diff --git a/CMakeLists.txt b/CMakeLists.txt index 214a16b79a..2c1b7d6421 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -1071,6 +1071,7 @@ set(PHY_SRC_COMMON ${OPENAIR1_DIR}/PHY/TOOLS/dB_routines.c ${OPENAIR1_DIR}/PHY/TOOLS/sqrt.c ${OPENAIR1_DIR}/PHY/TOOLS/lut.c + ${OPENAIR1_DIR}/PHY/TOOLS/simde_operations.c ) set(PHY_SRC @@ -1206,6 +1207,7 @@ set(PHY_SRC_UE ${OPENAIR1_DIR}/PHY/TOOLS/dB_routines.c ${OPENAIR1_DIR}/PHY/TOOLS/sqrt.c ${OPENAIR1_DIR}/PHY/TOOLS/lut.c + ${OPENAIR1_DIR}/PHY/TOOLS/simde_operations.c ${PHY_POLARSRC} ${PHY_SMALLBLOCKSRC} ${PHY_NR_CODINGIF} @@ -1258,6 +1260,7 @@ set(PHY_SRC_UE ${OPENAIR1_DIR}/PHY/TOOLS/dB_routines.c ${OPENAIR1_DIR}/PHY/TOOLS/sqrt.c ${OPENAIR1_DIR}/PHY/TOOLS/lut.c + ${OPENAIR1_DIR}/PHY/TOOLS/simde_operations.c ${OPENAIR1_DIR}/PHY/INIT/nr_init_ue.c # ${OPENAIR1_DIR}/SIMULATION/NR_UE_PHY/unit_tests/src/pucch_uci_test.c ${PHY_POLARSRC} @@ -1306,8 +1309,9 @@ set(PHY_MEX_UE ${OPENAIR1_DIR}/PHY/TOOLS/cadd_vv.c ${OPENAIR1_DIR}/PHY/TOOLS/cmult_sv.c ${OPENAIR1_DIR}/PHY/TOOLS/cmult_vv.c - ${OPENAIR1_DIR}/PHY/LTE_UE_TRANSPORT/dlsch_llr_computation_avx2.c ${OPENAIR1_DIR}/PHY/TOOLS/signal_energy.c + ${OPENAIR1_DIR}/PHY/TOOLS/simde_operations.c + ${OPENAIR1_DIR}/PHY/LTE_UE_TRANSPORT/dlsch_llr_computation_avx2.c ${OPENAIR1_DIR}/PHY/LTE_ESTIMATION/lte_ue_measurements.c ${OPENAIR_DIR}/common/utils/LOG/log.c ${OPENAIR_DIR}/common/utils/T/T.c diff --git a/doc/tutorial_resources/oai-cn5g/docker-compose.yml b/doc/tutorial_resources/oai-cn5g/docker-compose.yml index b57942fea0..fc9eab0ad5 100644 --- a/doc/tutorial_resources/oai-cn5g/docker-compose.yml +++ b/doc/tutorial_resources/oai-cn5g/docker-compose.yml @@ -36,7 +36,7 @@ services: ipv4_address: 192.168.70.139 oai-udr: container_name: "oai-udr" - image: oaisoftwarealliance/oai-udr:develop + image: oaisoftwarealliance/oai-udr:v1.5.1 environment: - TZ=Europe/Paris - UDR_NAME=OAI_UDR @@ -58,7 +58,7 @@ services: ipv4_address: 192.168.70.136 oai-udm: container_name: "oai-udm" - image: oaisoftwarealliance/oai-udm:develop + image: oaisoftwarealliance/oai-udm:v1.5.1 environment: - TZ=Europe/Paris - UDM_NAME=OAI_UDM @@ -77,7 +77,7 @@ services: ipv4_address: 192.168.70.137 oai-ausf: container_name: "oai-ausf" - image: oaisoftwarealliance/oai-ausf:develop + image: oaisoftwarealliance/oai-ausf:v1.5.1 environment: - TZ=Europe/Paris - AUSF_NAME=OAI_AUSF @@ -96,7 +96,7 @@ services: ipv4_address: 192.168.70.138 oai-nrf: container_name: "oai-nrf" - image: oaisoftwarealliance/oai-nrf:develop + image: oaisoftwarealliance/oai-nrf:v1.5.1 environment: - TZ=Europe/Paris - NRF_INTERFACE_NAME_FOR_SBI=eth0 @@ -105,7 +105,7 @@ services: ipv4_address: 192.168.70.130 oai-amf: container_name: "oai-amf" - image: oaisoftwarealliance/oai-amf:develop + image: oaisoftwarealliance/oai-amf:v1.5.1 environment: - TZ=Europe/Paris # PLMN informations @@ -151,7 +151,7 @@ services: ipv4_address: 192.168.70.132 oai-smf: container_name: "oai-smf" - image: oaisoftwarealliance/oai-smf:develop + image: oaisoftwarealliance/oai-smf:v1.5.1 environment: - TZ=Europe/Paris - SMF_INTERFACE_NAME_FOR_N4=eth0 @@ -202,7 +202,7 @@ services: ipv4_address: 192.168.70.133 oai-spgwu-tiny: container_name: "oai-spgwu-tiny" - image: oaisoftwarealliance/oai-spgwu-tiny:develop + image: oaisoftwarealliance/oai-spgwu-tiny:v1.5.1 environment: - TZ=Europe/Paris - SGW_INTERFACE_NAME_FOR_S1U_S12_S4_UP=eth0 diff --git a/executables/nr-uesoftmodem.c b/executables/nr-uesoftmodem.c index 7896e57145..f2420641fd 100644 --- a/executables/nr-uesoftmodem.c +++ b/executables/nr-uesoftmodem.c @@ -513,7 +513,7 @@ int main( int argc, char **argv ) { NR_UE_MAC_INST_t *mac = get_mac_inst(0); if (get_softmodem_params()->sa) { // set frame config to initial values from command line and assume that the SSB is centered on the grid - uint16_t nr_band = get_band(downlink_frequency[CC_id][0],uplink_frequency_offset[CC_id][0]); + uint16_t nr_band = get_softmodem_params()->band; mac->nr_band = nr_band; nr_init_frame_parms_ue_sa(&UE[CC_id]->frame_parms, downlink_frequency[CC_id][0], diff --git a/executables/softmodem-common.h b/executables/softmodem-common.h index 1f37c9bdea..aa26dd1afc 100644 --- a/executables/softmodem-common.h +++ b/executables/softmodem-common.h @@ -58,7 +58,7 @@ extern "C" #define CONFIG_HLP_PHYTST "test UE phy layer, mac disabled\n" #define CONFIG_HLP_DORA "test gNB and UE with RA procedures\n" #define CONFIG_HLP_SA "run gNB in standalone mode\n" -#define CONFIG_HLP_SL_MODE "sets the NR sidelink mode (0: not in sidelink mode, 1: in-coverage/gNB, 2: out-of-coverage/no gNB)" +#define CONFIG_HLP_SL_MODE "sets the NR sidelink mode (0: not in sidelink mode, 1: in-coverage/gNB, 2: out-of-coverage/no gNB)\n" #define CONFIG_HLP_EXTS "tells hardware to use an external timing reference\n" #define CONFIG_HLP_DMRSSYNC "tells RU to insert DMRS in subframe 1 slot 0" #define CONFIG_HLP_CLK "tells hardware to use a clock reference (0:internal, 1:external, 2:gpsdo)\n" @@ -106,6 +106,9 @@ extern "C" #define CONFIG_L1_EMULATOR "Run in L1 emulated mode (disable PHY layer)\n" #define CONFIG_HLP_CONTINUOUS_TX "perform continuous transmission, even in TDD mode (to work around USRP issues)\n" #define CONFIG_HLP_STATS_DISABLE "disable globally the stats generation and persistence" +#define CONFIG_HLP_SYNC_REF "Sync Reference in Sidelink\n" +#define CONFIG_HLP_NID1 "Set NID1 value in Sidelink\n" +#define CONFIG_HLP_NID2 "Set NID2 value in Sidelink\n" /*-----------------------------------------------------------------------------------------------------------------------------------------------------*/ /* command line parameters common to eNodeB and UE */ @@ -137,6 +140,10 @@ extern "C" #define NON_STOP softmodem_params.non_stop #define EMULATE_L1 softmodem_params.emulate_l1 #define CONTINUOUS_TX softmodem_params.continuous_tx +#define SYNC_REF softmodem_params.sync_ref +#define NID1 softmodem_params.nid1 +#define NID2 softmodem_params.nid2 + #define REORDER_THREAD_DISABLE softmodem_params.reorder_thread_disable #define DEFAULT_RFCONFIG_FILE "/usr/local/etc/syriq/ue.band7.tm1.PRB100.NR40.dat"; @@ -162,8 +169,8 @@ extern int usrp_tx_thread; {"d" , CONFIG_HLP_SOFTS, PARAMFLAG_BOOL, .uptr=&do_forms, .defintval=0, TYPE_UINT, 0}, \ {"dqt" , CONFIG_HLP_SOFTS_QT, PARAMFLAG_BOOL, .uptr=&do_forms_qt, .defintval=0, TYPE_UINT, 0}, \ {"q" , CONFIG_HLP_STMON, PARAMFLAG_BOOL, .iptr=&opp_enabled, .defintval=0, TYPE_INT, 0}, \ - {"numerology" , CONFIG_HLP_NUMEROLOGY, PARAMFLAG_BOOL, .iptr=&NUMEROLOGY, .defintval=1, TYPE_INT, 0}, \ - {"band" , CONFIG_HLP_BAND, PARAMFLAG_BOOL, .iptr=&BAND, .defintval=78, TYPE_INT, 0}, \ + {"numerology" , CONFIG_HLP_NUMEROLOGY, 0, .iptr=&NUMEROLOGY, .defintval=1, TYPE_INT, 0}, \ + {"band" , CONFIG_HLP_BAND, 0, .iptr=&BAND, .defintval=78, TYPE_INT, 0}, \ {"emulate-rf" , CONFIG_HLP_EMULATE_RF, PARAMFLAG_BOOL, .iptr=&EMULATE_RF, .defintval=0, TYPE_INT, 0}, \ {"parallel-config", CONFIG_HLP_PARALLEL_CMD, 0, .strptr=¶llel_config, .defstrval=NULL, TYPE_STRING, 0}, \ {"worker-config", CONFIG_HLP_WORKER_CMD, 0, .strptr=&worker_config, .defstrval=NULL, TYPE_STRING, 0}, \ @@ -181,6 +188,8 @@ extern int usrp_tx_thread; {"emulate-l1", CONFIG_L1_EMULATOR, PARAMFLAG_BOOL, .iptr=&EMULATE_L1, .defintval=0, TYPE_INT, 0}, \ {"continuous-tx", CONFIG_HLP_CONTINUOUS_TX, PARAMFLAG_BOOL, .iptr=&CONTINUOUS_TX, .defintval=0, TYPE_INT, 0}, \ {"disable-stats", CONFIG_HLP_STATS_DISABLE, PARAMFLAG_BOOL, .iptr=&stats_disabled, .defintval=0, TYPE_INT, 0}, \ + {"nid1", CONFIG_HLP_NID1, 0, .iptr=&NID1, .defintval=10, TYPE_INT, 0}, \ + {"nid2", CONFIG_HLP_NID2, 0, .iptr=&NID2, .defintval=1, TYPE_INT, 0}, \ } // clang-format on @@ -227,6 +236,8 @@ extern int usrp_tx_thread; { .s5 = { NULL } }, \ { .s5 = { NULL } }, \ { .s5 = { NULL } }, \ + { .s5 = { NULL } }, \ + { .s5 = { NULL } }, \ } // clang-format on @@ -335,6 +346,9 @@ typedef struct { int non_stop; int emulate_l1; int continuous_tx; + int sync_ref; + int nid1; + int nid2; } softmodem_params_t; extern uint64_t get_softmodem_optmask(void); diff --git a/openair1/PHY/CODING/nrPolar_tools/nr_polar_decoding_tools.c b/openair1/PHY/CODING/nrPolar_tools/nr_polar_decoding_tools.c index 20ecc690b3..7fd8d351ff 100644 --- a/openair1/PHY/CODING/nrPolar_tools/nr_polar_decoding_tools.c +++ b/openair1/PHY/CODING/nrPolar_tools/nr_polar_decoding_tools.c @@ -351,9 +351,17 @@ void applyGtoright(const t_nrPolar_params *pp,decoder_node_t *node) { ((__m64 *)alpha_r)[0] = _mm_subs_pi16(((__m64 *)alpha_v)[1],_mm_sign_pi16(((__m64 *)alpha_v)[0],((__m64 *)betal)[0])); } else - {// equivalent scalar code to above, activated only on non x86/ARM architectures or Nv=1,2 - for (int i=0;iNv/2;i++) { - alpha_r[i] = alpha_v[i+(node->Nv/2)] - (betal[i]*alpha_v[i]); + { + int temp_alpha_r; + for (int i = 0; i < node->Nv / 2; i++) { + temp_alpha_r = alpha_v[i + (node->Nv / 2)] - (betal[i] * alpha_v[i]); + if (temp_alpha_r > SHRT_MAX) { + alpha_r[i] = SHRT_MAX; + } else if (temp_alpha_r < -SHRT_MAX) { + alpha_r[i] = -SHRT_MAX; + } else { + alpha_r[i] = temp_alpha_r; + } } } if (node->Nv == 2) { // apply hard decision on right node diff --git a/openair1/PHY/CODING/nrPolar_tools/nr_polar_pbch_defs.h b/openair1/PHY/CODING/nrPolar_tools/nr_polar_pbch_defs.h index 2ced9633ce..1904d1bd83 100644 --- a/openair1/PHY/CODING/nrPolar_tools/nr_polar_pbch_defs.h +++ b/openair1/PHY/CODING/nrPolar_tools/nr_polar_pbch_defs.h @@ -35,6 +35,7 @@ #define NR_POLAR_PBCH_AGGREGATION_LEVEL 0 //uint8_t #define NR_POLAR_PBCH_MESSAGE_TYPE 0 //int8_t +#define NR_POLAR_PSBCH_MESSAGE_TYPE 3 //int8_t #define NR_POLAR_PBCH_PAYLOAD_BITS 32 //uint16_t #define NR_POLAR_PBCH_CRC_PARITY_BITS 24 #define NR_POLAR_PBCH_CRC_ERROR_CORRECTION_BITS 3 @@ -52,6 +53,8 @@ #define NR_POLAR_PBCH_I_BIL 0 //uint8_t #define NR_POLAR_PBCH_E 864 //uint16_t #define NR_POLAR_PBCH_E_DWORD 27 // NR_POLAR_PBCH_E/32 +#define NR_POLAR_PSBCH_E 1792 //uint16_t +#define NR_POLAR_PSBCH_E_DWORD 56 // NR_POLAR_PSBCH_E/32 /* * TEST CODE diff --git a/openair1/PHY/LTE_UE_TRANSPORT/dlsch_llr_computation_avx2.c b/openair1/PHY/LTE_UE_TRANSPORT/dlsch_llr_computation_avx2.c index c00040f668..88d7dfef6b 100644 --- a/openair1/PHY/LTE_UE_TRANSPORT/dlsch_llr_computation_avx2.c +++ b/openair1/PHY/LTE_UE_TRANSPORT/dlsch_llr_computation_avx2.c @@ -121,34 +121,6 @@ tmp_result2 = simde_mm256_slli_epi16(tmp_result2, 1); \ const simde__m256i a_sq = simde_mm256_adds_epi16(tmp_result, tmp_result2); - void seperate_real_imag_parts(__m256i *out_re, __m256i *out_im, __m256i in0, __m256i in1) - { - __m256i tmp0; - __m256i tmp1; - - in0 = simde_mm256_shufflelo_epi16(in0, 0xd8); //_MM_SHUFFLE(0,2,1,3)); - in0 = simde_mm256_shufflehi_epi16(in0, 0xd8); //_MM_SHUFFLE(0,2,1,3)); - in0 = simde_mm256_shuffle_epi32(in0, 0xd8); //_MM_SHUFFLE(0,2,1,3)); - - in1 = simde_mm256_shufflelo_epi16(in1, 0xd8); //_MM_SHUFFLE(0,2,1,3)); - in1 = simde_mm256_shufflehi_epi16(in1, 0xd8); //_MM_SHUFFLE(0,2,1,3)); - in1 = simde_mm256_shuffle_epi32(in1, 0xd8); //_MM_SHUFFLE(0,2,1,3)); - - // in0 = [Re(0,1,2,3) Im(0,1,2,3) Re(4,5,6,7) Im(4,5,6,7)] - // in0 = [Re(8,9,10,11) Im(8,9,10,11) Re(12,13,14,15) Im(12,13,14,15)] - - tmp0 = simde_mm256_unpacklo_epi64(in0, in1); - // axmm2 = [Re(0,1,2,3) Re(8,9,10,11) Re(4,5,6,7) Re(12,13,14,15)] - tmp0 = simde_mm256_permute4x64_epi64(tmp0, 0xd8); // Re(rho) - - tmp1 = simde_mm256_unpackhi_epi64(in0, in1); - // axmm3 = [Im(0,1,2,3) Im(8,9,10,11) Im(4,5,6,7) Im(12,13,14,15)] - tmp1 = simde_mm256_permute4x64_epi64(tmp1, 0xd8); // Im(rho) - - *out_re = tmp0; - *out_im = tmp1; -} - void qam64_qam16_avx2(short *stream0_in, short *stream1_in, short *ch_mag, @@ -245,7 +217,7 @@ void qam64_qam16_avx2(short *stream0_in, xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho) */ simde__m256i xmm0, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8; - seperate_real_imag_parts(&xmm2, &xmm3, rho01_256i[i], rho01_256i[i+1]); + simde_mm256_separate_real_imag_parts(&xmm2, &xmm3, rho01_256i[i], rho01_256i[i+1]); const simde__m256i rho_rpi = simde_mm256_adds_epi16(xmm2, xmm3); // rho = Re(rho) + Im(rho) const simde__m256i rho_rmi = simde_mm256_subs_epi16(xmm2, xmm3); // rho* = Re(rho) - Im(rho) @@ -323,7 +295,7 @@ void qam64_qam16_avx2(short *stream0_in, */ simde__m256i y1r, y1i; - seperate_real_imag_parts(&y1r, &y1i, stream1_256i_in[i], stream1_256i_in[i+1]); + simde_mm256_separate_real_imag_parts(&y1r, &y1i, stream1_256i_in[i], stream1_256i_in[i+1]); // Psi_r calculation from rho_rpi or rho_rmi xmm0 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(0));// ZERO for abs_pi16 @@ -604,7 +576,7 @@ void qam64_qam16_avx2(short *stream0_in, y0i = simde_mm256_unpackhi_epi64(xmm0,xmm1); */ simde__m256i y0r, y0i; - seperate_real_imag_parts(&y0r, &y0i, stream0_256i_in[i], stream0_256i_in[i+1]); + simde_mm256_separate_real_imag_parts(&y0r, &y0i, stream0_256i_in[i], stream0_256i_in[i+1]); /* // Rearrange desired channel magnitudes @@ -619,7 +591,7 @@ void qam64_qam16_avx2(short *stream0_in, ch_mag_des = simde_mm256_unpacklo_epi64(xmm2,xmm3); */ - seperate_real_imag_parts(&ch_mag_des, &xmm2, ch_mag_256i[i], ch_mag_256i[i+1]); + simde_mm256_separate_real_imag_parts(&ch_mag_des, &xmm2, ch_mag_256i[i], ch_mag_256i[i+1]); // Rearrange interfering channel magnitudes /* @@ -634,7 +606,7 @@ void qam64_qam16_avx2(short *stream0_in, ch_mag_int = simde_mm256_unpacklo_epi64(xmm2,xmm3); */ - seperate_real_imag_parts(&ch_mag_int, &xmm2, ch_mag_256i_i[i], ch_mag_256i_i[i+1]); + simde_mm256_separate_real_imag_parts(&ch_mag_int, &xmm2, ch_mag_256i_i[i], ch_mag_256i_i[i+1]); y0r_one_over_sqrt_21 = simde_mm256_mulhi_epi16(y0r, ONE_OVER_SQRT_42); y0r_three_over_sqrt_21 = simde_mm256_mulhi_epi16(y0r, THREE_OVER_SQRT_42); @@ -1819,7 +1791,7 @@ void qam64_qam64_avx2(int32_t *stream0_in, xmm3 = simde_mm256_permute4x64_epi64(xmm3,0xd8); // Im(rho) */ simde__m256i xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8; - seperate_real_imag_parts(&xmm2, &xmm3, rho01_256i[i], rho01_256i[i+1]); + simde_mm256_separate_real_imag_parts(&xmm2, &xmm3, rho01_256i[i], rho01_256i[i+1]); simde__m256i rho_rpi = simde_mm256_adds_epi16(xmm2, xmm3); // rho = Re(rho) + Im(rho) simde__m256i rho_rmi = simde_mm256_subs_epi16(xmm2, xmm3); // rho* = Re(rho) - Im(rho) @@ -1899,7 +1871,7 @@ void qam64_qam64_avx2(int32_t *stream0_in, y1i = simde_mm256_permute4x64_epi64(y1i,0xd8); // Im(y1) */ simde__m256i y1r, y1i, xmm0; - seperate_real_imag_parts(&y1r, &y1i, stream1_256i_in[i], stream1_256i_in[i+1]); + simde_mm256_separate_real_imag_parts(&y1r, &y1i, stream1_256i_in[i], stream1_256i_in[i+1]); // Psi_r calculation from rho_rpi or rho_rmi xmm0 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(0));// ZERO for abs_pi16 @@ -2181,7 +2153,7 @@ void qam64_qam64_avx2(int32_t *stream0_in, y0i = _mm_unpackhi_epi64(xmm0,xmm1); */ simde__m256i y0r, y0i; - seperate_real_imag_parts(&y0r, &y0i, stream0_256i_in[i], stream0_256i_in[i+1]); + simde_mm256_separate_real_imag_parts(&y0r, &y0i, stream0_256i_in[i], stream0_256i_in[i+1]); // Rearrange desired channel magnitudes // [|h|^2(1),|h|^2(1),|h|^2(2),|h|^2(2),...,,|h|^2(7),|h|^2(7)]*(2/sqrt(10)) @@ -2197,7 +2169,7 @@ void qam64_qam64_avx2(int32_t *stream0_in, ch_mag_des = _mm_unpacklo_epi64(xmm2,xmm3); */ // xmm2 is dummy variable that contains the same values as ch_mag_des - seperate_real_imag_parts(&ch_mag_des, &xmm2, ch_mag_256i[i], ch_mag_256i[i+1]); + simde_mm256_separate_real_imag_parts(&ch_mag_des, &xmm2, ch_mag_256i[i], ch_mag_256i[i+1]); // Rearrange interfering channel magnitudes @@ -2212,7 +2184,7 @@ void qam64_qam64_avx2(int32_t *stream0_in, xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3)); ch_mag_int = _mm_unpacklo_epi64(xmm2,xmm3); */ - seperate_real_imag_parts(&ch_mag_int, &xmm2, ch_mag_256i_i[i], ch_mag_256i_i[i+1]); + simde_mm256_separate_real_imag_parts(&ch_mag_int, &xmm2, ch_mag_256i_i[i], ch_mag_256i_i[i+1]); y0r_one_over_sqrt_21 = simde_mm256_mulhi_epi16(y0r, ONE_OVER_SQRT_42); y0r_three_over_sqrt_21 = simde_mm256_mulhi_epi16(y0r, THREE_OVER_SQRT_42); diff --git a/openair1/PHY/NR_REFSIG/ss_pbch_nr.h b/openair1/PHY/NR_REFSIG/ss_pbch_nr.h index ad96054c12..0929d483cd 100644 --- a/openair1/PHY/NR_REFSIG/ss_pbch_nr.h +++ b/openair1/PHY/NR_REFSIG/ss_pbch_nr.h @@ -41,6 +41,9 @@ /* PSS parameters */ #define NUMBER_PSS_SEQUENCE (3) +#define NUMBER_PSS_SEQUENCE_SL (2) +#define PSS_SSS_SUB_CARRIER_START (56) +#define PSS_SSS_SUB_CARRIER_START_SL (2) #define INVALID_PSS_SEQUENCE (NUMBER_PSS_SEQUENCE) #define LENGTH_PSS_NR (127) #define N_SC_RB (12) /* Resource block size in frequency domain expressed as a number if subcarriers */ @@ -62,6 +65,13 @@ #define SSS_SYMBOL_NB ((2) + OFFSET_SS_PBCH) #define PBCH_LAST_SYMBOL_NB ((3) + OFFSET_SS_PBCH) + +#define OFFSET_SS_PSBCH -1 +#define PSS0_SL_SYMBOL_NB ((1) + OFFSET_SS_PSBCH) +#define PSS1_SL_SYMBOL_NB ((2) + OFFSET_SS_PSBCH) +#define SSS0_SL_SYMBOL_NB ((3) + OFFSET_SS_PSBCH) +#define SSS1_SL_SYMBOL_NB ((4) + OFFSET_SS_PSBCH) + /* SS/PBCH parameters */ #define N_RB_SS_PBCH_BLOCK (20) #define NB_SYMBOLS_PBCH (3) diff --git a/openair1/PHY/NR_REFSIG/sss_nr.h b/openair1/PHY/NR_REFSIG/sss_nr.h index 7303d99d3b..38d6e72819 100644 --- a/openair1/PHY/NR_REFSIG/sss_nr.h +++ b/openair1/PHY/NR_REFSIG/sss_nr.h @@ -47,13 +47,22 @@ #define SCALING_METRIC_SSS_NR (15)//(19) #define N_ID_2_NUMBER (NUMBER_PSS_SEQUENCE) +#define N_ID_2_NUMBER_SL (NUMBER_PSS_SEQUENCE_SL) #define N_ID_1_NUMBER (NUMBER_SSS_SEQUENCE) #define GET_NID2(Nid_cell) (Nid_cell%3) #define GET_NID1(Nid_cell) (Nid_cell/3) +#define GET_NID2_SL(Nid_SL) (Nid_SL/NUMBER_SSS_SEQUENCE) +#define GET_NID1_SL(Nid_SL) (Nid_SL%NUMBER_SSS_SEQUENCE) + #define PSS_SC_START_NR (52) /* see from TS 38.211 table 7.4.3.1-1: Resources within an SS/PBCH block for PSS... */ +#define SSS_START_IDX (3) /* [0:PSBCH 1:PSS0 2:PSS1 3:SSS0 4:SSS1] */ +#define NUM_SSS_SYMBOLS (2) + +#define SSS_METRIC_FLOOR_NR (30000) + /************** VARIABLES *****************************************/ #define PHASE_HYPOTHESIS_NUMBER (16) diff --git a/openair1/PHY/NR_TRANSPORT/nr_transport_proto.h b/openair1/PHY/NR_TRANSPORT/nr_transport_proto.h index 0509ce7acf..2478748688 100644 --- a/openair1/PHY/NR_TRANSPORT/nr_transport_proto.h +++ b/openair1/PHY/NR_TRANSPORT/nr_transport_proto.h @@ -292,6 +292,19 @@ void nr_ulsch_compute_llr(int32_t *rxdataF_comp, void reset_active_stats(PHY_VARS_gNB *gNB, int frame); void reset_active_ulsch(PHY_VARS_gNB *gNB, int frame); +void nr_ulsch_compute_ML_llr(int32_t **rxdataF_comp, + int32_t **ul_ch_mag, + int32_t ***rho, + int16_t **llr_layers, + uint8_t nb_antennas_rx, + uint32_t rb_size, + uint32_t nb_re, + uint8_t symbol, + uint32_t rxdataF_ext_offset, + uint8_t mod_order); + +void nr_ulsch_shift_llr(int16_t **llr_layers, uint32_t nb_re, uint32_t rxdataF_ext_offset, uint8_t mod_order, int shift); + void nr_fill_ulsch(PHY_VARS_gNB *gNB, int frame, int slot, diff --git a/openair1/PHY/NR_TRANSPORT/nr_ulsch_demodulation.c b/openair1/PHY/NR_TRANSPORT/nr_ulsch_demodulation.c index 84baea9844..b06b690c0b 100644 --- a/openair1/PHY/NR_TRANSPORT/nr_ulsch_demodulation.c +++ b/openair1/PHY/NR_TRANSPORT/nr_ulsch_demodulation.c @@ -12,6 +12,7 @@ //#define DEBUG_CH_COMP //#define DEBUG_RB_EXT //#define DEBUG_CH_MAG +//#define ML_DEBUG #define INVALID_VALUE 255 @@ -636,7 +637,7 @@ void nr_ulsch_channel_compensation(int **rxdataF_ext, QAM_amp128 = _mm_set1_epi16(QAM16_n1); // 2/sqrt(10) QAM_amp128b = _mm_setzero_si128(); QAM_amp128c = _mm_setzero_si128(); - } + } else if (mod_order == 6) { QAM_amp128 = _mm_set1_epi16(QAM64_n1); // QAM_amp128b = _mm_set1_epi16(QAM64_n2); @@ -1081,7 +1082,7 @@ void nr_ulsch_detection_mrc(NR_DL_FRAME_PARMS *frame_parms, int32_t **ul_ch_mag, int32_t **ul_ch_magb, int32_t **ul_ch_magc, - int32_t ***rho, + int32_t ***rho, uint8_t nrOfLayers, uint8_t symbol, uint16_t nb_rb, @@ -1100,8 +1101,10 @@ void nr_ulsch_detection_mrc(NR_DL_FRAME_PARMS *frame_parms, if (n_rx > 1) { #if defined(__x86_64__) || defined(__i386__) - for (int aatx=0; aatxnb_antennas_rx][(symbol*(nb_re + off))]; ul_ch_mag128[0] = (__m128i *)&ul_ch_mag[aatx*frame_parms->nb_antennas_rx][(symbol*(nb_re + off))]; @@ -1113,7 +1116,7 @@ void nr_ulsch_detection_mrc(NR_DL_FRAME_PARMS *frame_parms, ul_ch_mag128[1] = (__m128i *)&ul_ch_mag[aatx*frame_parms->nb_antennas_rx+aa][(symbol*(nb_re + off))]; ul_ch_mag128b[1] = (__m128i *)&ul_ch_magb[aatx*frame_parms->nb_antennas_rx+aa][(symbol*(nb_re + off))]; ul_ch_mag128c[1] = (__m128i *)&ul_ch_magc[aatx*frame_parms->nb_antennas_rx+aa][(symbol*(nb_re + off))]; - + // MRC on each re of rb, both on MF output and magnitude (for 16QAM/64QAM llr computation) for (i=0; iN_RB_DL*12]; @@ -1982,11 +1999,20 @@ void nr_rx_pusch(PHY_VARS_gNB *gNB, uint32_t rxdataF_ext_offset = 0; uint8_t shift_ch_ext = rel15_ul->nrOfLayers > 1 ? log2_approx(max_ch >> 11) : 0; + // Flag to select the receiver: (true) Nonlinear ML receiver, (false) Linear MMSE receiver + // By default, we are using the Nonlinear ML receiver, except + // - for 256QAM as Nonlinear ML receiver is not implemented for 256QAM + // - for 64QAM as Nonlinear ML receiver requires more processing time than MMSE, and many machines are not powerful enough + bool ml_rx = true; + if (rel15_ul->nrOfLayers != 2 || rel15_ul->qam_mod_order >= 6) { + ml_rx = false; + } + int ad_shift = 0; if (rel15_ul->nrOfLayers == 1) { ad_shift = 1 + log2_approx(frame_parms->nb_antennas_rx >> 2); - } else { - ad_shift = -3; // For 2-layers, we are already doing a bit shift in the nr_ulsch_zero_forcing_rx_2layers() function, so we can use more bits + } else if (ml_rx == false) { + ad_shift = -3; // For 2-layers, we are already doing a bit shift in the nr_ulsch_mmse_2layers() function, so we can use more bits } for(uint8_t symbol = rel15_ul->start_symbol_index; symbol < (rel15_ul->start_symbol_index + rel15_ul->nr_of_symbols); symbol++) { @@ -2092,7 +2118,7 @@ void nr_rx_pusch(PHY_VARS_gNB *gNB, nb_re_pusch); // Apply MMSE for 2 Tx layers - if (rel15_ul->nrOfLayers == 2) { + if (ml_rx == false && rel15_ul->nrOfLayers == 2) { nr_ulsch_mmse_2layers(frame_parms, pusch_vars->rxdataF_comp, pusch_vars->ul_ch_mag0, @@ -2143,16 +2169,46 @@ void nr_rx_pusch(PHY_VARS_gNB *gNB, /*-------------------- LLRs computation -------------------------------------------------------------*/ /*-----------------------------------------------------------------------------------------------------*/ start_meas(&gNB->ulsch_llr_stats); - for (aatx=0; aatx < rel15_ul->nrOfLayers; aatx++) { - nr_ulsch_compute_llr(&pusch_vars->rxdataF_comp[aatx*frame_parms->nb_antennas_rx][symbol * (off + rel15_ul->rb_size * NR_NB_SC_PER_RB)], - pusch_vars->ul_ch_mag0[aatx*frame_parms->nb_antennas_rx], - pusch_vars->ul_ch_magb0[aatx*frame_parms->nb_antennas_rx], - pusch_vars->ul_ch_magc0[aatx*frame_parms->nb_antennas_rx], - &pusch_vars->llr_layers[aatx][rxdataF_ext_offset * rel15_ul->qam_mod_order], - rel15_ul->rb_size, - pusch_vars->ul_valid_re_per_slot[symbol], - symbol, - rel15_ul->qam_mod_order); + if (ml_rx == false || rel15_ul->nrOfLayers == 1) { + for (aatx=0; aatx < rel15_ul->nrOfLayers; aatx++) { + nr_ulsch_compute_llr(&pusch_vars->rxdataF_comp[aatx * frame_parms->nb_antennas_rx][symbol * (off + rel15_ul->rb_size * NR_NB_SC_PER_RB)], + pusch_vars->ul_ch_mag0[aatx * frame_parms->nb_antennas_rx], + pusch_vars->ul_ch_magb0[aatx * frame_parms->nb_antennas_rx], + pusch_vars->ul_ch_magc0[aatx * frame_parms->nb_antennas_rx], + &pusch_vars->llr_layers[aatx][rxdataF_ext_offset * rel15_ul->qam_mod_order], + rel15_ul->rb_size, + pusch_vars->ul_valid_re_per_slot[symbol], + symbol, + rel15_ul->qam_mod_order); + } + } else { + nr_ulsch_compute_ML_llr(pusch_vars->rxdataF_comp, + pusch_vars->ul_ch_mag0, + pusch_vars->rho, + pusch_vars->llr_layers, + frame_parms->nb_antennas_rx, + rel15_ul->rb_size, + nb_re_pusch, + symbol, + rxdataF_ext_offset, + rel15_ul->qam_mod_order); + + if (rel15_ul->qam_mod_order == 2) { + nr_ulsch_shift_llr(pusch_vars->llr_layers, nb_re_pusch, rxdataF_ext_offset, rel15_ul->qam_mod_order, 4); + } + +#ifdef ML_DEBUG + c16_t *llr_layers0 = (c16_t *)&pusch_vars->llr_layers[0][rxdataF_ext_offset * rel15_ul->qam_mod_order]; + c16_t *llr_layers1 = (c16_t *)&pusch_vars->llr_layers[1][rxdataF_ext_offset * rel15_ul->qam_mod_order]; + printf("===============================\n"); + printf("AFTER nr_ulsch_compute_ML_llr()\n"); + printf("===============================\n"); + for (int k = 0; k < nb_re_pusch; k++) { + printf("[%3i] llr_layers0 = (%6i, %6i), llr_layers1 = (%6i, %6i)\n", + k, llr_layers0[k].r, llr_layers0[k].i, llr_layers1[k].r, llr_layers1[k].i); + } + printf("\n"); +#endif } stop_meas(&gNB->ulsch_llr_stats); rxdataF_ext_offset += pusch_vars->ul_valid_re_per_slot[symbol]; diff --git a/openair1/PHY/NR_TRANSPORT/nr_ulsch_llr_computation.c b/openair1/PHY/NR_TRANSPORT/nr_ulsch_llr_computation.c index 4d4c81a986..f378830c23 100644 --- a/openair1/PHY/NR_TRANSPORT/nr_ulsch_llr_computation.c +++ b/openair1/PHY/NR_TRANSPORT/nr_ulsch_llr_computation.c @@ -509,3 +509,3440 @@ void nr_ulsch_compute_llr(int32_t *rxdataF_comp, break; } } + +/* + * 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. + * + * Input: + * stream0_in: MF filter output for 1st stream, i.e., y0' = h0'*y0 + * stream1_in: MF filter output for 2nd stream, i.e., y1' = h1'*y0 + * rho01: Channel cross correlation, i.e., rho01 = h0'*h1 + * length: Number of resource elements + * + * Output: + * stream0_out: Output LLRs for 1st stream + */ +void nr_ulsch_qpsk_qpsk(c16_t *stream0_in, c16_t *stream1_in, c16_t *stream0_out, c16_t *rho01, uint32_t length) +{ + __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_2_SQRT_2 = _mm_set1_epi16(23170); // round(2 ^ 16 / (2 * sqrt(2))) + + // In each iteration, we take 8 complex symbols + for (int i = 0; i < length >> 2; i += 2) { + + /// Compute real and imaginary parts of MF output for stream 0 (desired stream) + __m128i y0r, y0i; + simde_mm128_separate_real_imag_parts(&y0r, &y0i, stream0_128i_in[i], stream0_128i_in[i + 1]); + __m128i y0r_over2 = simde_mm_mulhi_epi16(y0r, ONE_OVER_2_SQRT_2); + y0r_over2 = _mm_slli_epi16(y0r_over2, 1); // y0r_over2 = Re(y0) / sqrt(2) + __m128i y0i_over2 = simde_mm_mulhi_epi16(y0i, ONE_OVER_2_SQRT_2); + y0i_over2 = _mm_slli_epi16(y0i_over2, 1); // y0i_over2 = Im(y0) / sqrt(2) + + /// Compute real and imaginary parts of MF output for stream 1 (interference stream) + __m128i y1r_over2, y1i_over2; + simde_mm128_separate_real_imag_parts(&y1r_over2, &y1i_over2, stream1_128i_in[i], stream1_128i_in[i + 1]); + y1r_over2 = simde_mm_srai_epi16(y1r_over2, 1); // y1r_over2 = Re(y1) / 2 + y1i_over2 = simde_mm_srai_epi16(y1i_over2, 1); // y1i_over2 = Im(y1) / 2 + + /// Get real and imaginary parts of rho + __m128i rhor, rhoi; + simde_mm128_separate_real_imag_parts(&rhor, &rhoi, rho01_128i[i], rho01_128i[i + 1]); + + /// Compute |psi_r| and |psi_i| + + // psi_r = rhor * xR + rhoi * xI + // psi_i = rhor * xI - rhoi * xR + + // Put (rho_r + rho_i)/(2*sqrt(2)) in rho_p + // rhor * xR + rhoi * xI --> xR = 1/sqrt(2) and xI = 1/sqrt(2) + // rhor * xI - rhoi * xR --> xR = -1/sqrt(2) and xI = 1/sqrt(2) + __m128i rho_p = simde_mm_adds_epi16(rhor, rhoi); // rho_p = Re(rho) + Im(rho) + rho_p = simde_mm_mulhi_epi16(rho_p, ONE_OVER_2_SQRT_2); // rho_p = rho_p / (2*sqrt(2)) + + // Put (rho_r - rho_i)/(2*sqrt(2)) in rho_m + // rhor * xR + rhoi * xI --> xR = 1/sqrt(2) and xI = -1/sqrt(2) + // rhor * xI - rhoi * xR --> xR = 1/sqrt(2) and xI = 1/sqrt(2) + __m128i rho_m = simde_mm_subs_epi16(rhor, rhoi); // rho_m = Re(rho) - Im(rho) + rho_m = simde_mm_mulhi_epi16(rho_m, ONE_OVER_2_SQRT_2); // rho_m = rho_m / (2*sqrt(2)) + + // xR = 1/sqrt(2) and xI = 1/sqrt(2) + __m128i abs_psi_rpm = simde_mm_subs_epi16(rho_p, y1r_over2); // psi_rpm = rho_p - y1r/2 + abs_psi_rpm = simde_mm_abs_epi16(abs_psi_rpm); // abs_psi_rpm = |psi_rpm| + + // xR = 1/sqrt(2) and xI = 1/sqrt(2) + __m128i abs_psi_imm = simde_mm_subs_epi16(rho_m, y1i_over2); // psi_imm = rho_m - y1i/2 + abs_psi_imm = simde_mm_abs_epi16(abs_psi_imm); // abs_psi_imm = |psi_imm| + + // xR = 1/sqrt(2) and xI = -1/sqrt(2) + __m128i abs_psi_rmm = simde_mm_subs_epi16(rho_m, y1r_over2); // psi_rmm = rho_m - y1r/2 + abs_psi_rmm = simde_mm_abs_epi16(abs_psi_rmm); // abs_psi_rmm = |psi_rmm| + + // xR = -1/sqrt(2) and xI = 1/sqrt(2) + __m128i abs_psi_ipm = simde_mm_subs_epi16(rho_p, y1i_over2); // psi_ipm = rho_p - y1i/2 + abs_psi_ipm = simde_mm_abs_epi16(abs_psi_ipm); // abs_psi_ipm = |psi_ipm| + + // xR = -1/sqrt(2) and xI = -1/sqrt(2) + __m128i abs_psi_rpp = simde_mm_adds_epi16(rho_p, y1r_over2); // psi_rpp = rho_p + y1r/2 + abs_psi_rpp = simde_mm_abs_epi16(abs_psi_rpp); // abs_psi_rpp = |psi_rpp| + + // xR = -1/sqrt(2) and xI = -1/sqrt(2) + __m128i abs_psi_imp = simde_mm_adds_epi16(rho_m, y1i_over2); // psi_imp = rho_m + y1i/2 + abs_psi_imp = simde_mm_abs_epi16(abs_psi_imp); // abs_psi_imp = |psi_imp| + + // xR = -1/sqrt(2) and xI = 1/sqrt(2) + __m128i abs_psi_rmp = simde_mm_adds_epi16(rho_m, y1r_over2); // psi_rmp = rho_m + y1r/2 + abs_psi_rmp = simde_mm_abs_epi16(abs_psi_rmp); // abs_psi_rmp = |psi_rmp| + + // xR = 1/sqrt(2) and xI = -1/sqrt(2) + __m128i abs_psi_ipp = simde_mm_adds_epi16(rho_p, y1i_over2); // psi_ipm = rho_p + y1i/2 + abs_psi_ipp = simde_mm_abs_epi16(abs_psi_ipp); // abs_psi_ipp = |psi_ipm| + + /// Compute bit metrics (lambda) + + // lambda = max { |psi_r - y1r| * |x2R| + |psi_i - y1i| * |x2I| + y0r * xR + y0i * xI} + + // xR = 1/sqrt(2) and xI = 1/sqrt(2) + // For numerator: bit_met_num_re_p = abs_psi_rpm + abs_psi_imm + y0r/sqrt(2) + y0i/sqrt(2) + __m128i bit_met_num_re_p = simde_mm_adds_epi16(abs_psi_rpm, abs_psi_imm); + bit_met_num_re_p = simde_mm_adds_epi16(bit_met_num_re_p, y0r_over2); + bit_met_num_re_p = simde_mm_adds_epi16(bit_met_num_re_p, y0i_over2); + + // xR = 1/sqrt(2) and xI = -1/sqrt(2) + // For numerator: bit_met_num_re_m = abs_psi_rmm + abs_psi_ipp + y0r/sqrt(2) - y0i/sqrt(2) + __m128i bit_met_num_re_m = simde_mm_adds_epi16(abs_psi_rmm, abs_psi_ipp); + bit_met_num_re_m = simde_mm_adds_epi16(bit_met_num_re_m, y0r_over2); + bit_met_num_re_m = simde_mm_subs_epi16(bit_met_num_re_m, y0i_over2); + + // xR = -1/sqrt(2) and xI = 1/sqrt(2) + // For denominator: bit_met_den_re_p = abs_psi_rmp + abs_psi_ipm - y0r/sqrt(2) + y0i/sqrt(2) + __m128i bit_met_den_re_p = simde_mm_adds_epi16(abs_psi_rmp, abs_psi_ipm); + bit_met_den_re_p = simde_mm_subs_epi16(bit_met_den_re_p, y0r_over2); + bit_met_den_re_p = simde_mm_adds_epi16(bit_met_den_re_p, y0i_over2); + + // xR = -1/sqrt(2) and xI = -1/sqrt(2) + // For denominator: bit_met_den_re_m = abs_psi_rpp + abs_psi_imp - y0r/sqrt(2) - y0i/sqrt(2) + __m128i bit_met_den_re_m = simde_mm_adds_epi16(abs_psi_rpp, abs_psi_imp); + bit_met_den_re_m = simde_mm_subs_epi16(bit_met_den_re_m, y0r_over2); + bit_met_den_re_m = simde_mm_subs_epi16(bit_met_den_re_m, y0i_over2); + + // xR = 1/sqrt(2) and xI = 1/sqrt(2) + // For numerator: bit_met_num_im_p = abs_psi_rpm + abs_psi_imm + y0r/sqrt(2) + y0i/sqrt(2) + __m128i bit_met_num_im_p = simde_mm_adds_epi16(abs_psi_rpm, abs_psi_imm); + bit_met_num_im_p = simde_mm_adds_epi16(bit_met_num_im_p, y0r_over2); + bit_met_num_im_p = simde_mm_adds_epi16(bit_met_num_im_p, y0i_over2); + + // xR = -1/sqrt(2) and xI = 1/sqrt(2) + // For numerator: bit_met_num_im_m = abs_psi_rmp + abs_psi_ipm - y0r/sqrt(2) + y0i/sqrt(2) + __m128i bit_met_num_im_m = simde_mm_adds_epi16(abs_psi_rmp, abs_psi_ipm); + bit_met_num_im_m = simde_mm_subs_epi16(bit_met_num_im_m, y0r_over2); + bit_met_num_im_m = simde_mm_adds_epi16(bit_met_num_im_m, y0i_over2); + + // xR = 1/sqrt(2) and xI = -1/sqrt(2) + // For denominator: bit_met_den_im_p = abs_psi_rmm + abs_psi_ipp + y0r/sqrt(2) - y0i/sqrt(2) + __m128i bit_met_den_im_p = simde_mm_adds_epi16(abs_psi_rmm, abs_psi_ipp); + bit_met_den_im_p = simde_mm_adds_epi16(bit_met_den_im_p, y0r_over2); + bit_met_den_im_p = simde_mm_subs_epi16(bit_met_den_im_p, y0i_over2); + + // xR = -1/sqrt(2) and xI = -1/sqrt(2) + // For denominator: bit_met_den_im_m = abs_psi_rpp + abs_psi_imp - y0r/sqrt(2)- y0i/sqrt(2) + __m128i bit_met_den_im_m = simde_mm_adds_epi16(abs_psi_rpp, abs_psi_imp); + bit_met_den_im_m = simde_mm_subs_epi16(bit_met_den_im_m, y0r_over2); + bit_met_den_im_m = simde_mm_subs_epi16(bit_met_den_im_m, y0i_over2); + + /// Compute the LLRs + + // LLR = lambda(c==1) - lambda(c==0) + + __m128i logmax_num_re0 = simde_mm_max_epi16(bit_met_num_re_p, bit_met_num_re_m); // LLR of the first bit: Bit = 1 + __m128i logmax_den_re0 = simde_mm_max_epi16(bit_met_den_re_p, bit_met_den_re_m); // LLR of the first bit: Bit = 0 + __m128i logmax_num_im0 = simde_mm_max_epi16(bit_met_num_im_p, bit_met_num_im_m); // LLR of the second bit: Bit = 1 + __m128i logmax_den_im0 = simde_mm_max_epi16(bit_met_den_im_p, bit_met_den_im_m); // LLR of the second bit: Bit = 0 + + y0r = simde_mm_subs_epi16(logmax_num_re0, logmax_den_re0); // LLR of first bit [L1(1), L1(2), L1(3), L1(4)] + y0i = simde_mm_subs_epi16(logmax_num_im0, logmax_den_im0); // LLR of second bit [L2(1), L2(2), L2(3), L2(4)] + + // [L1(1), L2(1), L1(2), L2(2)] + simde_mm_storeu_si128(&stream0_128i_out[i], simde_mm_unpacklo_epi16(y0r, y0i)); + + // false if only 2 REs remain + if (i < ((length >> 1) - 1)) { + simde_mm_storeu_si128(&stream0_128i_out[i + 1], simde_mm_unpackhi_epi16(y0r, y0i)); + } + } + + _mm_empty(); + _m_empty(); +} + +int16_t ones[8] __attribute__((aligned(16))) = {0xffff, 0xffff, 0xffff, 0xffff, 0xffff, 0xffff, 0xffff, 0xffff}; + +// 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 = simde_mm_mulhi_epi16(psi_r, a_r); \ + tmp_result = simde_mm_slli_epi16(tmp_result, 1); \ + tmp_result2 = simde_mm_mulhi_epi16(psi_i, a_i); \ + tmp_result2 = simde_mm_slli_epi16(tmp_result2, 1); \ + psi_a = simde_mm_adds_epi16(tmp_result, tmp_result2); + +// Calculate interference magnitude +#define interference_abs_epi16(psi, int_ch_mag, int_mag, c1, c2) \ + tmp_result = simde_mm_cmplt_epi16(psi, int_ch_mag); \ + tmp_result2 = simde_mm_xor_si128(tmp_result, (*(__m128i *)&ones[0])); \ + tmp_result = simde_mm_and_si128(tmp_result, c1); \ + tmp_result2 = simde_mm_and_si128(tmp_result2, c2); \ + int_mag = simde_mm_or_si128(tmp_result, tmp_result2); + +// 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 = simde_mm_mulhi_epi16(a_r, a_r); \ + tmp_result = simde_mm_slli_epi16(tmp_result, 1); \ + tmp_result = simde_mm_mulhi_epi16(tmp_result, scale_factor); \ + tmp_result = simde_mm_slli_epi16(tmp_result, 1); \ + tmp_result = simde_mm_mulhi_epi16(tmp_result, int_ch_mag); \ + tmp_result = simde_mm_slli_epi16(tmp_result, 1); \ + tmp_result2 = simde_mm_mulhi_epi16(a_i, a_i); \ + tmp_result2 = simde_mm_slli_epi16(tmp_result2, 1); \ + tmp_result2 = simde_mm_mulhi_epi16(tmp_result2, scale_factor); \ + tmp_result2 = simde_mm_slli_epi16(tmp_result2, 1); \ + tmp_result2 = simde_mm_mulhi_epi16(tmp_result2, int_ch_mag); \ + tmp_result2 = simde_mm_slli_epi16(tmp_result2, 1); \ + a_sq = simde_mm_adds_epi16(tmp_result, tmp_result2); + +__m128i max_epi16(__m128i m0, __m128i m1, __m128i m2, __m128i m3, __m128i m4, __m128i m5, __m128i m6, __m128i m7) +{ + __m128i a0 = simde_mm_max_epi16(m0, m1); + __m128i a1 = simde_mm_max_epi16(m2, m3); + __m128i a2 = simde_mm_max_epi16(m4, m5); + __m128i a3 = simde_mm_max_epi16(m6, m7); + __m128i b0 = simde_mm_max_epi16(a0, a1); + __m128i b1 = simde_mm_max_epi16(a2, a3); + return simde_mm_max_epi16(b0, b1); +} + +/* + * This function computes the LLRs of stream 0 (s_0) in presence of the interfering stream 1 (s_1) assuming that both symbols are + * 16QAM. It can be used for both MU-MIMO interference-aware receiver or for SU-MIMO receivers. + * + * Input: + * stream0_in: MF filter output for 1st stream, i.e., y0' = h0'*y0 + * stream1_in: MF filter output for 2nd stream, i.e., y1' = h1'*y0 + * ch_mag: 2*h0/sqrt(10), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + * ch_mag_i: 2*h1/sqrt(10), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + * rho01: Channel cross correlation, i.e., rho01 = h0'*h1 + * length: Number of resource elements + * + * Output: + * stream0_out: Output LLRs for 1st stream + */ +void nr_ulsch_qam16_qam16(c16_t *stream0_in, + c16_t *stream1_in, + c16_t *ch_mag, + c16_t *ch_mag_i, + c16_t *stream0_out, + c16_t *rho01, + uint32_t length) +{ + __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 = simde_mm_set1_epi16(20724); // round(1/sqrt(10)*2^16) + __m128i ONE_OVER_SQRT_10_Q15 = simde_mm_set1_epi16(10362); // round(1/sqrt(10)*2^15) + __m128i THREE_OVER_SQRT_10 = simde_mm_set1_epi16(31086); // round(3/sqrt(10)*2^15) + __m128i SQRT_10_OVER_FOUR = simde_mm_set1_epi16(25905); // round(sqrt(10)/4*2^15) + __m128i ONE_OVER_TWO_SQRT_10 = simde_mm_set1_epi16(10362); // round(1/2/sqrt(10)*2^16) + __m128i NINE_OVER_TWO_SQRT_10 = simde_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; + + __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 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_3_1 __attribute__((aligned(16))); + __m128i rho_rpi_3_3 __attribute__((aligned(16))); + __m128i rho_rmi_1_1 __attribute__((aligned(16))); + __m128i rho_rmi_1_3 __attribute__((aligned(16))); + __m128i rho_rmi_3_1 __attribute__((aligned(16))); + __m128i rho_rmi_3_3 __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_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_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_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_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_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_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_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 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_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_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_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_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_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_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_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 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_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_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_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 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_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_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_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 y0_p_1_1 __attribute__((aligned(16))); + __m128i y0_p_1_3 __attribute__((aligned(16))); + __m128i y0_p_3_1 __attribute__((aligned(16))); + __m128i y0_p_3_3 __attribute__((aligned(16))); + __m128i y0_m_1_1 __attribute__((aligned(16))); + __m128i y0_m_1_3 __attribute__((aligned(16))); + __m128i y0_m_3_1 __attribute__((aligned(16))); + __m128i y0_m_3_3 __attribute__((aligned(16))); + + __m128i y0r __attribute__((aligned(16))); + __m128i y0i __attribute__((aligned(16))); + __m128i y1r __attribute__((aligned(16))); + __m128i y1i __attribute__((aligned(16))); + + __m128i tmp_result __attribute__((aligned(16))); + __m128i tmp_result2 __attribute__((aligned(16))); + + // In one iteration, we deal with 8 REs + for (int i = 0; i < length >> 2; i += 2) { + + // Get rho + simde_mm128_separate_real_imag_parts(&xmm2, &xmm3, rho01_128i[i], rho01_128i[i + 1]); + rho_rpi = simde_mm_adds_epi16(xmm2, xmm3); // rho = Re(rho) + Im(rho) + rho_rmi = simde_mm_subs_epi16(xmm2, xmm3); // rho* = Re(rho) - Im(rho) + + // Compute the different rhos + rho_rpi_1_1 = simde_mm_mulhi_epi16(rho_rpi, ONE_OVER_SQRT_10); + rho_rmi_1_1 = simde_mm_mulhi_epi16(rho_rmi, ONE_OVER_SQRT_10); + rho_rpi_3_3 = simde_mm_mulhi_epi16(rho_rpi, THREE_OVER_SQRT_10); + rho_rmi_3_3 = simde_mm_mulhi_epi16(rho_rmi, THREE_OVER_SQRT_10); + rho_rpi_3_3 = simde_mm_slli_epi16(rho_rpi_3_3, 1); + rho_rmi_3_3 = simde_mm_slli_epi16(rho_rmi_3_3, 1); + + xmm4 = simde_mm_mulhi_epi16(xmm2, ONE_OVER_SQRT_10); // Re(rho) + xmm5 = simde_mm_mulhi_epi16(xmm3, THREE_OVER_SQRT_10); // Im(rho) + xmm5 = simde_mm_slli_epi16(xmm5, 1); + + rho_rpi_1_3 = simde_mm_adds_epi16(xmm4, xmm5); + rho_rmi_1_3 = simde_mm_subs_epi16(xmm4, xmm5); + + xmm6 = simde_mm_mulhi_epi16(xmm2, THREE_OVER_SQRT_10); // Re(rho) + xmm7 = simde_mm_mulhi_epi16(xmm3, ONE_OVER_SQRT_10); // Im(rho) + xmm6 = simde_mm_slli_epi16(xmm6, 1); + + rho_rpi_3_1 = simde_mm_adds_epi16(xmm6, xmm7); + rho_rmi_3_1 = simde_mm_subs_epi16(xmm6, xmm7); + + // Rearrange interfering MF output + simde_mm128_separate_real_imag_parts(&y1r, &y1i, stream1_128i_in[i], stream1_128i_in[i + 1]); + + xmm0 = simde_mm_setzero_si128(); // ZERO + xmm2 = simde_mm_subs_epi16(rho_rpi_1_1, y1r); // = [Re(rho)+ Im(rho)]/sqrt(10) - y1r + psi_r_p1_p1 = simde_mm_abs_epi16(xmm2); // = |[Re(rho)+ Im(rho)]/sqrt(10) - y1r| + + xmm2 = simde_mm_subs_epi16(rho_rmi_1_1, y1r); + psi_r_p1_m1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rmi_1_1, y1i); + psi_i_p1_p1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rpi_1_3, y1r); + psi_r_p1_p3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rmi_1_3, y1r); + psi_r_p1_m3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rmi_3_1, y1i); + psi_i_p1_p3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rpi_3_1, y1r); + psi_r_p3_p1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rmi_3_1, y1r); + psi_r_p3_m1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rmi_1_3, y1i); + psi_i_p3_p1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rpi_3_3, y1r); + psi_r_p3_p3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rmi_3_3, y1r); + psi_r_p3_m3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rmi_3_3, y1i); + psi_i_p3_p3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rpi_1_1, y1i); + psi_i_m1_p1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rpi_3_1, y1i); + psi_i_m1_p3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rpi_1_3, y1i); + psi_i_m3_p1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_subs_epi16(rho_rpi_3_3, y1i); + psi_i_m3_p3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(rho_rpi_1_1, y1i); + psi_i_p1_m1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(rho_rpi_3_1, y1i); + psi_i_p1_m3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(rho_rpi_1_3, y1i); + psi_i_p3_m1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(rho_rpi_3_3, y1i); + psi_i_p3_m3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(rho_rpi_1_1, y1r); + psi_r_m1_m1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(rho_rpi_1_3, y1r); + psi_r_m1_m3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(rho_rpi_3_1, y1r); + psi_r_m3_m1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(rho_rpi_3_3, y1r); + psi_r_m3_m3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(y1r, rho_rmi_1_1); + psi_r_m1_p1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(y1r, rho_rmi_1_3); + psi_r_m1_p3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(y1i, rho_rmi_1_1); + psi_i_m1_m1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(y1i, rho_rmi_3_1); + psi_i_m1_m3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(y1r, rho_rmi_3_1); + psi_r_m3_p1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(y1r, rho_rmi_3_3); + psi_r_m3_p3 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(y1i, rho_rmi_1_3); + psi_i_m3_m1 = simde_mm_abs_epi16(xmm2); + xmm2 = simde_mm_adds_epi16(y1i, rho_rmi_3_3); + psi_i_m3_m3 = simde_mm_abs_epi16(xmm2); + + // Rearrange desired MF output + simde_mm128_separate_real_imag_parts(&y0r, &y0i, stream0_128i_in[i], stream0_128i_in[i + 1]); + + // Rearrange desired channel magnitudes + // [|h|^2(1),|h|^2(2),|h|^2(3),|h|^2(4)]*(2/sqrt(10)) + simde_mm128_separate_real_imag_parts(&ch_mag_des, &xmm2, ch_mag_128i[i], ch_mag_128i[i + 1]); + + // Rearrange interfering channel magnitudes + simde_mm128_separate_real_imag_parts(&ch_mag_int, &xmm2, ch_mag_128i_i[i], ch_mag_128i_i[i + 1]); + + // Scale MF output of desired signal + y0r_over_sqrt10 = simde_mm_mulhi_epi16(y0r, ONE_OVER_SQRT_10); + y0i_over_sqrt10 = simde_mm_mulhi_epi16(y0i, ONE_OVER_SQRT_10); + y0r_three_over_sqrt10 = simde_mm_mulhi_epi16(y0r, THREE_OVER_SQRT_10); + y0i_three_over_sqrt10 = simde_mm_mulhi_epi16(y0i, THREE_OVER_SQRT_10); + y0r_three_over_sqrt10 = simde_mm_slli_epi16(y0r_three_over_sqrt10, 1); + y0i_three_over_sqrt10 = simde_mm_slli_epi16(y0i_three_over_sqrt10, 1); + + // Compute necessary combination of required terms + y0_p_1_1 = simde_mm_adds_epi16(y0r_over_sqrt10, y0i_over_sqrt10); + y0_m_1_1 = simde_mm_subs_epi16(y0r_over_sqrt10, y0i_over_sqrt10); + + y0_p_1_3 = simde_mm_adds_epi16(y0r_over_sqrt10, y0i_three_over_sqrt10); + y0_m_1_3 = simde_mm_subs_epi16(y0r_over_sqrt10, y0i_three_over_sqrt10); + + y0_p_3_1 = simde_mm_adds_epi16(y0r_three_over_sqrt10, y0i_over_sqrt10); + y0_m_3_1 = simde_mm_subs_epi16(y0r_three_over_sqrt10, y0i_over_sqrt10); + + y0_p_3_3 = simde_mm_adds_epi16(y0r_three_over_sqrt10, y0i_three_over_sqrt10); + y0_m_3_3 = simde_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 = simde_mm_mulhi_epi16(ch_mag_des, ONE_OVER_TWO_SQRT_10); + ch_mag_over_2 = simde_mm_mulhi_epi16(ch_mag_des, SQRT_10_OVER_FOUR); + ch_mag_over_2 = simde_mm_slli_epi16(ch_mag_over_2, 1); + ch_mag_9_over_10 = simde_mm_mulhi_epi16(ch_mag_des, NINE_OVER_TWO_SQRT_10); + ch_mag_9_over_10 = simde_mm_slli_epi16(ch_mag_9_over_10, 2); + + /// Compute bit metrics (lambda) + + __m128i bit_met_p1_p1 = simde_mm_subs_epi16(psi_a_p1_p1, a_sq_p1_p1); + bit_met_p1_p1 = simde_mm_adds_epi16(bit_met_p1_p1, y0_p_1_1); + bit_met_p1_p1 = simde_mm_subs_epi16(bit_met_p1_p1, ch_mag_over_10); + + __m128i bit_met_p1_p3 = simde_mm_subs_epi16(psi_a_p1_p3, a_sq_p1_p3); + bit_met_p1_p3 = simde_mm_adds_epi16(bit_met_p1_p3, y0_p_1_3); + bit_met_p1_p3 = simde_mm_subs_epi16(bit_met_p1_p3, ch_mag_over_2); + + __m128i bit_met_p1_m1 = simde_mm_subs_epi16(psi_a_p1_m1, a_sq_p1_m1); + bit_met_p1_m1 = simde_mm_adds_epi16(bit_met_p1_m1, y0_m_1_1); + bit_met_p1_m1 = simde_mm_subs_epi16(bit_met_p1_m1, ch_mag_over_10); + + __m128i bit_met_p1_m3 = simde_mm_subs_epi16(psi_a_p1_m3, a_sq_p1_m3); + bit_met_p1_m3 = simde_mm_adds_epi16(bit_met_p1_m3, y0_m_1_3); + bit_met_p1_m3 = simde_mm_subs_epi16(bit_met_p1_m3, ch_mag_over_2); + + __m128i bit_met_p3_p1 = simde_mm_subs_epi16(psi_a_p3_p1, a_sq_p3_p1); + bit_met_p3_p1 = simde_mm_adds_epi16(bit_met_p3_p1, y0_p_3_1); + bit_met_p3_p1 = simde_mm_subs_epi16(bit_met_p3_p1, ch_mag_over_2); + + __m128i bit_met_p3_p3 = simde_mm_subs_epi16(psi_a_p3_p3, a_sq_p3_p3); + bit_met_p3_p3 = simde_mm_adds_epi16(bit_met_p3_p3, y0_p_3_3); + bit_met_p3_p3 = simde_mm_subs_epi16(bit_met_p3_p3, ch_mag_9_over_10); + + __m128i bit_met_p3_m1 = simde_mm_subs_epi16(psi_a_p3_m1, a_sq_p3_m1); + bit_met_p3_m1 = simde_mm_adds_epi16(bit_met_p3_m1, y0_m_3_1); + bit_met_p3_m1 = simde_mm_subs_epi16(bit_met_p3_m1, ch_mag_over_2); + + __m128i bit_met_p3_m3 = simde_mm_subs_epi16(psi_a_p3_m3, a_sq_p3_m3); + bit_met_p3_m3 = simde_mm_adds_epi16(bit_met_p3_m3, y0_m_3_3); + bit_met_p3_m3 = simde_mm_subs_epi16(bit_met_p3_m3, ch_mag_9_over_10); + + __m128i bit_met_m1_p1 = simde_mm_subs_epi16(psi_a_m1_p1, a_sq_m1_p1); + bit_met_m1_p1 = simde_mm_subs_epi16(bit_met_m1_p1, y0_m_1_1); + bit_met_m1_p1 = simde_mm_subs_epi16(bit_met_m1_p1, ch_mag_over_10); + + __m128i bit_met_m1_p3 = simde_mm_subs_epi16(psi_a_m1_p3, a_sq_m1_p3); + bit_met_m1_p3 = simde_mm_subs_epi16(bit_met_m1_p3, y0_m_1_3); + bit_met_m1_p3 = simde_mm_subs_epi16(bit_met_m1_p3, ch_mag_over_2); + + __m128i bit_met_m1_m1 = simde_mm_subs_epi16(psi_a_m1_m1, a_sq_m1_m1); + bit_met_m1_m1 = simde_mm_subs_epi16(bit_met_m1_m1, y0_p_1_1); + bit_met_m1_m1 = simde_mm_subs_epi16(bit_met_m1_m1, ch_mag_over_10); + + __m128i bit_met_m1_m3 = simde_mm_subs_epi16(psi_a_m1_m3, a_sq_m1_m3); + bit_met_m1_m3 = simde_mm_subs_epi16(bit_met_m1_m3, y0_p_1_3); + bit_met_m1_m3 = simde_mm_subs_epi16(bit_met_m1_m3, ch_mag_over_2); + + __m128i bit_met_m3_p1 = simde_mm_subs_epi16(psi_a_m3_p1, a_sq_m3_p1); + bit_met_m3_p1 = simde_mm_subs_epi16(bit_met_m3_p1, y0_m_3_1); + bit_met_m3_p1 = simde_mm_subs_epi16(bit_met_m3_p1, ch_mag_over_2); + + __m128i bit_met_m3_p3 = simde_mm_subs_epi16(psi_a_m3_p3, a_sq_m3_p3); + bit_met_m3_p3 = simde_mm_subs_epi16(bit_met_m3_p3, y0_m_3_3); + bit_met_m3_p3 = simde_mm_subs_epi16(bit_met_m3_p3, ch_mag_9_over_10); + + __m128i bit_met_m3_m1 = simde_mm_subs_epi16(psi_a_m3_m1, a_sq_m3_m1); + bit_met_m3_m1 = simde_mm_subs_epi16(bit_met_m3_m1, y0_p_3_1); + bit_met_m3_m1 = simde_mm_subs_epi16(bit_met_m3_m1, ch_mag_over_2); + + __m128i bit_met_m3_m3 = simde_mm_subs_epi16(psi_a_m3_m3, a_sq_m3_m3); + bit_met_m3_m3 = simde_mm_subs_epi16(bit_met_m3_m3, y0_p_3_3); + bit_met_m3_m3 = simde_mm_subs_epi16(bit_met_m3_m3, ch_mag_9_over_10); + + /// Compute the LLRs + + // LLR = lambda(c==1) - lambda(c==0) + + // LLR of the first bit: Bit = 1 + __m128i logmax_num_re0 = max_epi16(bit_met_m1_p1, + bit_met_m1_p3, + bit_met_m1_m1, + bit_met_m1_m3, + bit_met_m3_p1, + bit_met_m3_p3, + bit_met_m3_m1, + bit_met_m3_m3); + + // LLR of the first bit: Bit = 0 + __m128i logmax_den_re0 = max_epi16(bit_met_p1_p1, + bit_met_p1_p3, + bit_met_p1_m1, + bit_met_p1_m3, + bit_met_p3_p1, + bit_met_p3_p3, + bit_met_p3_m1, + bit_met_p3_m3); + + // LLR of the second bit: Bit = 1 + __m128i logmax_num_re1 = max_epi16(bit_met_p1_m1, + bit_met_p3_m1, + bit_met_m1_m1, + bit_met_m3_m1, + bit_met_p1_m3, + bit_met_p3_m3, + bit_met_m1_m3, + bit_met_m3_m3); + + // LLR of the second bit: Bit = 0 + __m128i logmax_den_re1 = max_epi16(bit_met_p1_p1, + bit_met_p3_p1, + bit_met_m1_p1, + bit_met_m3_p1, + bit_met_p1_p3, + bit_met_p3_p3, + bit_met_m1_p3, + bit_met_m3_p3); + + // LLR of the third bit: Bit = 1 + __m128i logmax_num_im0 = max_epi16(bit_met_m3_p1, + bit_met_m3_p3, + bit_met_m3_m1, + bit_met_m3_m3, + bit_met_p3_p1, + bit_met_p3_p3, + bit_met_p3_m1, + bit_met_p3_m3); + + // LLR of the third bit: Bit = 0 + __m128i logmax_den_im0 = max_epi16(bit_met_m1_p1, + bit_met_m1_p3, + bit_met_m1_m1, + bit_met_m1_m3, + bit_met_p1_p1, + bit_met_p1_p3, + bit_met_p1_m1, + bit_met_p1_m3); + + // LLR of the fourth bit: Bit = 1 + __m128i logmax_num_im1 = max_epi16(bit_met_p1_m3, + bit_met_p3_m3, + bit_met_m1_m3, + bit_met_m3_m3, + bit_met_p1_p3, + bit_met_p3_p3, + bit_met_m1_p3, + bit_met_m3_p3); + + // LLR of the fourth bit: Bit = 0 + __m128i logmax_den_im1 = max_epi16(bit_met_p1_m1, + bit_met_p3_m1, + bit_met_m1_m1, + bit_met_m3_m1, + bit_met_p1_p1, + bit_met_p3_p1, + bit_met_m1_p1, + bit_met_m3_p1); + + y0r = simde_mm_subs_epi16(logmax_den_re0, logmax_num_re0); // LLR of first bit [L1(1), L1(2), L1(3), L1(4), L1(5), L1(6), L1(7), L1(8)] + y1r = simde_mm_subs_epi16(logmax_den_re1, logmax_num_re1); // LLR of second bit [L2(1), L2(2), L2(3), L2(4), L2(5), L2(6), L2(7), L2(8)] + y0i = simde_mm_subs_epi16(logmax_den_im0, logmax_num_im0); // LLR of third bit [L3(1), L3(2), L3(3), L3(4), L3(5), L3(6), L3(7), L3(8)] + y1i = simde_mm_subs_epi16(logmax_den_im1, logmax_num_im1); // LLR of fourth bit [L4(1), L4(2), L4(3), L4(4), L4(5), L4(6), L4(7), L4(8)] + + // Pack LLRs in output + xmm0 = simde_mm_unpacklo_epi16(y0r, y1r); // [L1(1), L2(1), L1(2), L2(2), L1(3), L2(3), L1(4), L2(4)] + xmm1 = simde_mm_unpackhi_epi16(y0r, y1r); // [L1(5), L2(5), L1(6), L2(6), L1(7), L2(7), L1(8), L2(8)] + xmm2 = simde_mm_unpacklo_epi16(y0i, y1i); // [L3(1), L4(1), L3(2), L4(2), L3(3), L4(3), L3(4), L4(4)] + xmm3 = simde_mm_unpackhi_epi16(y0i, y1i); // [L3(5), L4(5), L3(6), L4(6), L3(7), L4(7), L3(8), L4(8)] + stream0_128i_out[2 * i + 0] = simde_mm_unpacklo_epi32(xmm0, xmm2); // 8 LLRs, 2 REs + stream0_128i_out[2 * i + 1] = simde_mm_unpackhi_epi32(xmm0, xmm2); // 8 LLRs, 2 REs + stream0_128i_out[2 * i + 2] = simde_mm_unpacklo_epi32(xmm1, xmm3); // 8 LLRs, 2 REs + stream0_128i_out[2 * i + 3] = simde_mm_unpackhi_epi32(xmm1, xmm3); // 8 LLRs, 2 REs + } + + _mm_empty(); + _m_empty(); +} + +static const int16_t ones256[16] __attribute__((aligned(32))) = {0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff, + 0xffff}; + +/// Auxiliary Makros + +// 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_256(psi, int_ch_mag, int_two_ch_mag, int_three_ch_mag, a, c1, c3, c5, c7) \ + tmp_result = simde_mm256_cmpgt_epi16(int_two_ch_mag, psi); \ + tmp_result3 = simde_mm256_xor_si256(tmp_result, (*(__m256i *)&ones256[0])); \ + tmp_result2 = simde_mm256_cmpgt_epi16(int_ch_mag, psi); \ + tmp_result = simde_mm256_xor_si256(tmp_result, tmp_result2); \ + tmp_result4 = simde_mm256_cmpgt_epi16(psi, int_three_ch_mag); \ + tmp_result3 = simde_mm256_xor_si256(tmp_result3, tmp_result4); \ + tmp_result = simde_mm256_and_si256(tmp_result, c3); \ + tmp_result2 = simde_mm256_and_si256(tmp_result2, c1); \ + tmp_result3 = simde_mm256_and_si256(tmp_result3, c5); \ + tmp_result4 = simde_mm256_and_si256(tmp_result4, c7); \ + tmp_result = simde_mm256_or_si256(tmp_result, tmp_result2); \ + tmp_result3 = simde_mm256_or_si256(tmp_result3, tmp_result4); \ + const simde__m256i a = simde_mm256_or_si256(tmp_result, tmp_result3); + +// calculates psi_a = psi_r*a_r + psi_i*a_i +#define prodsum_psi_a_epi16_256(psi_r, a_r, psi_i, a_i, psi_a) \ + tmp_result = simde_mm256_mulhi_epi16(psi_r, a_r); \ + tmp_result = simde_mm256_slli_epi16(tmp_result, 1); \ + tmp_result2 = simde_mm256_mulhi_epi16(psi_i, a_i); \ + tmp_result2 = simde_mm256_slli_epi16(tmp_result2, 1); \ + simde__m256i psi_a = simde_mm256_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_256(a_r, a_i, int_ch_mag, scale_factor, a_sq) \ + tmp_result = simde_mm256_mulhi_epi16(a_r, a_r); \ + tmp_result = simde_mm256_slli_epi16(tmp_result, 1); \ + tmp_result = simde_mm256_mulhi_epi16(tmp_result, scale_factor); \ + tmp_result = simde_mm256_slli_epi16(tmp_result, 3); \ + tmp_result = simde_mm256_mulhi_epi16(tmp_result, int_ch_mag); \ + tmp_result = simde_mm256_slli_epi16(tmp_result, 1); \ + tmp_result2 = simde_mm256_mulhi_epi16(a_i, a_i); \ + tmp_result2 = simde_mm256_slli_epi16(tmp_result2, 1); \ + tmp_result2 = simde_mm256_mulhi_epi16(tmp_result2, scale_factor); \ + tmp_result2 = simde_mm256_slli_epi16(tmp_result2, 3); \ + tmp_result2 = simde_mm256_mulhi_epi16(tmp_result2, int_ch_mag); \ + tmp_result2 = simde_mm256_slli_epi16(tmp_result2, 1); \ + const simde__m256i a_sq = simde_mm256_adds_epi16(tmp_result, tmp_result2); + +/* + * This function computes the LLRs of stream 0 (s_0) in presence of the interfering stream 1 (s_1) assuming that both symbols are + * 64QAM. It can be used for both MU-MIMO interference-aware