7 +701,7 @@ static void *UE_thread_rxn_txnp4(void *arg) { #if UE_TIMING_TRACE start_meas(&UE->generic_stat); #endif -printf(">>> mac init\n"); + //printf(">>> mac init\n"); if (UE->mac_enabled==1) { @@ -770,7 +772,7 @@ printf(">>> mac init\n"); #if UE_TIMING_TRACE stop_meas(&UE->generic_stat); #endif -printf(">>> mac ended\n"); + //printf(">>> mac ended\n"); // Prepare the future Tx data #if 0 @@ -874,13 +876,14 @@ void *UE_thread(void *arg) { usleep(500*1000); } #endif + if (instance_cnt_synch < 0) { // we can invoke the synch // grab 10 ms of signal and wakeup synch thread for (int i=0; iframe_parms.nb_antennas_rx; i++) rxp[i] = (void*)&UE->common_vars.rxdata[i][0]; if (UE->mode != loop_through_memory) - AssertFatal( UE->frame_parms.samples_per_subframe*10 == + AssertFatal( UE->frame_parms.samples_per_subframe*10== UE->rfdevice.trx_read_func(&UE->rfdevice, ×tamp, rxp, @@ -933,10 +936,10 @@ void *UE_thread(void *arg) { } UE->rx_offset=0; UE->time_sync_cell=0; - //UE->proc.proc_rxtx[0].frame_rx++; + UE->proc.proc_rxtx[0].frame_rx++; //UE->proc.proc_rxtx[1].frame_rx++; - for (th_id=0; th_id < RX_NB_TH; th_id++) { - UE->proc.proc_rxtx[th_id].frame_rx++; + for (th_id=1; th_id < RX_NB_TH; th_id++) { + UE->proc.proc_rxtx[th_id].frame_rx = UE->proc.proc_rxtx[0].frame_rx; } //printf("first stream frame rx %d\n",UE->proc.proc_rxtx[0].frame_rx); @@ -955,6 +958,10 @@ void *UE_thread(void *arg) { rt_sleep_ns(1000*1000); } else { + thread_idx++; + if(thread_idx>=RX_NB_TH) + thread_idx = 0; + subframe_nr++; subframe_nr %= NR_NUMBER_OF_SUBFRAMES_PER_FRAME; UE_nr_rxtx_proc_t *proc = &UE->proc.proc_rxtx[thread_idx]; @@ -975,9 +982,6 @@ void *UE_thread(void *arg) { LOG_D(PHY,"Process subframe %d thread Idx %d \n", subframe_nr, UE->current_thread_id[subframe_nr]); - thread_idx++; - if(thread_idx>=RX_NB_TH) - thread_idx = 0; if (UE->mode != loop_through_memory) { for (i=0; iframe_parms.nb_antennas_rx; i++) @@ -1046,10 +1050,10 @@ void *UE_thread(void *arg) { // operate on thread sf mod 2 AssertFatal(pthread_mutex_lock(&proc->mutex_rxtx) ==0,""); if(subframe_nr == 0) { - //UE->proc.proc_rxtx[0].frame_rx++; + UE->proc.proc_rxtx[0].frame_rx++; //UE->proc.proc_rxtx[1].frame_rx++; - for (th_id=0; th_id < RX_NB_TH; th_id++) { - UE->proc.proc_rxtx[th_id].frame_rx++; + for (th_id=1; th_id < RX_NB_TH; th_id++) { + UE->proc.proc_rxtx[th_id].frame_rx = UE->proc.proc_rxtx[0].frame_rx; } #ifdef SAIF_ENABLED if (!