receiver or for SU-MIMO receivers. + * + * Input: + * stream0_in: MF filter output for 1st stream, i.e., y0' = h0'*y0 + * stream1_in: MF filter output for 2nd stream, i.e., y1' = h1'*y0 + * ch_mag: 4*h0/sqrt(42), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + * ch_mag_i: 4*h0/sqrt(42), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc + * rho01: Channel cross correlation, i.e., rho01 = h0'*h1 + * length: Number of resource elements + * + * Output: + * stream0_out: Output LLRs for 1st stream + */ +void nr_ulsch_qam64_qam64(c16_t *stream0_in, + c16_t *stream1_in, + c16_t *ch_mag, + c16_t *ch_mag_i, + c16_t *stream0_out, + c16_t *rho01, + uint32_t length) +{ + __m256i *rho01_256i = (__m256i *)rho01; + __m256i *stream0_256i_in = (__m256i *)stream0_in; + __m256i *stream1_256i_in = (__m256i *)stream1_in; + __m256i *ch_mag_256i = (__m256i *)ch_mag; + __m256i *ch_mag_256i_i = (__m256i *)ch_mag_i; + + __m256i ONE_OVER_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(10112)); // round(1/sqrt(42)*2^16) + __m256i THREE_OVER_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(30337)); // round(3/sqrt(42)*2^16) + __m256i FIVE_OVER_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(25281)); // round(5/sqrt(42)*2^15) + __m256i SEVEN_OVER_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(17697)); // round(7/sqrt(42)*2^14) Q2.14 + __m256i ONE_OVER_SQRT_2 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(23170)); // round(1/sqrt(2)*2^15) + __m256i ONE_OVER_SQRT_2_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(3575)); // round(1/sqrt(2*42)*2^15) + __m256i THREE_OVER_SQRT_2_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(10726)); // round(3/sqrt(2*42)*2^15) + __m256i FIVE_OVER_SQRT_2_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(17876)); // round(5/sqrt(2*42)*2^15) + __m256i SEVEN_OVER_SQRT_2_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(25027)); // round(7/sqrt(2*42)*2^15) + __m256i FORTYNINE_OVER_FOUR_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(30969)); // round(49/(4*sqrt(42))*2^14), Q2.14 + __m256i THIRTYSEVEN_OVER_FOUR_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(23385)); // round(37/(4*sqrt(42))*2^14), Q2.14 + __m256i TWENTYFIVE_OVER_FOUR_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(31601)); // round(25/(4*sqrt(42))*2^15) + __m256i TWENTYNINE_OVER_FOUR_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(18329)); // round(29/(4*sqrt(42))*2^15), Q2.14 + __m256i SEVENTEEN_OVER_FOUR_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(21489)); // round(17/(4*sqrt(42))*2^15) + __m256i NINE_OVER_FOUR_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(11376)); // round(9/(4*sqrt(42))*2^15) + __m256i THIRTEEN_OVER_FOUR_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(16433)); // round(13/(4*sqrt(42))*2^15) + __m256i FIVE_OVER_FOUR_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(6320)); // round(5/(4*sqrt(42))*2^15) + __m256i ONE_OVER_FOUR_SQRT_42 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(1264)); // round(1/(4*sqrt(42))*2^15) + __m256i SQRT_42_OVER_FOUR = simde_mm256_broadcastw_epi16(_mm_set1_epi16(13272)); // round(sqrt(42)/4*2^13), Q3.12 + + __m256i ch_mag_des; + __m256i ch_mag_int; + __m256i ch_mag_98_over_42_with_sigma2; + __m256i ch_mag_74_over_42_with_sigma2; + __m256i ch_mag_58_over_42_with_sigma2; + __m256i ch_mag_50_over_42_with_sigma2; + __m256i ch_mag_34_over_42_with_sigma2; + __m256i ch_mag_18_over_42_with_sigma2; + __m256i ch_mag_26_over_42_with_sigma2; + __m256i ch_mag_10_over_42_with_sigma2; + __m256i ch_mag_2_over_42_with_sigma2; + __m256i y0r_one_over_sqrt_21; + __m256i y0r_three_over_sqrt_21; + __m256i y0r_five_over_sqrt_21; + __m256i y0r_seven_over_sqrt_21; + __m256i y0i_one_over_sqrt_21; + __m256i y0i_three_over_sqrt_21; + __m256i y0i_five_over_sqrt_21; + __m256i y0i_seven_over_sqrt_21; + __m256i ch_mag_int_with_sigma2; + __m256i two_ch_mag_int_with_sigma2; + __m256i three_ch_mag_int_with_sigma2; + + uint32_t len256 = length >> 3; + + for (int i = 0; i < len256; i += 2) { + + // Get rho + simde__m256i xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8; + simde_mm256_separate_real_imag_parts(&xmm2, &xmm3, rho01_256i[i], rho01_256i[i + 1]); + + simde__m256i rho_rpi = simde_mm256_adds_epi16(xmm2, xmm3); // rho = Re(rho) + Im(rho) + simde__m256i rho_rmi = simde_mm256_subs_epi16(xmm2, xmm3); // rho* = Re(rho) - Im(rho) + + // Compute the different rhos + simde__m256i rho_rpi_1_1 = simde_mm256_mulhi_epi16(rho_rpi, ONE_OVER_SQRT_42); + simde__m256i rho_rmi_1_1 = simde_mm256_mulhi_epi16(rho_rmi, ONE_OVER_SQRT_42); + simde__m256i rho_rpi_3_3 = simde_mm256_mulhi_epi16(rho_rpi, THREE_OVER_SQRT_42); + simde__m256i rho_rmi_3_3 = simde_mm256_mulhi_epi16(rho_rmi, THREE_OVER_SQRT_42); + simde__m256i rho_rpi_5_5 = simde_mm256_mulhi_epi16(rho_rpi, FIVE_OVER_SQRT_42); + simde__m256i rho_rmi_5_5 = simde_mm256_mulhi_epi16(rho_rmi, FIVE_OVER_SQRT_42); + simde__m256i rho_rpi_7_7 = simde_mm256_mulhi_epi16(rho_rpi, SEVEN_OVER_SQRT_42); + simde__m256i rho_rmi_7_7 = simde_mm256_mulhi_epi16(rho_rmi, SEVEN_OVER_SQRT_42); + + rho_rpi_5_5 = simde_mm256_slli_epi16(rho_rpi_5_5, 1); + rho_rmi_5_5 = simde_mm256_slli_epi16(rho_rmi_5_5, 1); + rho_rpi_7_7 = simde_mm256_slli_epi16(rho_rpi_7_7, 2); + rho_rmi_7_7 = simde_mm256_slli_epi16(rho_rmi_7_7, 2); + + xmm4 = simde_mm256_mulhi_epi16(xmm2, ONE_OVER_SQRT_42); + xmm5 = simde_mm256_mulhi_epi16(xmm3, ONE_OVER_SQRT_42); + xmm6 = simde_mm256_mulhi_epi16(xmm3, THREE_OVER_SQRT_42); + xmm7 = simde_mm256_mulhi_epi16(xmm3, FIVE_OVER_SQRT_42); + xmm8 = simde_mm256_mulhi_epi16(xmm3, SEVEN_OVER_SQRT_42); + xmm7 = simde_mm256_slli_epi16(xmm7, 1); + xmm8 = simde_mm256_slli_epi16(xmm8, 2); + + simde__m256i rho_rpi_1_3 = simde_mm256_adds_epi16(xmm4, xmm6); + simde__m256i rho_rmi_1_3 = simde_mm256_subs_epi16(xmm4, xmm6); + simde__m256i rho_rpi_1_5 = simde_mm256_adds_epi16(xmm4, xmm7); + simde__m256i rho_rmi_1_5 = simde_mm256_subs_epi16(xmm4, xmm7); + simde__m256i rho_rpi_1_7 = simde_mm256_adds_epi16(xmm4, xmm8); + simde__m256i rho_rmi_1_7 = simde_mm256_subs_epi16(xmm4, xmm8); + + xmm4 = simde_mm256_mulhi_epi16(xmm2, THREE_OVER_SQRT_42); + simde__m256i rho_rpi_3_1 = simde_mm256_adds_epi16(xmm4, xmm5); + simde__m256i rho_rmi_3_1 = simde_mm256_subs_epi16(xmm4, xmm5); + simde__m256i rho_rpi_3_5 = simde_mm256_adds_epi16(xmm4, xmm7); + simde__m256i rho_rmi_3_5 = simde_mm256_subs_epi16(xmm4, xmm7); + simde__m256i rho_rpi_3_7 = simde_mm256_adds_epi16(xmm4, xmm8); + simde__m256i rho_rmi_3_7 = simde_mm256_subs_epi16(xmm4, xmm8); + + xmm4 = simde_mm256_mulhi_epi16(xmm2, FIVE_OVER_SQRT_42); + xmm4 = simde_mm256_slli_epi16(xmm4, 1); + simde__m256i rho_rpi_5_1 = simde_mm256_adds_epi16(xmm4, xmm5); + simde__m256i rho_rmi_5_1 = simde_mm256_subs_epi16(xmm4, xmm5); + simde__m256i rho_rpi_5_3 = simde_mm256_adds_epi16(xmm4, xmm6); + simde__m256i rho_rmi_5_3 = simde_mm256_subs_epi16(xmm4, xmm6); + simde__m256i rho_rpi_5_7 = simde_mm256_adds_epi16(xmm4, xmm8); + simde__m256i rho_rmi_5_7 = simde_mm256_subs_epi16(xmm4, xmm8); + + xmm4 = simde_mm256_mulhi_epi16(xmm2, SEVEN_OVER_SQRT_42); + xmm4 = simde_mm256_slli_epi16(xmm4, 2); + simde__m256i rho_rpi_7_1 = simde_mm256_adds_epi16(xmm4, xmm5); + simde__m256i rho_rmi_7_1 = simde_mm256_subs_epi16(xmm4, xmm5); + simde__m256i rho_rpi_7_3 = simde_mm256_adds_epi16(xmm4, xmm6); + simde__m256i rho_rmi_7_3 = simde_mm256_subs_epi16(xmm4, xmm6); + simde__m256i rho_rpi_7_5 = simde_mm256_adds_epi16(xmm4, xmm7); + simde__m256i rho_rmi_7_5 = simde_mm256_subs_epi16(xmm4, xmm7); + + // Rearrange interfering MF output + simde__m256i y1r, y1i, xmm0; + simde_mm256_separate_real_imag_parts(&y1r, &y1i, stream1_256i_in[i], stream1_256i_in[i + 1]); + + // Psi_r calculation from rho_rpi or rho_rmi + xmm0 = simde_mm256_broadcastw_epi16(_mm_set1_epi16(0)); // ZERO for abs_pi16 + xmm2 = simde_mm256_subs_epi16(rho_rpi_7_7, y1r); + + simde__m256i psi_r_p7_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_7_5, y1r); + simde__m256i psi_r_p7_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_7_3, y1r); + simde__m256i psi_r_p7_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_7_1, y1r); + simde__m256i psi_r_p7_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_7_1, y1r); + simde__m256i psi_r_p7_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_7_3, y1r); + simde__m256i psi_r_p7_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_7_5, y1r); + simde__m256i psi_r_p7_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_7_7, y1r); + simde__m256i psi_r_p7_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_5_7, y1r); + simde__m256i psi_r_p5_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_5_5, y1r); + simde__m256i psi_r_p5_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_5_3, y1r); + simde__m256i psi_r_p5_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_5_1, y1r); + simde__m256i psi_r_p5_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_5_1, y1r); + simde__m256i psi_r_p5_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_5_3, y1r); + simde__m256i psi_r_p5_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_5_5, y1r); + simde__m256i psi_r_p5_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_5_7, y1r); + simde__m256i psi_r_p5_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_3_7, y1r); + simde__m256i psi_r_p3_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_3_5, y1r); + simde__m256i psi_r_p3_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_3_3, y1r); + simde__m256i psi_r_p3_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_3_1, y1r); + simde__m256i psi_r_p3_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_3_1, y1r); + simde__m256i psi_r_p3_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_3_3, y1r); + simde__m256i psi_r_p3_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_3_5, y1r); + simde__m256i psi_r_p3_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_3_7, y1r); + simde__m256i psi_r_p3_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_1_7, y1r); + simde__m256i psi_r_p1_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_1_5, y1r); + simde__m256i psi_r_p1_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_1_3, y1r); + simde__m256i psi_r_p1_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_1_1, y1r); + simde__m256i psi_r_p1_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_1_1, y1r); + simde__m256i psi_r_p1_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_1_3, y1r); + simde__m256i psi_r_p1_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_1_5, y1r); + simde__m256i psi_r_p1_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_1_7, y1r); + simde__m256i psi_r_p1_m7 = simde_mm256_abs_epi16(xmm2); + + xmm2 = simde_mm256_adds_epi16(rho_rmi_1_7, y1r); + simde__m256i psi_r_m1_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_1_5, y1r); + simde__m256i psi_r_m1_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_1_3, y1r); + simde__m256i psi_r_m1_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_1_1, y1r); + simde__m256i psi_r_m1_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_1_1, y1r); + simde__m256i psi_r_m1_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_1_3, y1r); + simde__m256i psi_r_m1_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_1_5, y1r); + simde__m256i psi_r_m1_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_1_7, y1r); + simde__m256i psi_r_m1_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_3_7, y1r); + simde__m256i psi_r_m3_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_3_5, y1r); + simde__m256i psi_r_m3_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_3_3, y1r); + simde__m256i psi_r_m3_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_3_1, y1r); + simde__m256i psi_r_m3_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_3_1, y1r); + simde__m256i psi_r_m3_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_3_3, y1r); + simde__m256i psi_r_m3_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_3_5, y1r); + simde__m256i psi_r_m3_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_3_7, y1r); + simde__m256i psi_r_m3_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_5_7, y1r); + simde__m256i psi_r_m5_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_5_5, y1r); + simde__m256i psi_r_m5_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_5_3, y1r); + simde__m256i psi_r_m5_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_5_1, y1r); + simde__m256i psi_r_m5_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_5_1, y1r); + simde__m256i psi_r_m5_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_5_3, y1r); + simde__m256i psi_r_m5_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_5_5, y1r); + simde__m256i psi_r_m5_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_5_7, y1r); + simde__m256i psi_r_m5_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_7_7, y1r); + simde__m256i psi_r_m7_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_7_5, y1r); + simde__m256i psi_r_m7_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_7_3, y1r); + simde__m256i psi_r_m7_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_7_1, y1r); + simde__m256i psi_r_m7_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_7_1, y1r); + simde__m256i psi_r_m7_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_7_3, y1r); + simde__m256i psi_r_m7_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_7_5, y1r); + simde__m256i psi_r_m7_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_7_7, y1r); + simde__m256i psi_r_m7_m7 = simde_mm256_abs_epi16(xmm2); + + // Simde__M256i Psi_i calculation from rho_rpi or rho_rmi + xmm2 = simde_mm256_subs_epi16(rho_rmi_7_7, y1i); + simde__m256i psi_i_p7_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_5_7, y1i); + simde__m256i psi_i_p7_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_3_7, y1i); + simde__m256i psi_i_p7_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_1_7, y1i); + simde__m256i psi_i_p7_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_1_7, y1i); + simde__m256i psi_i_p7_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_3_7, y1i); + simde__m256i psi_i_p7_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_5_7, y1i); + simde__m256i psi_i_p7_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_7_7, y1i); + simde__m256i psi_i_p7_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_7_5, y1i); + simde__m256i psi_i_p5_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_5_5, y1i); + simde__m256i psi_i_p5_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_3_5, y1i); + simde__m256i psi_i_p5_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_1_5, y1i); + simde__m256i psi_i_p5_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_1_5, y1i); + simde__m256i psi_i_p5_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_3_5, y1i); + simde__m256i psi_i_p5_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_5_5, y1i); + simde__m256i psi_i_p5_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_7_5, y1i); + simde__m256i psi_i_p5_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_7_3, y1i); + simde__m256i psi_i_p3_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_5_3, y1i); + simde__m256i psi_i_p3_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_3_3, y1i); + simde__m256i psi_i_p3_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_1_3, y1i); + simde__m256i psi_i_p3_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_1_3, y1i); + simde__m256i psi_i_p3_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_3_3, y1i); + simde__m256i psi_i_p3_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_5_3, y1i); + simde__m256i psi_i_p3_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_7_3, y1i); + simde__m256i psi_i_p3_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_7_1, y1i); + simde__m256i psi_i_p1_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_5_1, y1i); + simde__m256i psi_i_p1_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_3_1, y1i); + simde__m256i psi_i_p1_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rmi_1_1, y1i); + simde__m256i psi_i_p1_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_1_1, y1i); + simde__m256i psi_i_p1_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_3_1, y1i); + simde__m256i psi_i_p1_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_5_1, y1i); + simde__m256i psi_i_p1_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rpi_7_1, y1i); + simde__m256i psi_i_p1_m7 = simde_mm256_abs_epi16(xmm2); + + xmm2 = simde_mm256_subs_epi16(rho_rpi_7_1, y1i); + simde__m256i psi_i_m1_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_5_1, y1i); + simde__m256i psi_i_m1_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_3_1, y1i); + simde__m256i psi_i_m1_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_1_1, y1i); + simde__m256i psi_i_m1_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_1_1, y1i); + simde__m256i psi_i_m1_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_3_1, y1i); + simde__m256i psi_i_m1_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_5_1, y1i); + simde__m256i psi_i_m1_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_7_1, y1i); + simde__m256i psi_i_m1_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_7_3, y1i); + simde__m256i psi_i_m3_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_5_3, y1i); + simde__m256i psi_i_m3_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_3_3, y1i); + simde__m256i psi_i_m3_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_1_3, y1i); + simde__m256i psi_i_m3_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_1_3, y1i); + simde__m256i psi_i_m3_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_3_3, y1i); + simde__m256i psi_i_m3_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_5_3, y1i); + simde__m256i psi_i_m3_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_7_3, y1i); + simde__m256i psi_i_m3_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_7_5, y1i); + simde__m256i psi_i_m5_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_5_5, y1i); + simde__m256i psi_i_m5_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_3_5, y1i); + simde__m256i psi_i_m5_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_1_5, y1i); + simde__m256i psi_i_m5_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_1_5, y1i); + simde__m256i psi_i_m5_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_3_5, y1i); + simde__m256i psi_i_m5_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_5_5, y1i); + simde__m256i psi_i_m5_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_7_5, y1i); + simde__m256i psi_i_m5_m7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_7_7, y1i); + simde__m256i psi_i_m7_p7 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_5_7, y1i); + simde__m256i psi_i_m7_p5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_3_7, y1i); + simde__m256i psi_i_m7_p3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_subs_epi16(rho_rpi_1_7, y1i); + simde__m256i psi_i_m7_p1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_1_7, y1i); + simde__m256i psi_i_m7_m1 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_3_7, y1i); + simde__m256i psi_i_m7_m3 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_5_7, y1i); + simde__m256i psi_i_m7_m5 = simde_mm256_abs_epi16(xmm2); + xmm2 = simde_mm256_adds_epi16(rho_rmi_7_7, y1i); + simde__m256i psi_i_m7_m7 = simde_mm256_abs_epi16(xmm2); + + // Rearrange desired MF output + simde__m256i y0r, y0i; + simde_mm256_separate_real_imag_parts(&y0r, &y0i, stream0_256i_in[i], stream0_256i_in[i + 1]); + + // Rearrange desired channel magnitudes + // [|h|^2(1),|h|^2(1),|h|^2(2),|h|^2(2),...,,|h|^2(7),|h|^2(7)]*(2/sqrt(10)) + // xmm2 is dummy variable that contains the same values as ch_mag_des + simde_mm256_separate_real_imag_parts(&ch_mag_des, &xmm2, ch_mag_256i[i], ch_mag_256i[i + 1]); + + // Rearrange interfering channel magnitudes + simde_mm256_separate_real_imag_parts(&ch_mag_int, &xmm2, ch_mag_256i_i[i], ch_mag_256i_i[i + 1]); + + y0r_one_over_sqrt_21 = simde_mm256_mulhi_epi16(y0r, ONE_OVER_SQRT_42); + y0r_three_over_sqrt_21 = simde_mm256_mulhi_epi16(y0r, THREE_OVER_SQRT_42); + y0r_five_over_sqrt_21 = simde_mm256_mulhi_epi16(y0r, FIVE_OVER_SQRT_42); + y0r_five_over_sqrt_21 = simde_mm256_slli_epi16(y0r_five_over_sqrt_21, 1); + y0r_seven_over_sqrt_21 = simde_mm256_mulhi_epi16(y0r, SEVEN_OVER_SQRT_42); + y0r_seven_over_sqrt_21 = simde_mm256_slli_epi16(y0r_seven_over_sqrt_21, 2); // Q2.14 + + y0i_one_over_sqrt_21 = simde_mm256_mulhi_epi16(y0i, ONE_OVER_SQRT_42); + y0i_three_over_sqrt_21 = simde_mm256_mulhi_epi16(y0i, THREE_OVER_SQRT_42); + y0i_five_over_sqrt_21 = simde_mm256_mulhi_epi16(y0i, FIVE_OVER_SQRT_42); + y0i_five_over_sqrt_21 = simde_mm256_slli_epi16(y0i_five_over_sqrt_21, 1); + y0i_seven_over_sqrt_21 = simde_mm256_mulhi_epi16(y0i, SEVEN_OVER_SQRT_42); + y0i_seven_over_sqrt_21 = simde_mm256_slli_epi16(y0i_seven_over_sqrt_21, 2); // Q2.14 + + simde__m256i y0_p_7_1 = simde_mm256_adds_epi16(y0r_seven_over_sqrt_21, y0i_one_over_sqrt_21); + simde__m256i y0_p_7_3 = simde_mm256_adds_epi16(y0r_seven_over_sqrt_21, y0i_three_over_sqrt_21); + simde__m256i y0_p_7_5 = simde_mm256_adds_epi16(y0r_seven_over_sqrt_21, y0i_five_over_sqrt_21); + simde__m256i y0_p_7_7 = simde_mm256_adds_epi16(y0r_seven_over_sqrt_21, y0i_seven_over_sqrt_21); + simde__m256i y0_p_5_1 = simde_mm256_adds_epi16(y0r_five_over_sqrt_21, y0i_one_over_sqrt_21); + simde__m256i y0_p_5_3 = simde_mm256_adds_epi16(y0r_five_over_sqrt_21, y0i_three_over_sqrt_21); + simde__m256i y0_p_5_5 = simde_mm256_adds_epi16(y0r_five_over_sqrt_21, y0i_five_over_sqrt_21); + simde__m256i y0_p_5_7 = simde_mm256_adds_epi16(y0r_five_over_sqrt_21, y0i_seven_over_sqrt_21); + simde__m256i y0_p_3_1 = simde_mm256_adds_epi16(y0r_three_over_sqrt_21, y0i_one_over_sqrt_21); + simde__m256i y0_p_3_3 = simde_mm256_adds_epi16(y0r_three_over_sqrt_21, y0i_three_over_sqrt_21); + simde__m256i y0_p_3_5 = simde_mm256_adds_epi16(y0r_three_over_sqrt_21, y0i_five_over_sqrt_21); + simde__m256i y0_p_3_7 = simde_mm256_adds_epi16(y0r_three_over_sqrt_21, y0i_seven_over_sqrt_21); + simde__m256i y0_p_1_1 = simde_mm256_adds_epi16(y0r_one_over_sqrt_21, y0i_one_over_sqrt_21); + simde__m256i y0_p_1_3 = simde_mm256_adds_epi16(y0r_one_over_sqrt_21, y0i_three_over_sqrt_21); + simde__m256i y0_p_1_5 = simde_mm256_adds_epi16(y0r_one_over_sqrt_21, y0i_five_over_sqrt_21); + simde__m256i y0_p_1_7 = simde_mm256_adds_epi16(y0r_one_over_sqrt_21, y0i_seven_over_sqrt_21); + + simde__m256i y0_m_1_1 = simde_mm256_subs_epi16(y0r_one_over_sqrt_21, y0i_one_over_sqrt_21); + simde__m256i y0_m_1_3 = simde_mm256_subs_epi16(y0r_one_over_sqrt_21, y0i_three_over_sqrt_21); + simde__m256i y0_m_1_5 = simde_mm256_subs_epi16(y0r_one_over_sqrt_21, y0i_five_over_sqrt_21); + simde__m256i y0_m_1_7 = simde_mm256_subs_epi16(y0r_one_over_sqrt_21, y0i_seven_over_sqrt_21); + simde__m256i y0_m_3_1 = simde_mm256_subs_epi16(y0r_three_over_sqrt_21, y0i_one_over_sqrt_21); + simde__m256i y0_m_3_3 = simde_mm256_subs_epi16(y0r_three_over_sqrt_21, y0i_three_over_sqrt_21); + simde__m256i y0_m_3_5 = simde_mm256_subs_epi16(y0r_three_over_sqrt_21, y0i_five_over_sqrt_21); + simde__m256i y0_m_3_7 = simde_mm256_subs_epi16(y0r_three_over_sqrt_21, y0i_seven_over_sqrt_21); + simde__m256i y0_m_5_1 = simde_mm256_subs_epi16(y0r_five_over_sqrt_21, y0i_one_over_sqrt_21); + simde__m256i y0_m_5_3 = simde_mm256_subs_epi16(y0r_five_over_sqrt_21, y0i_three_over_sqrt_21); + simde__m256i y0_m_5_5 = simde_mm256_subs_epi16(y0r_five_over_sqrt_21, y0i_five_over_sqrt_21); + simde__m256i y0_m_5_7 = simde_mm256_subs_epi16(y0r_five_over_sqrt_21, y0i_seven_over_sqrt_21); + simde__m256i y0_m_7_1 = simde_mm256_subs_epi16(y0r_seven_over_sqrt_21, y0i_one_over_sqrt_21); + simde__m256i y0_m_7_3 = simde_mm256_subs_epi16(y0r_seven_over_sqrt_21, y0i_three_over_sqrt_21); + simde__m256i y0_m_7_5 = simde_mm256_subs_epi16(y0r_seven_over_sqrt_21, y0i_five_over_sqrt_21); + simde__m256i y0_m_7_7 = simde_mm256_subs_epi16(y0r_seven_over_sqrt_21, y0i_seven_over_sqrt_21); + + // Detection of interference term + ch_mag_int_with_sigma2 = simde_mm256_srai_epi16(ch_mag_int, 1); // *2 + two_ch_mag_int_with_sigma2 = ch_mag_int; // *4 + three_ch_mag_int_with_sigma2 = simde_mm256_adds_epi16(ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2); // *6 + simde__m256i tmp_result, tmp_result2, tmp_result3, tmp_result4; + interference_abs_64qam_epi16_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(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_256(psi_r_p7_p7, a_r_p7_p7, psi_i_p7_p7, a_i_p7_p7, psi_a_p7_p7); + prodsum_psi_a_epi16_256(psi_r_p7_p5, a_r_p7_p5, psi_i_p7_p5, a_i_p7_p5, psi_a_p7_p5); + prodsum_psi_a_epi16_256(psi_r_p7_p3, a_r_p7_p3, psi_i_p7_p3, a_i_p7_p3, psi_a_p7_p3); + prodsum_psi_a_epi16_256(psi_r_p7_p1, a_r_p7_p1, psi_i_p7_p1, a_i_p7_p1, psi_a_p7_p1); + prodsum_psi_a_epi16_256(psi_r_p7_m1, a_r_p7_m1, psi_i_p7_m1, a_i_p7_m1, psi_a_p7_m1); + prodsum_psi_a_epi16_256(psi_r_p7_m3, a_r_p7_m3, psi_i_p7_m3, a_i_p7_m3, psi_a_p7_m3); + prodsum_psi_a_epi16_256(psi_r_p7_m5, a_r_p7_m5, psi_i_p7_m5, a_i_p7_m5, psi_a_p7_m5); + prodsum_psi_a_epi16_256(psi_r_p7_m7, a_r_p7_m7, psi_i_p7_m7, a_i_p7_m7, psi_a_p7_m7); + prodsum_psi_a_epi16_256(psi_r_p5_p7, a_r_p5_p7, psi_i_p5_p7, a_i_p5_p7, psi_a_p5_p7); + prodsum_psi_a_epi16_256(psi_r_p5_p5, a_r_p5_p5, psi_i_p5_p5, a_i_p5_p5, psi_a_p5_p5); + prodsum_psi_a_epi16_256(psi_r_p5_p3, a_r_p5_p3, psi_i_p5_p3, a_i_p5_p3, psi_a_p5_p3); + prodsum_psi_a_epi16_256(psi_r_p5_p1, a_r_p5_p1, psi_i_p5_p1, a_i_p5_p1, psi_a_p5_p1); + prodsum_psi_a_epi16_256(psi_r_p5_m1, a_r_p5_m1, psi_i_p5_m1, a_i_p5_m1, psi_a_p5_m1); + prodsum_psi_a_epi16_256(psi_r_p5_m3, a_r_p5_m3, psi_i_p5_m3, a_i_p5_m3, psi_a_p5_m3); + prodsum_psi_a_epi16_256(psi_r_p5_m5, a_r_p5_m5, psi_i_p5_m5, a_i_p5_m5, psi_a_p5_m5); + prodsum_psi_a_epi16_256(psi_r_p5_m7, a_r_p5_m7, psi_i_p5_m7, a_i_p5_m7, psi_a_p5_m7); + prodsum_psi_a_epi16_256(psi_r_p3_p7, a_r_p3_p7, psi_i_p3_p7, a_i_p3_p7, psi_a_p3_p7); + prodsum_psi_a_epi16_256(psi_r_p3_p5, a_r_p3_p5, psi_i_p3_p5, a_i_p3_p5, psi_a_p3_p5); + prodsum_psi_a_epi16_256(psi_r_p3_p3, a_r_p3_p3, psi_i_p3_p3, a_i_p3_p3, psi_a_p3_p3); + prodsum_psi_a_epi16_256(psi_r_p3_p1, a_r_p3_p1, psi_i_p3_p1, a_i_p3_p1, psi_a_p3_p1); + prodsum_psi_a_epi16_256(psi_r_p3_m1, a_r_p3_m1, psi_i_p3_m1, a_i_p3_m1, psi_a_p3_m1); + prodsum_psi_a_epi16_256(psi_r_p3_m3, a_r_p3_m3, psi_i_p3_m3, a_i_p3_m3, psi_a_p3_m3); + prodsum_psi_a_epi16_256(psi_r_p3_m5, a_r_p3_m5, psi_i_p3_m5, a_i_p3_m5, psi_a_p3_m5); + prodsum_psi_a_epi16_256(psi_r_p3_m7, a_r_p3_m7, psi_i_p3_m7, a_i_p3_m7, psi_a_p3_m7); + prodsum_psi_a_epi16_256(psi_r_p1_p7, a_r_p1_p7, psi_i_p1_p7, a_i_p1_p7, psi_a_p1_p7); + prodsum_psi_a_epi16_256(psi_r_p1_p5, a_r_p1_p5, psi_i_p1_p5, a_i_p1_p5, psi_a_p1_p5); + prodsum_psi_a_epi16_256(psi_r_p1_p3, a_r_p1_p3, psi_i_p1_p3, a_i_p1_p3, psi_a_p1_p3); + prodsum_psi_a_epi16_256(psi_r_p1_p1, a_r_p1_p1, psi_i_p1_p1, a_i_p1_p1, psi_a_p1_p1); + prodsum_psi_a_epi16_256(psi_r_p1_m1, a_r_p1_m1, psi_i_p1_m1, a_i_p1_m1, psi_a_p1_m1); + prodsum_psi_a_epi16_256(psi_r_p1_m3, a_r_p1_m3, psi_i_p1_m3, a_i_p1_m3, psi_a_p1_m3); + prodsum_psi_a_epi16_256(psi_r_p1_m5, a_r_p1_m5, psi_i_p1_m5, a_i_p1_m5, psi_a_p1_m5); + prodsum_psi_a_epi16_256(psi_r_p1_m7, a_r_p1_m7, psi_i_p1_m7, a_i_p1_m7, psi_a_p1_m7); + prodsum_psi_a_epi16_256(psi_r_m1_p7, a_r_m1_p7, psi_i_m1_p7, a_i_m1_p7, psi_a_m1_p7); + prodsum_psi_a_epi16_256(psi_r_m1_p5, a_r_m1_p5, psi_i_m1_p5, a_i_m1_p5, psi_a_m1_p5); + prodsum_psi_a_epi16_256(psi_r_m1_p3, a_r_m1_p3, psi_i_m1_p3, a_i_m1_p3, psi_a_m1_p3); + prodsum_psi_a_epi16_256(psi_r_m1_p1, a_r_m1_p1, psi_i_m1_p1, a_i_m1_p1, psi_a_m1_p1); + prodsum_psi_a_epi16_256(psi_r_m1_m1, a_r_m1_m1, psi_i_m1_m1, a_i_m1_m1, psi_a_m1_m1); + prodsum_psi_a_epi16_256(psi_r_m1_m3, a_r_m1_m3, psi_i_m1_m3, a_i_m1_m3, psi_a_m1_m3); + prodsum_psi_a_epi16_256(psi_r_m1_m5, a_r_m1_m5, psi_i_m1_m5, a_i_m1_m5, psi_a_m1_m5); + prodsum_psi_a_epi16_256(psi_r_m1_m7, a_r_m1_m7, psi_i_m1_m7, a_i_m1_m7, psi_a_m1_m7); + prodsum_psi_a_epi16_256(psi_r_m3_p7, a_r_m3_p7, psi_i_m3_p7, a_i_m3_p7, psi_a_m3_p7); + prodsum_psi_a_epi16_256(psi_r_m3_p5, a_r_m3_p5, psi_i_m3_p5, a_i_m3_p5, psi_a_m3_p5); + prodsum_psi_a_epi16_256(psi_r_m3_p3, a_r_m3_p3, psi_i_m3_p3, a_i_m3_p3, psi_a_m3_p3); + prodsum_psi_a_epi16_256(psi_r_m3_p1, a_r_m3_p1, psi_i_m3_p1, a_i_m3_p1, psi_a_m3_p1); + prodsum_psi_a_epi16_256(psi_r_m3_m1, a_r_m3_m1, psi_i_m3_m1, a_i_m3_m1, psi_a_m3_m1); + prodsum_psi_a_epi16_256(psi_r_m3_m3, a_r_m3_m3, psi_i_m3_m3, a_i_m3_m3, psi_a_m3_m3); + prodsum_psi_a_epi16_256(psi_r_m3_m5, a_r_m3_m5, psi_i_m3_m5, a_i_m3_m5, psi_a_m3_m5); + prodsum_psi_a_epi16_256(psi_r_m3_m7, a_r_m3_m7, psi_i_m3_m7, a_i_m3_m7, psi_a_m3_m7); + prodsum_psi_a_epi16_256(psi_r_m5_p7, a_r_m5_p7, psi_i_m5_p7, a_i_m5_p7, psi_a_m5_p7); + prodsum_psi_a_epi16_256(psi_r_m5_p5, a_r_m5_p5, psi_i_m5_p5, a_i_m5_p5, psi_a_m5_p5); + prodsum_psi_a_epi16_256(psi_r_m5_p3, a_r_m5_p3, psi_i_m5_p3, a_i_m5_p3, psi_a_m5_p3); + prodsum_psi_a_epi16_256(psi_r_m5_p1, a_r_m5_p1, psi_i_m5_p1, a_i_m5_p1, psi_a_m5_p1); + prodsum_psi_a_epi16_256(psi_r_m5_m1, a_r_m5_m1, psi_i_m5_m1, a_i_m5_m1, psi_a_m5_m1); + prodsum_psi_a_epi16_256(psi_r_m5_m3, a_r_m5_m3, psi_i_m5_m3, a_i_m5_m3, psi_a_m5_m3); + prodsum_psi_a_epi16_256(psi_r_m5_m5, a_r_m5_m5, psi_i_m5_m5, a_i_m5_m5, psi_a_m5_m5); + prodsum_psi_a_epi16_256(psi_r_m5_m7, a_r_m5_m7, psi_i_m5_m7, a_i_m5_m7, psi_a_m5_m7); + prodsum_psi_a_epi16_256(psi_r_m7_p7, a_r_m7_p7, psi_i_m7_p7, a_i_m7_p7, psi_a_m7_p7); + prodsum_psi_a_epi16_256(psi_r_m7_p5, a_r_m7_p5, psi_i_m7_p5, a_i_m7_p5, psi_a_m7_p5); + prodsum_psi_a_epi16_256(psi_r_m7_p3, a_r_m7_p3, psi_i_m7_p3, a_i_m7_p3, psi_a_m7_p3); + prodsum_psi_a_epi16_256(psi_r_m7_p1, a_r_m7_p1, psi_i_m7_p1, a_i_m7_p1, psi_a_m7_p1); + prodsum_psi_a_epi16_256(psi_r_m7_m1, a_r_m7_m1, psi_i_m7_m1, a_i_m7_m1, psi_a_m7_m1); + prodsum_psi_a_epi16_256(psi_r_m7_m3, a_r_m7_m3, psi_i_m7_m3, a_i_m7_m3, psi_a_m7_m3); + prodsum_psi_a_epi16_256(psi_r_m7_m5, a_r_m7_m5, psi_i_m7_m5, a_i_m7_m5, psi_a_m7_m5); + prodsum_psi_a_epi16_256(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 = simde_mm256_mulhi_epi16(psi_a_p7_p7, ONE_OVER_SQRT_2); + psi_a_p7_p7 = simde_mm256_slli_epi16(psi_a_p7_p7, 2); + psi_a_p7_p5 = simde_mm256_mulhi_epi16(psi_a_p7_p5, ONE_OVER_SQRT_2); + psi_a_p7_p5 = simde_mm256_slli_epi16(psi_a_p7_p5, 2); + psi_a_p7_p3 = simde_mm256_mulhi_epi16(psi_a_p7_p3, ONE_OVER_SQRT_2); + psi_a_p7_p3 = simde_mm256_slli_epi16(psi_a_p7_p3, 2); + psi_a_p7_p1 = simde_mm256_mulhi_epi16(psi_a_p7_p1, ONE_OVER_SQRT_2); + psi_a_p7_p1 = simde_mm256_slli_epi16(psi_a_p7_p1, 2); + psi_a_p7_m1 = simde_mm256_mulhi_epi16(psi_a_p7_m1, ONE_OVER_SQRT_2); + psi_a_p7_m1 = simde_mm256_slli_epi16(psi_a_p7_m1, 2); + psi_a_p7_m3 = simde_mm256_mulhi_epi16(psi_a_p7_m3, ONE_OVER_SQRT_2); + psi_a_p7_m3 = simde_mm256_slli_epi16(psi_a_p7_m3, 2); + psi_a_p7_m5 = simde_mm256_mulhi_epi16(psi_a_p7_m5, ONE_OVER_SQRT_2); + psi_a_p7_m5 = simde_mm256_slli_epi16(psi_a_p7_m5, 2); + psi_a_p7_m7 = simde_mm256_mulhi_epi16(psi_a_p7_m7, ONE_OVER_SQRT_2); + psi_a_p7_m7 = simde_mm256_slli_epi16(psi_a_p7_m7, 2); + psi_a_p5_p7 = simde_mm256_mulhi_epi16(psi_a_p5_p7, ONE_OVER_SQRT_2); + psi_a_p5_p7 = simde_mm256_slli_epi16(psi_a_p5_p7, 2); + psi_a_p5_p5 = simde_mm256_mulhi_epi16(psi_a_p5_p5, ONE_OVER_SQRT_2); + psi_a_p5_p5 = simde_mm256_slli_epi16(psi_a_p5_p5, 2); + psi_a_p5_p3 = simde_mm256_mulhi_epi16(psi_a_p5_p3, ONE_OVER_SQRT_2); + psi_a_p5_p3 = simde_mm256_slli_epi16(psi_a_p5_p3, 2); + psi_a_p5_p1 = simde_mm256_mulhi_epi16(psi_a_p5_p1, ONE_OVER_SQRT_2); + psi_a_p5_p1 = simde_mm256_slli_epi16(psi_a_p5_p1, 2); + psi_a_p5_m1 = simde_mm256_mulhi_epi16(psi_a_p5_m1, ONE_OVER_SQRT_2); + psi_a_p5_m1 = simde_mm256_slli_epi16(psi_a_p5_m1, 2); + psi_a_p5_m3 = simde_mm256_mulhi_epi16(psi_a_p5_m3, ONE_OVER_SQRT_2); + psi_a_p5_m3 = simde_mm256_slli_epi16(psi_a_p5_m3, 2); + psi_a_p5_m5 = simde_mm256_mulhi_epi16(psi_a_p5_m5, ONE_OVER_SQRT_2); + psi_a_p5_m5 = simde_mm256_slli_epi16(psi_a_p5_m5, 2); + psi_a_p5_m7 = simde_mm256_mulhi_epi16(psi_a_p5_m7, ONE_OVER_SQRT_2); + psi_a_p5_m7 = simde_mm256_slli_epi16(psi_a_p5_m7, 2); + psi_a_p3_p7 = simde_mm256_mulhi_epi16(psi_a_p3_p7, ONE_OVER_SQRT_2); + psi_a_p3_p7 = simde_mm256_slli_epi16(psi_a_p3_p7, 2); + psi_a_p3_p5 = simde_mm256_mulhi_epi16(psi_a_p3_p5, ONE_OVER_SQRT_2); + psi_a_p3_p5 = simde_mm256_slli_epi16(psi_a_p3_p5, 2); + psi_a_p3_p3 = simde_mm256_mulhi_epi16(psi_a_p3_p3, ONE_OVER_SQRT_2); + psi_a_p3_p3 = simde_mm256_slli_epi16(psi_a_p3_p3, 2); + psi_a_p3_p1 = simde_mm256_mulhi_epi16(psi_a_p3_p1, ONE_OVER_SQRT_2); + psi_a_p3_p1 = simde_mm256_slli_epi16(psi_a_p3_p1, 2); + psi_a_p3_m1 = simde_mm256_mulhi_epi16(psi_a_p3_m1, ONE_OVER_SQRT_2); + psi_a_p3_m1 = simde_mm256_slli_epi16(psi_a_p3_m1, 2); + psi_a_p3_m3 = simde_mm256_mulhi_epi16(psi_a_p3_m3, ONE_OVER_SQRT_2); + psi_a_p3_m3 = simde_mm256_slli_epi16(psi_a_p3_m3, 2); + psi_a_p3_m5 = simde_mm256_mulhi_epi16(psi_a_p3_m5, ONE_OVER_SQRT_2); + psi_a_p3_m5 = simde_mm256_slli_epi16(psi_a_p3_m5, 2); + psi_a_p3_m7 = simde_mm256_mulhi_epi16(psi_a_p3_m7, ONE_OVER_SQRT_2); + psi_a_p3_m7 = simde_mm256_slli_epi16(psi_a_p3_m7, 2); + psi_a_p1_p7 = simde_mm256_mulhi_epi16(psi_a_p1_p7, ONE_OVER_SQRT_2); + psi_a_p1_p7 = simde_mm256_slli_epi16(psi_a_p1_p7, 2); + psi_a_p1_p5 = simde_mm256_mulhi_epi16(psi_a_p1_p5, ONE_OVER_SQRT_2); + psi_a_p1_p5 = simde_mm256_slli_epi16(psi_a_p1_p5, 2); + psi_a_p1_p3 = simde_mm256_mulhi_epi16(psi_a_p1_p3, ONE_OVER_SQRT_2); + psi_a_p1_p3 = simde_mm256_slli_epi16(psi_a_p1_p3, 2); + psi_a_p1_p1 = simde_mm256_mulhi_epi16(psi_a_p1_p1, ONE_OVER_SQRT_2); + psi_a_p1_p1 = simde_mm256_slli_epi16(psi_a_p1_p1, 2); + psi_a_p1_m1 = simde_mm256_mulhi_epi16(psi_a_p1_m1, ONE_OVER_SQRT_2); + psi_a_p1_m1 = simde_mm256_slli_epi16(psi_a_p1_m1, 2); + psi_a_p1_m3 = simde_mm256_mulhi_epi16(psi_a_p1_m3, ONE_OVER_SQRT_2); + psi_a_p1_m3 = simde_mm256_slli_epi16(psi_a_p1_m3, 2); + psi_a_p1_m5 = simde_mm256_mulhi_epi16(psi_a_p1_m5, ONE_OVER_SQRT_2); + psi_a_p1_m5 = simde_mm256_slli_epi16(psi_a_p1_m5, 2); + psi_a_p1_m7 = simde_mm256_mulhi_epi16(psi_a_p1_m7, ONE_OVER_SQRT_2); + psi_a_p1_m7 = simde_mm256_slli_epi16(psi_a_p1_m7, 2); + psi_a_m1_p7 = simde_mm256_mulhi_epi16(psi_a_m1_p7, ONE_OVER_SQRT_2); + psi_a_m1_p7 = simde_mm256_slli_epi16(psi_a_m1_p7, 2); + psi_a_m1_p5 = simde_mm256_mulhi_epi16(psi_a_m1_p5, ONE_OVER_SQRT_2); + psi_a_m1_p5 = simde_mm256_slli_epi16(psi_a_m1_p5, 2); + psi_a_m1_p3 = simde_mm256_mulhi_epi16(psi_a_m1_p3, ONE_OVER_SQRT_2); + psi_a_m1_p3 = simde_mm256_slli_epi16(psi_a_m1_p3, 2); + psi_a_m1_p1 = simde_mm256_mulhi_epi16(psi_a_m1_p1, ONE_OVER_SQRT_2); + psi_a_m1_p1 = simde_mm256_slli_epi16(psi_a_m1_p1, 2); + psi_a_m1_m1 = simde_mm256_mulhi_epi16(psi_a_m1_m1, ONE_OVER_SQRT_2); + psi_a_m1_m1 = simde_mm256_slli_epi16(psi_a_m1_m1, 2); + psi_a_m1_m3 = simde_mm256_mulhi_epi16(psi_a_m1_m3, ONE_OVER_SQRT_2); + psi_a_m1_m3 = simde_mm256_slli_epi16(psi_a_m1_m3, 2); + psi_a_m1_m5 = simde_mm256_mulhi_epi16(psi_a_m1_m5, ONE_OVER_SQRT_2); + psi_a_m1_m5 = simde_mm256_slli_epi16(psi_a_m1_m5, 2); + psi_a_m1_m7 = simde_mm256_mulhi_epi16(psi_a_m1_m7, ONE_OVER_SQRT_2); + psi_a_m1_m7 = simde_mm256_slli_epi16(psi_a_m1_m7, 2); + psi_a_m3_p7 = simde_mm256_mulhi_epi16(psi_a_m3_p7, ONE_OVER_SQRT_2); + psi_a_m3_p7 = simde_mm256_slli_epi16(psi_a_m3_p7, 2); + psi_a_m3_p5 = simde_mm256_mulhi_epi16(psi_a_m3_p5, ONE_OVER_SQRT_2); + psi_a_m3_p5 = simde_mm256_slli_epi16(psi_a_m3_p5, 2); + psi_a_m3_p3 = simde_mm256_mulhi_epi16(psi_a_m3_p3, ONE_OVER_SQRT_2); + psi_a_m3_p3 = simde_mm256_slli_epi16(psi_a_m3_p3, 2); + psi_a_m3_p1 = simde_mm256_mulhi_epi16(psi_a_m3_p1, ONE_OVER_SQRT_2); + psi_a_m3_p1 = simde_mm256_slli_epi16(psi_a_m3_p1, 2); + psi_a_m3_m1 = simde_mm256_mulhi_epi16(psi_a_m3_m1, ONE_OVER_SQRT_2); + psi_a_m3_m1 = simde_mm256_slli_epi16(psi_a_m3_m1, 2); + psi_a_m3_m3 = simde_mm256_mulhi_epi16(psi_a_m3_m3, ONE_OVER_SQRT_2); + psi_a_m3_m3 = simde_mm256_slli_epi16(psi_a_m3_m3, 2); + psi_a_m3_m5 = simde_mm256_mulhi_epi16(psi_a_m3_m5, ONE_OVER_SQRT_2); + psi_a_m3_m5 = simde_mm256_slli_epi16(psi_a_m3_m5, 2); + psi_a_m3_m7 = simde_mm256_mulhi_epi16(psi_a_m3_m7, ONE_OVER_SQRT_2); + psi_a_m3_m7 = simde_mm256_slli_epi16(psi_a_m3_m7, 2); + psi_a_m5_p7 = simde_mm256_mulhi_epi16(psi_a_m5_p7, ONE_OVER_SQRT_2); + psi_a_m5_p7 = simde_mm256_slli_epi16(psi_a_m5_p7, 2); + psi_a_m5_p5 = simde_mm256_mulhi_epi16(psi_a_m5_p5, ONE_OVER_SQRT_2); + psi_a_m5_p5 = simde_mm256_slli_epi16(psi_a_m5_p5, 2); + psi_a_m5_p3 = simde_mm256_mulhi_epi16(psi_a_m5_p3, ONE_OVER_SQRT_2); + psi_a_m5_p3 = simde_mm256_slli_epi16(psi_a_m5_p3, 2); + psi_a_m5_p1 = simde_mm256_mulhi_epi16(psi_a_m5_p1, ONE_OVER_SQRT_2); + psi_a_m5_p1 = simde_mm256_slli_epi16(psi_a_m5_p1, 2); + psi_a_m5_m1 = simde_mm256_mulhi_epi16(psi_a_m5_m1, ONE_OVER_SQRT_2); + psi_a_m5_m1 = simde_mm256_slli_epi16(psi_a_m5_m1, 2); + psi_a_m5_m3 = simde_mm256_mulhi_epi16(psi_a_m5_m3, ONE_OVER_SQRT_2); + psi_a_m5_m3 = simde_mm256_slli_epi16(psi_a_m5_m3, 2); + psi_a_m5_m5 = simde_mm256_mulhi_epi16(psi_a_m5_m5, ONE_OVER_SQRT_2); + psi_a_m5_m5 = simde_mm256_slli_epi16(psi_a_m5_m5, 2); + psi_a_m5_m7 = simde_mm256_mulhi_epi16(psi_a_m5_m7, ONE_OVER_SQRT_2); + psi_a_m5_m7 = simde_mm256_slli_epi16(psi_a_m5_m7, 2); + psi_a_m7_p7 = simde_mm256_mulhi_epi16(psi_a_m7_p7, ONE_OVER_SQRT_2); + psi_a_m7_p7 = simde_mm256_slli_epi16(psi_a_m7_p7, 2); + psi_a_m7_p5 = simde_mm256_mulhi_epi16(psi_a_m7_p5, ONE_OVER_SQRT_2); + psi_a_m7_p5 = simde_mm256_slli_epi16(psi_a_m7_p5, 2); + psi_a_m7_p3 = simde_mm256_mulhi_epi16(psi_a_m7_p3, ONE_OVER_SQRT_2); + psi_a_m7_p3 = simde_mm256_slli_epi16(psi_a_m7_p3, 2); + psi_a_m7_p1 = simde_mm256_mulhi_epi16(psi_a_m7_p1, ONE_OVER_SQRT_2); + psi_a_m7_p1 = simde_mm256_slli_epi16(psi_a_m7_p1, 2); + psi_a_m7_m1 = simde_mm256_mulhi_epi16(psi_a_m7_m1, ONE_OVER_SQRT_2); + psi_a_m7_m1 = simde_mm256_slli_epi16(psi_a_m7_m1, 2); + psi_a_m7_m3 = simde_mm256_mulhi_epi16(psi_a_m7_m3, ONE_OVER_SQRT_2); + psi_a_m7_m3 = simde_mm256_slli_epi16(psi_a_m7_m3, 2); + psi_a_m7_m5 = simde_mm256_mulhi_epi16(psi_a_m7_m5, ONE_OVER_SQRT_2); + psi_a_m7_m5 = simde_mm256_slli_epi16(psi_a_m7_m5, 2); + psi_a_m7_m7 = simde_mm256_mulhi_epi16(psi_a_m7_m7, ONE_OVER_SQRT_2); + psi_a_m7_m7 = simde_mm256_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_256(a_r_p7_p7, a_i_p7_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_p7); + square_a_64qam_epi16_256(a_r_p7_p5, a_i_p7_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_p5); + square_a_64qam_epi16_256(a_r_p7_p3, a_i_p7_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_p3); + square_a_64qam_epi16_256(a_r_p7_p1, a_i_p7_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_p1); + square_a_64qam_epi16_256(a_r_p7_m1, a_i_p7_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_m1); + square_a_64qam_epi16_256(a_r_p7_m3, a_i_p7_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_m3); + square_a_64qam_epi16_256(a_r_p7_m5, a_i_p7_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_m5); + square_a_64qam_epi16_256(a_r_p7_m7, a_i_p7_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p7_m7); + square_a_64qam_epi16_256(a_r_p5_p7, a_i_p5_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_p7); + square_a_64qam_epi16_256(a_r_p5_p5, a_i_p5_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_p5); + square_a_64qam_epi16_256(a_r_p5_p3, a_i_p5_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_p3); + square_a_64qam_epi16_256(a_r_p5_p1, a_i_p5_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_p1); + square_a_64qam_epi16_256(a_r_p5_m1, a_i_p5_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_m1); + square_a_64qam_epi16_256(a_r_p5_m3, a_i_p5_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_m3); + square_a_64qam_epi16_256(a_r_p5_m5, a_i_p5_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_m5); + square_a_64qam_epi16_256(a_r_p5_m7, a_i_p5_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p5_m7); + square_a_64qam_epi16_256(a_r_p3_p7, a_i_p3_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_p7); + square_a_64qam_epi16_256(a_r_p3_p5, a_i_p3_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_p5); + square_a_64qam_epi16_256(a_r_p3_p3, a_i_p3_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_p3); + square_a_64qam_epi16_256(a_r_p3_p1, a_i_p3_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_p1); + square_a_64qam_epi16_256(a_r_p3_m1, a_i_p3_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_m1); + square_a_64qam_epi16_256(a_r_p3_m3, a_i_p3_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_m3); + square_a_64qam_epi16_256(a_r_p3_m5, a_i_p3_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_m5); + square_a_64qam_epi16_256(a_r_p3_m7, a_i_p3_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p3_m7); + square_a_64qam_epi16_256(a_r_p1_p7, a_i_p1_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_p7); + square_a_64qam_epi16_256(a_r_p1_p5, a_i_p1_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_p5); + square_a_64qam_epi16_256(a_r_p1_p3, a_i_p1_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_p3); + square_a_64qam_epi16_256(a_r_p1_p1, a_i_p1_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_p1); + square_a_64qam_epi16_256(a_r_p1_m1, a_i_p1_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_m1); + square_a_64qam_epi16_256(a_r_p1_m3, a_i_p1_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_m3); + square_a_64qam_epi16_256(a_r_p1_m5, a_i_p1_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_m5); + square_a_64qam_epi16_256(a_r_p1_m7, a_i_p1_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_m7); + square_a_64qam_epi16_256(a_r_m1_p7, a_i_m1_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_p7); + square_a_64qam_epi16_256(a_r_m1_p5, a_i_m1_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_p5); + square_a_64qam_epi16_256(a_r_m1_p3, a_i_m1_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_p3); + square_a_64qam_epi16_256(a_r_m1_p1, a_i_m1_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_p1); + square_a_64qam_epi16_256(a_r_m1_m1, a_i_m1_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_m1); + square_a_64qam_epi16_256(a_r_m1_m3, a_i_m1_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_m3); + square_a_64qam_epi16_256(a_r_m1_m5, a_i_m1_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_m5); + square_a_64qam_epi16_256(a_r_m1_m7, a_i_m1_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_m7); + square_a_64qam_epi16_256(a_r_m3_p7, a_i_m3_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_p7); + square_a_64qam_epi16_256(a_r_m3_p5, a_i_m3_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_p5); + square_a_64qam_epi16_256(a_r_m3_p3, a_i_m3_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_p3); + square_a_64qam_epi16_256(a_r_m3_p1, a_i_m3_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_p1); + square_a_64qam_epi16_256(a_r_m3_m1, a_i_m3_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_m1); + square_a_64qam_epi16_256(a_r_m3_m3, a_i_m3_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_m3); + square_a_64qam_epi16_256(a_r_m3_m5, a_i_m3_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_m5); + square_a_64qam_epi16_256(a_r_m3_m7, a_i_m3_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m3_m7); + square_a_64qam_epi16_256(a_r_m5_p7, a_i_m5_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_p7); + square_a_64qam_epi16_256(a_r_m5_p5, a_i_m5_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_p5); + square_a_64qam_epi16_256(a_r_m5_p3, a_i_m5_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_p3); + square_a_64qam_epi16_256(a_r_m5_p1, a_i_m5_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_p1); + square_a_64qam_epi16_256(a_r_m5_m1, a_i_m5_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_m1); + square_a_64qam_epi16_256(a_r_m5_m3, a_i_m5_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_m3); + square_a_64qam_epi16_256(a_r_m5_m5, a_i_m5_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_m5); + square_a_64qam_epi16_256(a_r_m5_m7, a_i_m5_m7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m5_m7); + square_a_64qam_epi16_256(a_r_m7_p7, a_i_m7_p7, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_p7); + square_a_64qam_epi16_256(a_r_m7_p5, a_i_m7_p5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_p5); + square_a_64qam_epi16_256(a_r_m7_p3, a_i_m7_p3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_p3); + square_a_64qam_epi16_256(a_r_m7_p1, a_i_m7_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_p1); + square_a_64qam_epi16_256(a_r_m7_m1, a_i_m7_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_m1); + square_a_64qam_epi16_256(a_r_m7_m3, a_i_m7_m3, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_m3); + square_a_64qam_epi16_256(a_r_m7_m5, a_i_m7_m5, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m7_m5); + square_a_64qam_epi16_256(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 = simde_mm256_mulhi_epi16(ch_mag_des, ONE_OVER_FOUR_SQRT_42); + ch_mag_2_over_42_with_sigma2 = simde_mm256_slli_epi16(ch_mag_2_over_42_with_sigma2, 1); + // x=1, y=3 + ch_mag_10_over_42_with_sigma2 = simde_mm256_mulhi_epi16(ch_mag_des, FIVE_OVER_FOUR_SQRT_42); + ch_mag_10_over_42_with_sigma2 = simde_mm256_slli_epi16(ch_mag_10_over_42_with_sigma2, 1); + // x=1, x=5 + ch_mag_26_over_42_with_sigma2 = simde_mm256_mulhi_epi16(ch_mag_des, THIRTEEN_OVER_FOUR_SQRT_42); + ch_mag_26_over_42_with_sigma2 = simde_mm256_slli_epi16(ch_mag_26_over_42_with_sigma2, 1); + // x=1, y=7 + ch_mag_50_over_42_with_sigma2 = simde_mm256_mulhi_epi16(ch_mag_des, TWENTYFIVE_OVER_FOUR_SQRT_42); + ch_mag_50_over_42_with_sigma2 = simde_mm256_slli_epi16(ch_mag_50_over_42_with_sigma2, 1); + // x=3, y=3 + ch_mag_18_over_42_with_sigma2 = simde_mm256_mulhi_epi16(ch_mag_des, NINE_OVER_FOUR_SQRT_42); + ch_mag_18_over_42_with_sigma2 = simde_mm256_slli_epi16(ch_mag_18_over_42_with_sigma2, 1); + // x=3, y=5 + ch_mag_34_over_42_with_sigma2 = simde_mm256_mulhi_epi16(ch_mag_des, SEVENTEEN_OVER_FOUR_SQRT_42); + ch_mag_34_over_42_with_sigma2 = simde_mm256_slli_epi16(ch_mag_34_over_42_with_sigma2, 1); + // x=3, y=7 + ch_mag_58_over_42_with_sigma2 = simde_mm256_mulhi_epi16(ch_mag_des, TWENTYNINE_OVER_FOUR_SQRT_42); + ch_mag_58_over_42_with_sigma2 = simde_mm256_slli_epi16(ch_mag_58_over_42_with_sigma2, 2); + // x=5, y=5 + ch_mag_50_over_42_with_sigma2 = simde_mm256_mulhi_epi16(ch_mag_des, TWENTYFIVE_OVER_FOUR_SQRT_42); + ch_mag_50_over_42_with_sigma2 = simde_mm256_slli_epi16(ch_mag_50_over_42_with_sigma2, 1); + // x=5, y=7 + ch_mag_74_over_42_with_sigma2 = simde_mm256_mulhi_epi16(ch_mag_des, THIRTYSEVEN_OVER_FOUR_SQRT_42); + ch_mag_74_over_42_with_sigma2 = simde_mm256_slli_epi16(ch_mag_74_over_42_with_sigma2, 2); + // x=7, y=7 + ch_mag_98_over_42_with_sigma2 = simde_mm256_mulhi_epi16(ch_mag_des, FORTYNINE_OVER_FOUR_SQRT_42); + ch_mag_98_over_42_with_sigma2 = simde_mm256_slli_epi16(ch_mag_98_over_42_with_sigma2, 2); + + // Computing Metrics + xmm0 = simde_mm256_subs_epi16(psi_a_p7_p7, a_sq_p7_p7); + simde__m256i xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_7_7); + simde__m256i bit_met_p7_p7 = simde_mm256_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p7_p5, a_sq_p7_p5); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_7_5); + simde__m256i bit_met_p7_p5 = simde_mm256_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p7_p3, a_sq_p7_p3); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_7_3); + simde__m256i bit_met_p7_p3 = simde_mm256_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p7_p1, a_sq_p7_p1); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_7_1); + simde__m256i bit_met_p7_p1 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p7_m1, a_sq_p7_m1); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_7_1); + simde__m256i bit_met_p7_m1 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p7_m3, a_sq_p7_m3); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_7_3); + simde__m256i bit_met_p7_m3 = simde_mm256_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p7_m5, a_sq_p7_m5); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_7_5); + simde__m256i bit_met_p7_m5 = simde_mm256_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p7_m7, a_sq_p7_m7); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_7_7); + simde__m256i bit_met_p7_m7 = simde_mm256_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p5_p7, a_sq_p5_p7); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_5_7); + simde__m256i bit_met_p5_p7 = simde_mm256_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p5_p5, a_sq_p5_p5); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_5_5); + simde__m256i bit_met_p5_p5 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p5_p3, a_sq_p5_p3); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_5_3); + simde__m256i bit_met_p5_p3 = simde_mm256_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p5_p1, a_sq_p5_p1); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_5_1); + simde__m256i bit_met_p5_p1 = simde_mm256_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p5_m1, a_sq_p5_m1); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_5_1); + simde__m256i bit_met_p5_m1 = simde_mm256_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p5_m3, a_sq_p5_m3); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_5_3); + simde__m256i bit_met_p5_m3 = simde_mm256_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p5_m5, a_sq_p5_m5); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_5_5); + simde__m256i bit_met_p5_m5 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p5_m7, a_sq_p5_m7); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_5_7); + simde__m256i bit_met_p5_m7 = simde_mm256_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p3_p7, a_sq_p3_p7); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_3_7); + simde__m256i bit_met_p3_p7 = simde_mm256_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p3_p5, a_sq_p3_p5); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_3_5); + simde__m256i bit_met_p3_p5 = simde_mm256_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p3_p3, a_sq_p3_p3); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_3_3); + simde__m256i bit_met_p3_p3 = simde_mm256_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p3_p1, a_sq_p3_p1); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_3_1); + simde__m256i bit_met_p3_p1 = simde_mm256_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p3_m1, a_sq_p3_m1); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_3_1); + simde__m256i bit_met_p3_m1 = simde_mm256_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p3_m3, a_sq_p3_m3); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_3_3); + simde__m256i bit_met_p3_m3 = simde_mm256_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p3_m5, a_sq_p3_m5); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_3_5); + simde__m256i bit_met_p3_m5 = simde_mm256_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p3_m7, a_sq_p3_m7); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_3_7); + simde__m256i bit_met_p3_m7 = simde_mm256_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p1_p7, a_sq_p1_p7); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_1_7); + simde__m256i bit_met_p1_p7 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p1_p5, a_sq_p1_p5); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_1_5); + simde__m256i bit_met_p1_p5 = simde_mm256_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p1_p3, a_sq_p1_p3); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_1_3); + simde__m256i bit_met_p1_p3 = simde_mm256_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p1_p1, a_sq_p1_p1); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_p_1_1); + simde__m256i bit_met_p1_p1 = simde_mm256_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p1_m1, a_sq_p1_m1); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_1_1); + simde__m256i bit_met_p1_m1 = simde_mm256_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p1_m3, a_sq_p1_m3); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_1_3); + simde__m256i bit_met_p1_m3 = simde_mm256_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p1_m5, a_sq_p1_m5); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_1_5); + simde__m256i bit_met_p1_m5 = simde_mm256_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_p1_m7, a_sq_p1_m7); + xmm1 = simde_mm256_adds_epi16(xmm0, y0_m_1_7); + simde__m256i bit_met_p1_m7 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + + xmm0 = simde_mm256_subs_epi16(psi_a_m1_p7, a_sq_m1_p7); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_1_7); + simde__m256i bit_met_m1_p7 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m1_p5, a_sq_m1_p5); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_1_5); + simde__m256i bit_met_m1_p5 = simde_mm256_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m1_p3, a_sq_m1_p3); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_1_3); + simde__m256i bit_met_m1_p3 = simde_mm256_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m1_p1, a_sq_m1_p1); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_1_1); + simde__m256i bit_met_m1_p1 = simde_mm256_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m1_m1, a_sq_m1_m1); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_1_1); + simde__m256i bit_met_m1_m1 = simde_mm256_subs_epi16(xmm1, ch_mag_2_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m1_m3, a_sq_m1_m3); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_1_3); + simde__m256i bit_met_m1_m3 = simde_mm256_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m1_m5, a_sq_m1_m5); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_1_5); + simde__m256i bit_met_m1_m5 = simde_mm256_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m1_m7, a_sq_m1_m7); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_1_7); + simde__m256i bit_met_m1_m7 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m3_p7, a_sq_m3_p7); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_3_7); + simde__m256i bit_met_m3_p7 = simde_mm256_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m3_p5, a_sq_m3_p5); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_3_5); + simde__m256i bit_met_m3_p5 = simde_mm256_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m3_p3, a_sq_m3_p3); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_3_3); + simde__m256i bit_met_m3_p3 = simde_mm256_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m3_p1, a_sq_m3_p1); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_3_1); + simde__m256i bit_met_m3_p1 = simde_mm256_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m3_m1, a_sq_m3_m1); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_3_1); + simde__m256i bit_met_m3_m1 = simde_mm256_subs_epi16(xmm1, ch_mag_10_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m3_m3, a_sq_m3_m3); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_3_3); + simde__m256i bit_met_m3_m3 = simde_mm256_subs_epi16(xmm1, ch_mag_18_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m3_m5, a_sq_m3_m5); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_3_5); + simde__m256i bit_met_m3_m5 = simde_mm256_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m3_m7, a_sq_m3_m7); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_3_7); + simde__m256i bit_met_m3_m7 = simde_mm256_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m5_p7, a_sq_m5_p7); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_5_7); + simde__m256i bit_met_m5_p7 = simde_mm256_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m5_p5, a_sq_m5_p5); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_5_5); + simde__m256i bit_met_m5_p5 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m5_p3, a_sq_m5_p3); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_5_3); + simde__m256i bit_met_m5_p3 = simde_mm256_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m5_p1, a_sq_m5_p1); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_5_1); + simde__m256i bit_met_m5_p1 = simde_mm256_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m5_m1, a_sq_m5_m1); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_5_1); + simde__m256i bit_met_m5_m1 = simde_mm256_subs_epi16(xmm1, ch_mag_26_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m5_m3, a_sq_m5_m3); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_5_3); + simde__m256i bit_met_m5_m3 = simde_mm256_subs_epi16(xmm1, ch_mag_34_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m5_m5, a_sq_m5_m5); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_5_5); + simde__m256i bit_met_m5_m5 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m5_m7, a_sq_m5_m7); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_5_7); + simde__m256i bit_met_m5_m7 = simde_mm256_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m7_p7, a_sq_m7_p7); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_7_7); + simde__m256i bit_met_m7_p7 = simde_mm256_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m7_p5, a_sq_m7_p5); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_7_5); + simde__m256i bit_met_m7_p5 = simde_mm256_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m7_p3, a_sq_m7_p3); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_7_3); + simde__m256i bit_met_m7_p3 = simde_mm256_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m7_p1, a_sq_m7_p1); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_m_7_1); + simde__m256i bit_met_m7_p1 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m7_m1, a_sq_m7_m1); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_7_1); + simde__m256i bit_met_m7_m1 = simde_mm256_subs_epi16(xmm1, ch_mag_50_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m7_m3, a_sq_m7_m3); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_7_3); + simde__m256i bit_met_m7_m3 = simde_mm256_subs_epi16(xmm1, ch_mag_58_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m7_m5, a_sq_m7_m5); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_7_5); + simde__m256i bit_met_m7_m5 = simde_mm256_subs_epi16(xmm1, ch_mag_74_over_42_with_sigma2); + xmm0 = simde_mm256_subs_epi16(psi_a_m7_m7, a_sq_m7_m7); + xmm1 = simde_mm256_subs_epi16(xmm0, y0_p_7_7); + simde__m256i bit_met_m7_m7 = simde_mm256_subs_epi16(xmm1, ch_mag_98_over_42_with_sigma2); + + // Detection for 1st bit + // bit = 1 + xmm0 = simde_mm256_max_epi16(bit_met_m7_p7, bit_met_m7_p5); + xmm1 = simde_mm256_max_epi16(bit_met_m7_p3, bit_met_m7_p1); + xmm2 = simde_mm256_max_epi16(bit_met_m7_m1, bit_met_m7_m3); + xmm3 = simde_mm256_max_epi16(bit_met_m7_m5, bit_met_m7_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + simde__m256i logmax_den_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_m5_p7, bit_met_m5_p5); + xmm1 = simde_mm256_max_epi16(bit_met_m5_p3, bit_met_m5_p1); + xmm2 = simde_mm256_max_epi16(bit_met_m5_m1, bit_met_m5_m3); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m5, bit_met_m5_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_m3_p7, bit_met_m3_p5); + xmm1 = simde_mm256_max_epi16(bit_met_m3_p3, bit_met_m3_p1); + xmm2 = simde_mm256_max_epi16(bit_met_m3_m1, bit_met_m3_m3); + xmm3 = simde_mm256_max_epi16(bit_met_m3_m5, bit_met_m3_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_m1_p7, bit_met_m1_p5); + xmm1 = simde_mm256_max_epi16(bit_met_m1_p3, bit_met_m1_p1); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m1, bit_met_m1_m3); + xmm3 = simde_mm256_max_epi16(bit_met_m1_m5, bit_met_m1_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + + // bit = 0 + xmm0 = simde_mm256_max_epi16(bit_met_p7_p7, bit_met_p7_p5); + xmm1 = simde_mm256_max_epi16(bit_met_p7_p3, bit_met_p7_p1); + xmm2 = simde_mm256_max_epi16(bit_met_p7_m1, bit_met_p7_m3); + xmm3 = simde_mm256_max_epi16(bit_met_p7_m5, bit_met_p7_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + simde__m256i logmax_num_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p5_p7, bit_met_p5_p5); + xmm1 = simde_mm256_max_epi16(bit_met_p5_p3, bit_met_p5_p1); + xmm2 = simde_mm256_max_epi16(bit_met_p5_m1, bit_met_p5_m3); + xmm3 = simde_mm256_max_epi16(bit_met_p5_m5, bit_met_p5_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p3_p7, bit_met_p3_p5); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p3, bit_met_p3_p1); + xmm2 = simde_mm256_max_epi16(bit_met_p3_m1, bit_met_p3_m3); + xmm3 = simde_mm256_max_epi16(bit_met_p3_m5, bit_met_p3_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p1_p7, bit_met_p1_p5); + xmm1 = simde_mm256_max_epi16(bit_met_p1_p3, bit_met_p1_p1); + xmm2 = simde_mm256_max_epi16(bit_met_p1_m1, bit_met_p1_m3); + xmm3 = simde_mm256_max_epi16(bit_met_p1_m5, bit_met_p1_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + + y0r = simde_mm256_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 2nd bit + // bit = 1 + xmm0 = simde_mm256_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + + // bit = 0 + xmm0 = simde_mm256_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p1, bit_met_m7_p1); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p5, bit_met_m7_p5); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + + y1r = simde_mm256_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 3rd bit + xmm0 = simde_mm256_max_epi16(bit_met_m7_m7, bit_met_m7_m5); + xmm1 = simde_mm256_max_epi16(bit_met_m7_m3, bit_met_m7_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m7_p1, bit_met_m7_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m7_p5, bit_met_m7_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_m5_m7, bit_met_m5_m5); + xmm1 = simde_mm256_max_epi16(bit_met_m5_m3, bit_met_m5_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m5_p1, bit_met_m5_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p5, bit_met_m5_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p5_m7, bit_met_p5_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p5_m3, bit_met_p5_m1); + xmm2 = simde_mm256_max_epi16(bit_met_p5_p1, bit_met_p5_p3); + xmm3 = simde_mm256_max_epi16(bit_met_p5_p5, bit_met_p5_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m7, bit_met_p7_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p7_m3, bit_met_p7_m1); + xmm2 = simde_mm256_max_epi16(bit_met_p7_p1, bit_met_p7_p3); + xmm3 = simde_mm256_max_epi16(bit_met_p7_p5, bit_met_p7_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + + xmm0 = simde_mm256_max_epi16(bit_met_m3_m7, bit_met_m3_m5); + xmm1 = simde_mm256_max_epi16(bit_met_m3_m3, bit_met_m3_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m3_p1, bit_met_m3_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m3_p5, bit_met_m3_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_m1_m7, bit_met_m1_m5); + xmm1 = simde_mm256_max_epi16(bit_met_m1_m3, bit_met_m1_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p1, bit_met_m1_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m1_p5, bit_met_m1_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p1_m7, bit_met_p1_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p1_m3, bit_met_p1_m1); + xmm2 = simde_mm256_max_epi16(bit_met_p1_p1, bit_met_p1_p3); + xmm3 = simde_mm256_max_epi16(bit_met_p1_p5, bit_met_p1_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p3_m7, bit_met_p3_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m3, bit_met_p3_m1); + xmm2 = simde_mm256_max_epi16(bit_met_p3_p1, bit_met_p3_p3); + xmm3 = simde_mm256_max_epi16(bit_met_p3_p5, bit_met_p3_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + + simde__m256i y2r = simde_mm256_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 4th bit + xmm0 = simde_mm256_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p5, bit_met_m5_p5); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + + xmm0 = simde_mm256_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p1, bit_met_m7_p1); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + + y0i = simde_mm256_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 5th bit + xmm0 = simde_mm256_max_epi16(bit_met_m7_m7, bit_met_m7_m5); + xmm1 = simde_mm256_max_epi16(bit_met_m7_m3, bit_met_m7_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m7_p1, bit_met_m7_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m7_p5, bit_met_m7_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_m1_m7, bit_met_m1_m5); + xmm1 = simde_mm256_max_epi16(bit_met_m1_m3, bit_met_m1_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p1, bit_met_m1_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m1_p5, bit_met_m1_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p1_m7, bit_met_p1_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p1_m3, bit_met_p1_m1); + xmm2 = simde_mm256_max_epi16(bit_met_p1_p1, bit_met_p1_p3); + xmm3 = simde_mm256_max_epi16(bit_met_p1_p5, bit_met_p1_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m7, bit_met_p7_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p7_m3, bit_met_p7_m1); + xmm2 = simde_mm256_max_epi16(bit_met_p7_p1, bit_met_p7_p3); + xmm3 = simde_mm256_max_epi16(bit_met_p7_p5, bit_met_p7_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + + xmm0 = simde_mm256_max_epi16(bit_met_m5_m7, bit_met_m5_m5); + xmm1 = simde_mm256_max_epi16(bit_met_m5_m3, bit_met_m5_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m5_p1, bit_met_m5_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p5, bit_met_m5_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_m3_m7, bit_met_m3_m5); + xmm1 = simde_mm256_max_epi16(bit_met_m3_m3, bit_met_m3_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m3_p1, bit_met_m3_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m3_p5, bit_met_m3_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p3_m7, bit_met_p3_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m3, bit_met_p3_m1); + xmm2 = simde_mm256_max_epi16(bit_met_p3_p1, bit_met_p3_p3); + xmm3 = simde_mm256_max_epi16(bit_met_p3_p5, bit_met_p3_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p5_m7, bit_met_p5_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p5_m3, bit_met_p5_m1); + xmm2 = simde_mm256_max_epi16(bit_met_p5_p1, bit_met_p5_p3); + xmm3 = simde_mm256_max_epi16(bit_met_p5_p5, bit_met_p5_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + + y1i = simde_mm256_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Detection for 6th bit + xmm0 = simde_mm256_max_epi16(bit_met_p7_p7, bit_met_p5_p7); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p7, bit_met_p1_p7); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p7, bit_met_m3_p7); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p7, bit_met_m7_p7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_p1, bit_met_p5_p1); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p1, bit_met_p1_p1); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p1, bit_met_m3_p1); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p1, bit_met_m5_p1); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m1, bit_met_p5_m1); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m1, bit_met_p1_m1); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m1, bit_met_m3_m1); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m1, bit_met_m7_m1); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m7, bit_met_p5_m7); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m7, bit_met_p1_m7); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m7, bit_met_m3_m7); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m7, bit_met_m7_m7); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm4); + logmax_den_re0 = simde_mm256_max_epi16(logmax_den_re0, xmm5); + + xmm0 = simde_mm256_max_epi16(bit_met_p7_m5, bit_met_p5_m5); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m5, bit_met_p1_m5); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m5, bit_met_m3_m5); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m5, bit_met_m7_m5); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(xmm4, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_m3, bit_met_p5_m3); + xmm1 = simde_mm256_max_epi16(bit_met_p3_m3, bit_met_p1_m3); + xmm2 = simde_mm256_max_epi16(bit_met_m1_m3, bit_met_m3_m3); + xmm3 = simde_mm256_max_epi16(bit_met_m5_m3, bit_met_m7_m3); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_p3, bit_met_p5_p3); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p3, bit_met_p1_p3); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p3, bit_met_m3_p3); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p3, bit_met_m7_p3); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + xmm0 = simde_mm256_max_epi16(bit_met_p7_p5, bit_met_p5_p5); + xmm1 = simde_mm256_max_epi16(bit_met_p3_p5, bit_met_p1_p5); + xmm2 = simde_mm256_max_epi16(bit_met_m1_p5, bit_met_m3_p5); + xmm3 = simde_mm256_max_epi16(bit_met_m5_p5, bit_met_m7_p5); + xmm4 = simde_mm256_max_epi16(xmm0, xmm1); + xmm5 = simde_mm256_max_epi16(xmm2, xmm3); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm4); + logmax_num_re0 = simde_mm256_max_epi16(logmax_num_re0, xmm5); + + simde__m256i y2i = simde_mm256_subs_epi16(logmax_num_re0, logmax_den_re0); + + // Map to output stream, difficult to do in SIMD since we have 6 16bit LLRs + int idx0 = 24 * i; + for (int re = 0; re < 16; re++) { + stream0_out[idx0 + 0].r = ((short *)&y0r)[re]; + stream0_out[idx0 + 0].i = ((short *)&y1r)[re]; + stream0_out[idx0 + 1].r = ((short *)&y2r)[re]; + stream0_out[idx0 + 1].i = ((short *)&y0i)[re]; + stream0_out[idx0 + 2].r = ((short *)&y1i)[re]; + stream0_out[idx0 + 2].i = ((short *)&y2i)[re]; + idx0 += 3; + } + } + + _mm_empty(); + _m_empty(); +} + +void nr_ulsch_compute_ML_llr(int32_t **rxdataF_comp, + int32_t **ul_ch_mag, + int32_t ***rho, + int16_t **llr_layers, + uint8_t nb_antennas_rx, + uint32_t rb_size, + uint32_t nb_re, + uint8_t symbol, + uint32_t rxdataF_ext_offset, + uint8_t mod_order) +{ + int off = ((rb_size & 1) == 1) ? 