(proc->frame_rx%4000)) @@ -1069,7 +1073,7 @@ void *UE_thread(void *arg) { proc->nr_tti_rx=subframe_nr; proc->subframe_rx=subframe_nr; - + proc->frame_tx = proc->frame_rx; proc->nr_tti_tx= subframe_nr + DURATION_RX_TO_TX; if (proc->nr_tti_tx > NR_NUMBER_OF_SUBFRAMES_PER_FRAME) { @@ -1194,11 +1198,11 @@ void init_UE_threads(PHY_VARS_NR_UE *UE) { #ifdef UE_DLSCH_PARALLELISATION pthread_mutex_init(&UE->proc.proc_rxtx[i].mutex_dlsch_td,NULL); pthread_cond_init(&UE->proc.proc_rxtx[i].cond_dlsch_td,NULL); - pthread_create(&UE->proc.proc_rxtx[i].pthread_dlsch_td,NULL,dlsch_decoding_2thread0, rtd); + pthread_create(&UE->proc.proc_rxtx[i].pthread_dlsch_td,NULL,nr_dlsch_decoding_2thread0, rtd); //thread 2 pthread_mutex_init(&UE->proc.proc_rxtx[i].mutex_dlsch_td1,NULL); pthread_cond_init(&UE->proc.proc_rxtx[i].cond_dlsch_td1,NULL); - pthread_create(&UE->proc.proc_rxtx[i].pthread_dlsch_td1,NULL,dlsch_decoding_2thread1, rtd); + pthread_create(&UE->proc.proc_rxtx[i].pthread_dlsch_td1,NULL,nr_dlsch_decoding_2thread1, rtd); #endif #ifdef UE_SLOT_PARALLELISATION diff --git a/targets/RT/USER/nr-uesoftmodem.c b/targets/RT/USER/nr-uesoftmodem.c index 02db7d1a02..68f0254762 100644 --- a/targets/RT/USER/nr-uesoftmodem.c +++ b/targets/RT/USER/nr-uesoftmodem.c @@ -85,10 +85,10 @@ unsigned short config_frames[4] = {2,9,11,13}; // current status is that every UE has a DL scope for a SINGLE eNB (eNB_id=0) // at eNB 0, an UL scope for every UE FD_lte_phy_scope_ue *form_ue[NUMBER_OF_UE_MAX]; -FD_lte_phy_scope_enb *form_enb[MAX_NUM_CCs][NUMBER_OF_UE_MAX]; -FD_stats_form *form_stats=NULL,*form_stats_l2=NULL; +//FD_lte_phy_scope_enb *form_enb[MAX_NUM_CCs][NUMBER_OF_UE_MAX]; +//FD_stats_form *form_stats=NULL,*form_stats_l2=NULL; char title[255]; -unsigned char scope_enb_num_ue = 2; +//unsigned char scope_enb_num_ue = 2; static pthread_t forms_thread; //xforms #endif //XFORMS #include "nr-uesoftmodem.h" @@ -119,6 +119,8 @@ int threequarter_fs=0; uint32_t downlink_frequency[MAX_NUM_CCs][4]; int32_t uplink_frequency_offset[MAX_NUM_CCs][4]; +extern int16_t nr_dlsch_demod_shift; + //static char *conf_config_file_name = NULL; @@ -435,21 +437,21 @@ static void *scope_thread(void *arg) { #endif - while (!oai_exit) { + while (!oai_exit) { //len = dump_ue_stats (PHY_vars_UE_g[0][0], &PHY_vars_UE_g[0][0]->proc.proc_rxtx[0],stats_buffer, 0, mode,rx_input_level_dBm); //fl_set_object_label(form_stats->stats_text, stats_buffer); - fl_clear_browser(form_stats->stats_text); - fl_add_browser_line(form_stats->stats_text, stats_buffer); + //fl_clear_browser(form_stats->stats_text); + //fl_add_browser_line(form_stats->stats_text, stats_buffer); - if (PHY_vars_UE_g[0][0]->is_synchronized == 1) + //if (PHY_vars_UE_g[0][0]->is_synchronized == 1) phy_scope_UE(form_ue[0], PHY_vars_UE_g[0][0], - 0,0,7); + 0,0,1); //else it is done in the synch thread //printf("doing forms\n"); //usleep(100000); // 100 ms - sleep(1); + sleep(0.1); } // printf("%s",stats_buffer); @@ -547,7 +549,7 @@ static void get_options(void) { set_glog_onlinelog(online_log_messages); } if(config_isparamset(cmdline_logparams,CMDLINE_GLOGLEVEL_IDX)) { - set_glog(glog_level); + set_glog(glog_level); } if (start_telnetsrv) { @@ -872,23 +874,6 @@ int main( int argc, char **argv ) { T_Config_Init(); #endif - // initialize the log (see log.h for details) - set_glog(OAILOG_DEBUG); - - set_log(HW, OAILOG_DEBUG); - set_log(PHY, OAILOG_DEBUG); - set_log(MAC, OAILOG_INFO); - set_log(RLC, OAILOG_INFO); - set_log(PDCP, OAILOG_INFO); - set_log(OTG, OAILOG_INFO); - set_log(RRC, OAILOG_INFO); -#if defined(ENABLE_ITTI) - set_log(SIM, OAILOG_INFO); -# if defined(ENABLE_USE_MME) - set_log(NAS, OAILOG_INFO); -# endif -#endif - //randominit (0); set_taus_seed (0); @@ -1131,6 +1116,8 @@ int main( int argc, char **argv ) { pthread_cond_init(&sync_cond,NULL); pthread_mutex_init(&sync_mutex, NULL); + init_UE(1); + #ifdef XFORMS int UE_id; int fl_argc=1; @@ -1140,8 +1127,8 @@ int main( int argc, char **argv ) { fl_initialize (&fl_argc, argv, NULL, 0, 0); // restore the original command line args // argv = fl_get_cmdline_args( &argc ); - form_stats = create_form_stats_form(); - fl_show_form (form_stats->stats_form, FL_PLACE_HOTSPOT, FL_FULLBORDER, "stats"); + //form_stats = create_form_stats_form(); + //fl_show_form (form_stats->stats_form, FL_PLACE_HOTSPOT, FL_FULLBORDER, "stats"); UE_id = 0; form_ue[UE_id] = create_lte_phy_scope_ue(); sprintf (title, "NR DL SCOPE UE"); @@ -1171,7 +1158,6 @@ int main( int argc, char **argv ) { rt_sleep_ns(10*100000000ULL); - init_UE(1); number_of_cards = 1; for(CC_id=0; CC_idstats_form); - fl_free_form(form_stats->stats_form); + //fl_hide_form(form_stats->stats_form); + //fl_free_form(form_stats->stats_form); fl_hide_form(form_ue[0]->lte_phy_scope_ue); fl_free_form(form_ue[0]->lte_phy_scope_ue); } diff --git a/targets/RT/USER/nr-uesoftmodem.h b/targets/RT/USER/nr-uesoftmodem.h index ff3ec107de..99fb64f175 100644 --- a/targets/RT/USER/nr-uesoftmodem.h +++ b/targets/RT/USER/nr-uesoftmodem.h @@ -163,7 +163,7 @@ {"threadThreeSubframe", NULL, 0, iptr:&(threads.three), defintval:1, TYPE_INT, 0}, \ {"threadSlot1ProcOne", NULL, 0, iptr:&(threads.slot1_proc_one), defintval:1, TYPE_INT, 0}, \ {"threadSlot1ProcTwo", NULL, 0, iptr:&(threads.slot1_proc_two), defintval:1, TYPE_INT, 0}, \ -{"dlsch-demod-shift", CONFIG_HLP_DLSHIFT, 0, iptr:(int32_t *)&dlsch_demod_shift, defintval:0, TYPE_INT, 0}, \ +{"nr_dlsch-demod-shift", CONFIG_HLP_DLSHIFT, 0, iptr:(int32_t *)&nr_dlsch_demod_shift, defintval:0, TYPE_INT, 0}, \ {"A" , CONFIG_HLP_TADV, 0, uptr:&timing_advance, defintval:0, TYPE_UINT, 0}, \ {"C" , CONFIG_HLP_DLF, 0, uptr:&(downlink_frequency[0][0]), defuintval:2680000000, TYPE_UINT, 0}, \ {"a" , CONFIG_HLP_CHOFF, 0, iptr:&chain_offset, defintval:0, TYPE_INT, 0}, \