4 : 0; + c16_t *rxdataF_comp0 = (c16_t *)&rxdataF_comp[0][symbol * (off + (rb_size * NR_NB_SC_PER_RB))]; + c16_t *rxdataF_comp1 = (c16_t *)&rxdataF_comp[nb_antennas_rx][symbol * (off + (rb_size * NR_NB_SC_PER_RB))]; + c16_t *ul_ch_mag0 = (c16_t *)&ul_ch_mag[0][symbol * (off + (rb_size * NR_NB_SC_PER_RB))]; + c16_t *ul_ch_mag1 = (c16_t *)&ul_ch_mag[nb_antennas_rx][symbol * (off + (rb_size * NR_NB_SC_PER_RB))]; + c16_t *llr_layers0 = (c16_t *)&llr_layers[0][rxdataF_ext_offset * mod_order]; + c16_t *llr_layers1 = (c16_t *)&llr_layers[1][rxdataF_ext_offset * mod_order]; + c16_t *rho0 = (c16_t *)&rho[0][1][symbol * (off + (rb_size * NR_NB_SC_PER_RB))]; + c16_t *rho1 = (c16_t *)&rho[0][2][symbol * (off + (rb_size * NR_NB_SC_PER_RB))]; + + switch (mod_order) { + case 2: + nr_ulsch_qpsk_qpsk(rxdataF_comp0, rxdataF_comp1, llr_layers0, rho0, nb_re); + nr_ulsch_qpsk_qpsk(rxdataF_comp1, rxdataF_comp0, llr_layers1, rho1, nb_re); + break; + case 4: + nr_ulsch_qam16_qam16(rxdataF_comp0, rxdataF_comp1, ul_ch_mag0, ul_ch_mag1, llr_layers0, rho0, nb_re); + nr_ulsch_qam16_qam16(rxdataF_comp1, rxdataF_comp0, ul_ch_mag1, ul_ch_mag0, llr_layers1, rho1, nb_re); + break; + case 6: + nr_ulsch_qam64_qam64(rxdataF_comp0, rxdataF_comp1, ul_ch_mag0, ul_ch_mag1, llr_layers0, rho0, nb_re); + nr_ulsch_qam64_qam64(rxdataF_comp1, rxdataF_comp0, ul_ch_mag1, ul_ch_mag0, llr_layers1, rho1, nb_re); + break; + default: + AssertFatal(1 == 0, "nr_ulsch_compute_llr: invalid Qm value, symbol = %d, Qm = %d\n", symbol, mod_order); + } +} + +void nr_ulsch_shift_llr(int16_t **llr_layers, uint32_t nb_re, uint32_t rxdataF_ext_offset, uint8_t mod_order, int shift) +{ + __m128i *llr_layers0 = (__m128i *)&llr_layers[0][rxdataF_ext_offset * mod_order]; + __m128i *llr_layers1 = (__m128i *)&llr_layers[1][rxdataF_ext_offset * mod_order]; + + uint8_t mem_offset = ((16 - ((long)llr_layers0)) & 0xF) >> 2; + + if (mem_offset > 0) { + c16_t *llr_layers0_c16 = (c16_t *)&llr_layers[0][rxdataF_ext_offset * mod_order]; + c16_t *llr_layers1_c16 = (c16_t *)&llr_layers[1][rxdataF_ext_offset * mod_order]; + for (int i = 0; i < mem_offset; i++) { + llr_layers0_c16[i] = c16Shift(llr_layers0_c16[i], shift); + llr_layers1_c16[i] = c16Shift(llr_layers1_c16[i], shift); + } + llr_layers0 = (__m128i *)&llr_layers[0][rxdataF_ext_offset * mod_order + (mem_offset << 1)]; + llr_layers1 = (__m128i *)&llr_layers[1][rxdataF_ext_offset * mod_order + (mem_offset << 1)]; + } + + for (int i = 0; i < nb_re >> 2; i++) { + llr_layers0[i] = simde_mm_srai_epi16(llr_layers0[i], shift); + llr_layers1[i] = simde_mm_srai_epi16(llr_layers1[i], shift); + } +} \ No newline at end of file diff --git a/openair1/PHY/NR_UE_TRANSPORT/pss_nr.c b/openair1/PHY/NR_UE_TRANSPORT/pss_nr.c index 48486bbdee..7323d27dbb 100644 --- a/openair1/PHY/NR_UE_TRANSPORT/pss_nr.c +++ b/openair1/PHY/NR_UE_TRANSPORT/pss_nr.c @@ -73,7 +73,7 @@ int16_t *get_primary_synchro_nr2(const int nid2) * *********************************************************************/ -void generate_pss_nr(NR_DL_FRAME_PARMS *fp,int N_ID_2) +void generate_pss_nr(NR_DL_FRAME_PARMS *fp, int N_ID_2) { AssertFatal(fp->ofdm_symbol_size > 127,"Illegal ofdm_symbol_size %d\n",fp->ofdm_symbol_size); AssertFatal(N_ID_2>=0 && N_ID_2 <=2,"Illegal N_ID_2 %d\n",N_ID_2); @@ -88,9 +88,8 @@ void generate_pss_nr(NR_DL_FRAME_PARMS *fp,int N_ID_2) assert(N_ID_2 < NUMBER_PSS_SEQUENCE); memcpy(x, x_initial, sizeof(x_initial)); - for (int i=0; i < (LENGTH_PSS_NR - INITIAL_PSS_NR); i++) { - x[i+INITIAL_PSS_NR] = (x[i + 4] + x[i])%(2); - } + for (int i = 0; i < (LENGTH_PSS_NR - INITIAL_PSS_NR); i++) + x[i + INITIAL_PSS_NR] = (x[i + 4] + x[i]) % (2); for (int n=0; n < LENGTH_PSS_NR; n++) { const int m = (n + 43 * N_ID_2) % (LENGTH_PSS_NR); @@ -139,7 +138,8 @@ void generate_pss_nr(NR_DL_FRAME_PARMS *fp,int N_ID_2) * sample 0 is for continuous frequency which is used here */ - unsigned int k = fp->first_carrier_offset + fp->ssb_start_subcarrier + 56; //and + unsigned int subcarrier_start = get_softmodem_params()->sl_mode == 0 ? PSS_SSS_SUB_CARRIER_START : PSS_SSS_SUB_CARRIER_START_SL; + unsigned int k = fp->first_carrier_offset + fp->ssb_start_subcarrier + subcarrier_start; if (k>= fp->ofdm_symbol_size) k-=fp->ofdm_symbol_size; c16_t synchroF_tmp[fp->ofdm_symbol_size] __attribute__((aligned(32))); memset(synchroF_tmp, 0, sizeof(synchroF_tmp)); @@ -237,7 +237,8 @@ void generate_pss_nr(NR_DL_FRAME_PARMS *fp,int N_ID_2) static void init_context_pss_nr(NR_DL_FRAME_PARMS *frame_parms_ue) { AssertFatal(frame_parms_ue->ofdm_symbol_size > 127, "illegal ofdm_symbol_size %d\n", frame_parms_ue->ofdm_symbol_size); - for (int i = 0; i < NUMBER_PSS_SEQUENCE; i++) { + int pss_sequence = get_softmodem_params()->sl_mode == 0 ? NUMBER_PSS_SEQUENCE : NUMBER_PSS_SEQUENCE_SL; + for (int i = 0; i < pss_sequence; i++) { primary_synchro_nr2[i] = malloc16_clear(LENGTH_PSS_NR * sizeof(int16_t)); AssertFatal(primary_synchro_nr2[i], "Fatal memory allocation problem \n"); primary_synchro_time_nr[i] = malloc16_clear(frame_parms_ue->ofdm_symbol_size * sizeof(c16_t)); @@ -583,7 +584,8 @@ static int pss_search_time_nr(c16_t **rxdata, PHY_VARS_NR_UE *ue, int fo_flag, i pss_source = 0; int maxval=0; - for (int j = 0; j < NUMBER_PSS_SEQUENCE; j++) + int max_size = get_softmodem_params()->sl_mode == 0 ? NUMBER_PSS_SEQUENCE : NUMBER_PSS_SEQUENCE_SL; + for (int j = 0; j < max_size; j++) for (int i = 0; i < frame_parms->ofdm_symbol_size; i++) { maxval = max(maxval, abs(primary_synchro_time_nr[j][i].r)); maxval = max(maxval, abs(primary_synchro_time_nr[j][i].i)); @@ -595,7 +597,7 @@ static int pss_search_time_nr(c16_t **rxdata, PHY_VARS_NR_UE *ue, int fo_flag, i /* Correlation computation is based on a a dot product which is realized thank to SIMS extensions */ uint16_t pss_index_start = 0; - uint16_t pss_index_end = NUMBER_PSS_SEQUENCE; + uint16_t pss_index_end = get_softmodem_params()->sl_mode == 0 ? NUMBER_PSS_SEQUENCE : NUMBER_PSS_SEQUENCE_SL; if (ue->target_Nid_cell != -1) { pss_index_start = GET_NID2(ue->target_Nid_cell); pss_index_end = pss_index_start + 1; diff --git a/openair1/PHY/NR_UE_TRANSPORT/sss_nr.c b/openair1/PHY/NR_UE_TRANSPORT/sss_nr.c index 2f53a7b95d..cb8e4f40e9 100644 --- a/openair1/PHY/NR_UE_TRANSPORT/sss_nr.c +++ b/openair1/PHY/NR_UE_TRANSPORT/sss_nr.c @@ -36,7 +36,7 @@ #include "PHY/defs_nr_UE.h" #include "PHY/MODULATION/modulation_UE.h" - +#include "executables/softmodem-common.h" #include "PHY/NR_REFSIG/ss_pbch_nr.h" #define DEFINE_VARIABLES_SSS_NR_H @@ -78,31 +78,25 @@ void init_context_sss_nr(int amp) int16_t x1[LENGTH_SSS_NR]; int16_t dss_current; int m0, m1; + int nid_2_num = get_softmodem_params()->sl_mode == 0 ? N_ID_2_NUMBER : N_ID_2_NUMBER_SL; const int x0_initial[INITIAL_SSS_NR] = { 1, 0, 0, 0, 0, 0, 0 }; const int x1_initial[INITIAL_SSS_NR] = { 1, 0, 0, 0, 0, 0, 0 }; - - for (int i=0; i < INITIAL_SSS_NR; i++) { + for (int i = 0; i < INITIAL_SSS_NR; i++) { x0[i] = x0_initial[i]; x1[i] = x1_initial[i]; } - - for (int i=0; i < (LENGTH_SSS_NR - INITIAL_SSS_NR); i++) { - x0[i+7] = (x0[i + 4] + x0[i])%(2); - x1[i+7] = (x1[i + 1] + x1[i])%(2); + for (int i = 0; i < (LENGTH_SSS_NR - INITIAL_SSS_NR); i++) { + x0[i + 7] = (x0[i + 4] + x0[i]) % (2); + x1[i + 7] = (x1[i + 1] + x1[i]) % (2); } - for (int N_ID_2 = 0; N_ID_2 < N_ID_2_NUMBER; N_ID_2++) { - + for (int N_ID_2 = 0; N_ID_2 < nid_2_num; N_ID_2++) { for (int N_ID_1 = 0; N_ID_1 < N_ID_1_NUMBER; N_ID_1++) { - - m0 = 15*(N_ID_1/112) + (5*N_ID_2); - m1 = N_ID_1%112; - + m0 = 15 * (N_ID_1 / 112) + (5 * N_ID_2); + m1 = N_ID_1 % 112; for (int n = 0; n < LENGTH_SSS_NR; n++) { - - dss_current = (1 - 2*x0[(n + m0)%(LENGTH_SSS_NR)])*(1 - 2*x1[(n + m1)%(LENGTH_SSS_NR)]); - + dss_current = (1 - 2 * x0 [(n + m0) % (LENGTH_SSS_NR)]) * (1 - 2 * x1[(n + m1) % (LENGTH_SSS_NR)]); /* Modulation of SSS is a BPSK TS 36.211 chapter 5.1.2 BPSK */ #if 1 d_sss[N_ID_2][N_ID_1][n] = dss_current;// * amp; @@ -276,8 +270,9 @@ static int do_pss_sss_extract_nr( for (int aarx = 0; aarx < frame_parms->nb_antennas_rx; aarx++) { int pss_symbol = 0; - int sss_symbol = SSS_SYMBOL_NB - PSS_SYMBOL_NB; - + int sss_symbol = get_softmodem_params()->sl_mode == 0 ? + (SSS_SYMBOL_NB - PSS_SYMBOL_NB) : + (SSS0_SL_SYMBOL_NB - PSS0_SL_SYMBOL_NB) ; unsigned int ofdm_symbol_size = frame_parms->ofdm_symbol_size; c16_t *pss_rxF = rxdataF[aarx] + pss_symbol * ofdm_symbol_size; @@ -286,12 +281,16 @@ static int do_pss_sss_extract_nr( c16_t *pss_rxF_ext = pss_ext[aarx]; c16_t *sss_rxF_ext = sss_ext[aarx]; - unsigned int k = frame_parms->first_carrier_offset + frame_parms->ssb_start_subcarrier + 56; + unsigned int k = frame_parms->first_carrier_offset + + frame_parms->ssb_start_subcarrier + + ((get_softmodem_params()->sl_mode == 0) ? + PSS_SSS_SUB_CARRIER_START : + PSS_SSS_SUB_CARRIER_START_SL); if (k>= frame_parms->ofdm_symbol_size) k-=frame_parms->ofdm_symbol_size; for (int i=0; i < LENGTH_PSS_NR; i++) { - if (doPss) { + if (doPss) { pss_rxF_ext[i] = pss_rxF[k]; } @@ -300,9 +299,7 @@ static int do_pss_sss_extract_nr( } k++; - - if (k == ofdm_symbol_size) k=0; - + if (k == frame_parms->ofdm_symbol_size) k = 0; } } @@ -365,7 +362,12 @@ static int pss_sss_extract_nr(PHY_VARS_NR_UE *phy_vars_ue, * *********************************************************************/ -int rx_sss_nr(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, int32_t *tot_metric, uint8_t *phase_max, int *freq_offset_sss, c16_t rxdataF[][ue->frame_parms.samples_per_slot_wCP]) +int rx_sss_nr(PHY_VARS_NR_UE *ue, + UE_nr_rxtx_proc_t *proc, + int32_t *tot_metric, + uint8_t *phase_max, + int *freq_offset_sss, + c16_t rxdataF[][ue->frame_parms.samples_per_slot_wCP]) { uint8_t i; c16_t pss_ext[NB_ANTENNAS_RX][LENGTH_PSS_NR]; diff --git a/openair1/PHY/TOOLS/simde_operations.c b/openair1/PHY/TOOLS/simde_operations.c new file mode 100644 index 0000000000..a61956d87b --- /dev/null +++ b/openair1/PHY/TOOLS/simde_operations.c @@ -0,0 +1,60 @@ +/* + * 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 + +void simde_mm128_separate_real_imag_parts(__m128i *out_re, __m128i *out_im, __m128i in0, __m128i in1) +{ + // Put in0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)] + in0 = simde_mm_shufflelo_epi16(in0, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + in0 = simde_mm_shufflehi_epi16(in0, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + in0 = simde_mm_shuffle_epi32(in0, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + + // Put xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)] + in1 = simde_mm_shufflelo_epi16(in1, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + in1 = simde_mm_shufflehi_epi16(in1, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + in1 = simde_mm_shuffle_epi32(in1, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + + *out_re = simde_mm_unpacklo_epi64(in0, in1); + *out_im = simde_mm_unpackhi_epi64(in0, in1); +} + +void simde_mm256_separate_real_imag_parts(__m256i *out_re, __m256i *out_im, __m256i in0, __m256i in1) +{ + // Put in0 = [Re(0,1,2,3) Im(0,1,2,3) Re(4,5,6,7) Im(4,5,6,7)] + in0 = simde_mm256_shufflelo_epi16(in0, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + in0 = simde_mm256_shufflehi_epi16(in0, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + in0 = simde_mm256_shuffle_epi32(in0, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + + // Put in1 = [Re(8,9,10,11) Im(8,9,10,11) Re(12,13,14,15) Im(12,13,14,15)] + in1 = simde_mm256_shufflelo_epi16(in1, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + in1 = simde_mm256_shufflehi_epi16(in1, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + in1 = simde_mm256_shuffle_epi32(in1, 0xd8); //_MM_SHUFFLE(0,2,1,3)); + + // Put tmp0 =[Re(0,1,2,3) Re(8,9,10,11) Re(4,5,6,7) Re(12,13,14,15)] + __m256i tmp0 = simde_mm256_unpacklo_epi64(in0, in1); + + // Put tmp1 = [Im(0,1,2,3) Im(8,9,10,11) Im(4,5,6,7) Im(12,13,14,15)] + __m256i tmp1 = simde_mm256_unpackhi_epi64(in0, in1); + + *out_re = simde_mm256_permute4x64_epi64(tmp0, 0xd8); + *out_im = simde_mm256_permute4x64_epi64(tmp1, 0xd8); +} \ No newline at end of file diff --git a/openair1/PHY/TOOLS/tools_defs.h b/openair1/PHY/TOOLS/tools_defs.h index 4076352455..060b85f94e 100644 --- a/openair1/PHY/TOOLS/tools_defs.h +++ b/openair1/PHY/TOOLS/tools_defs.h @@ -851,6 +851,8 @@ c32_t dot_product(const c16_t *x, double interp(double x, double *xs, double *ys, int count); +void simde_mm128_separate_real_imag_parts(__m128i *out_re, __m128i *out_im, __m128i in0, __m128i in1); +void simde_mm256_separate_real_imag_parts(__m256i *out_re, __m256i *out_im, __m256i in0, __m256i in1); #ifdef __cplusplus } diff --git a/openair1/PHY/defs_nr_UE.h b/openair1/PHY/defs_nr_UE.h index 870f093c74..d4c5a2da49 100644 --- a/openair1/PHY/defs_nr_UE.h +++ b/openair1/PHY/defs_nr_UE.h @@ -318,6 +318,22 @@ typedef struct { fapi_nr_dl_config_dci_dl_pdu_rel15_t pdcch_config[FAPI_NR_MAX_SS]; } NR_UE_PDCCH_CONFIG; +#define NR_PSBCH_MAX_NB_CARRIERS 132 +#define NR_PSBCH_MAX_NB_MOD_SYMBOLS 99 +#define NR_PSBCH_DMRS_LENGTH 297 // in mod symbols +#define NR_PSBCH_DMRS_LENGTH_DWORD 20 // ceil(2(QPSK)*NR_PBCH_DMRS_LENGTH/32) + +/* NR Sidelink PSBCH payload fields + TODO: This will be removed in the future and + filled in by the upper layers once developed. */ +typedef struct { + uint32_t coverageIndicator : 1; + uint32_t tddConfig : 12; + uint32_t DFN : 10; + uint32_t slotIndex : 7; + uint32_t reserved : 2; +} PSBCH_payload; + #define PBCH_A 24 typedef struct { @@ -378,8 +394,12 @@ typedef struct { int if_freq_off; /// \brief Indicator that UE is synchronized to a gNB int is_synchronized; + /// \brief Indicator that UE is synchronized to a SyncRef UE on Sidelink + int is_synchronized_sl; /// \brief Target gNB Nid_cell when UE is resynchronizing int target_Nid_cell; + /// \brief Indicator that UE is an SynchRef UE + int sync_ref; /// Data structure for UE process scheduling UE_nr_proc_t proc; /// Flag to indicate the UE shouldn't do timing correction at all @@ -660,6 +680,7 @@ typedef struct nr_rxtx_thread_data_s { UE_nr_rxtx_proc_t proc; PHY_VARS_NR_UE *UE; int writeBlockSize; + notifiedFIFO_t txFifo; nr_phy_data_t phy_data; int tx_wait_for_dlsch; } nr_rxtx_thread_data_t; diff --git a/openair1/PHY/defs_nr_common.h b/openair1/PHY/defs_nr_common.h index e8dba45f2b..2b0a07fda7 100644 --- a/openair1/PHY/defs_nr_common.h +++ b/openair1/PHY/defs_nr_common.h @@ -169,6 +169,8 @@ struct NR_DL_FRAME_PARMS { /// Frame type (0 FDD, 1 TDD) frame_type_t frame_type; uint8_t tdd_config; + /// Sidelink Cell ID + uint16_t Nid_SL; /// Cell ID uint16_t Nid_cell; /// subcarrier spacing (15,30,60,120) @@ -260,6 +262,8 @@ struct NR_DL_FRAME_PARMS { uint8_t ssb_index; /// OFDM symbol offset divisor for UL uint32_t ofdm_offset_divisor; + uint16_t tdd_slot_config; + uint8_t tdd_period; }; // PRS config structures diff --git a/openair1/SIMULATION/NR_PHY/nr_dummy_functions.c b/openair1/SIMULATION/NR_PHY/nr_dummy_functions.c index dccdce2934..6d3948974d 100644 --- a/openair1/SIMULATION/NR_PHY/nr_dummy_functions.c +++ b/openair1/SIMULATION/NR_PHY/nr_dummy_functions.c @@ -32,6 +32,8 @@ void nr_mac_rrc_sync_ind(const module_id_t module_id, const frame_t frame, const bool in_sync) {} +void nr_mac_rrc_ra_ind(const module_id_t mod_id, int frame, bool success) {} + void rrc_data_ind(const protocol_ctxt_t *const ctxt_pP, const rb_id_t Srb_id, const sdu_size_t sdu_sizeP, diff --git a/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/pss_test.c b/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/pss_test.c index de1249457f..c9e76846e9 100644 --- a/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/pss_test.c +++ b/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/pss_test.c @@ -119,14 +119,15 @@ void test_synchro_pss_nr(PHY_VARS_NR_UE *PHY_VARS_NR_UE, int position_symbol, in /* search pss */ synchro_position = pss_synchro_nr(PHY_VARS_NR_UE, rate_change); + int pss_sequence = get_softmodem_params()->sl_mode == 0 ? NUMBER_PSS_SEQUENCE : NUMBER_PSS_SEQUENCE_SL; - if (pss_sequence_number < NUMBER_PSS_SEQUENCE) { + if (pss_sequence_number < pss_sequence) { NID2_value = pss_sequence_number; } else { - NID2_value = NUMBER_PSS_SEQUENCE; + NID2_value = pss_sequence; } - if (NID2_value < NUMBER_PSS_SEQUENCE) { + if (NID2_value < pss_sequence) { test->number_of_tests++; /* position should be adjusted with i&q samples which are successively stored as int16 in the received buffer */ int test_margin = PSS_DETECTION_MARGIN_MAX; /* warning correlation results give an offset position between 0 and 12 */ @@ -213,9 +214,10 @@ int main(int argc, char *argv[]) test_synchro_pss_nr(PHY_vars_UE, 0, INVALID_PSS_SEQUENCE, &test); #endif + int pss_sequence = get_softmodem_params()->sl_mode == 0 ? NUMBER_PSS_SEQUENCE : NUMBER_PSS_SEQUENCE_SL; for (int index_position = 0; index_position < size_test_position; index_position++) { - for (int number_pss_sequence = 0; number_pss_sequence < NUMBER_PSS_SEQUENCE; number_pss_sequence++) { + for (int number_pss_sequence = 0; number_pss_sequence < pss_sequence; number_pss_sequence++) { p_test_synchro_pss(PHY_vars_UE, test_position[index_position], number_pss_sequence, &test); diff --git a/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/pss_util_test.c b/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/pss_util_test.c index 9c96d39534..815c12a6a6 100644 --- a/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/pss_util_test.c +++ b/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/pss_util_test.c @@ -111,12 +111,15 @@ void phase_shift_samples(int16_t *samples, int length, int16_t phase_shift_re, i void display_data(int pss_sequence_number, int16_t *rxdata, int position) { #ifdef DEBUG_TEST_PSS - int16_t *pss_sequence[NUMBER_PSS_SEQUENCE] = {primary_synch0_time, primary_synch1_time, primary_synch2_time}; - int16_t *pss_sequence_time = pss_sequence[pss_sequence_number]; + int pss_sequence = get_softmodem_params()->sl_mode == 0 ? NUMBER_PSS_SEQUENCE : NUMBER_PSS_SEQUENCE_SL; + int16_t *pss_sequence[pss_sequence] = {primary_synch0_time, primary_synch1_time, primary_synch2_time}; + int16_t *pss_sequence_sl[pss_sequence] = {primary_synch0_time, primary_synch1_time}; + int16_t *pss_sequence_time = pss_sequence[pss_sequence_number];; + if (get_softmodem_params()->sl_mode != 0) { + pss_sequence_time = pss_sequence_sl[pss_sequence_number]; printf(" pss %6d data \n", pss_sequence_number); - for (int i = 0; i < 4; i++) { - if (pss_sequence_number < NUMBER_PSS_SEQUENCE) { + if (pss_sequence_number < pss_sequence) { printf("[i %6d] : %4d [i %6d] : %8i at address : %p \n", i, pss_sequence_time[2*i], (i + position), rxdata[2*i + (position*2)], &(rxdata[2*i + (position*2)])); printf("[q %6d] : %4d [q %6d] : %8i at address : %p \n", i, pss_sequence_time[2*i+1], (i + position), rxdata[2*i + 1 + (position*2)], &(rxdata[2*i + 1 + (position*2)])); } else { @@ -242,7 +245,8 @@ int init_test(unsigned char N_tx, unsigned char N_rx, unsigned char transmission int n_ssb_crb = 0; int ssb_subcarrier_offset = 0; nr_init_frame_parms_ue(frame_parms, mu, extended_prefix_flag, N_RB_DL, n_ssb_crb, ssb_subcarrier_offset); - PHY_vars_UE->frame_parms.Nid_cell = (3 * N_ID_1_NUMBER) + N_ID_2_NUMBER; /* set to unvalid value */ + int nid_2_num = get_softmodem_params()->sl_mode == 0 ? N_ID_2_NUMBER : N_ID_2_NUMBER_SL; + PHY_vars_UE->frame_parms.Nid_cell = (3 * N_ID_1_NUMBER) + nid_2_num; /* set to unvalid value */ //phy_init_nr_top(frame_parms); @@ -447,8 +451,8 @@ int set_pss_in_rx_buffer(PHY_VARS_NR_UE *PHY_vars_UE, int position_symbol, int p printf("This pss sequence can not be fully written in the received window \n"); return (-1); } - - if ((pss_sequence_number >= NUMBER_PSS_SEQUENCE) && (pss_sequence_number < 0)) { + int pss_sequence = get_softmodem_params()->sl_mode == 0 ? NUMBER_PSS_SEQUENCE : NUMBER_PSS_SEQUENCE_SL; + if ((pss_sequence_number >= pss_sequence) && (pss_sequence_number < 0)) { printf("Unknow pss sequence %d \n", pss_sequence_number); return (-1); } @@ -504,8 +508,9 @@ void set_sequence_pss(PHY_VARS_NR_UE *PHY_vars_UE, int position_symbol, int pss_ if (position_symbol < 0) { set_pss_in_rx_buffer_from_external_buffer(PHY_vars_UE, input_buffer); } + int pss_sequence = get_softmodem_params()->sl_mode == 0 ? NUMBER_PSS_SEQUENCE : NUMBER_PSS_SEQUENCE_SL; /* write pss sequence in received ue buffer */ - else if (pss_sequence_number < NUMBER_PSS_SEQUENCE) { + else if (pss_sequence_number < pss_sequence) { if (position_symbol > (samples_for_frame - frame_parms->ofdm_symbol_size)) { printf("This position for pss sequence %d is not supported because it exceeds the frame length %d!\n", position_symbol, samples_for_frame); exit(0); diff --git a/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/sss_test.c b/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/sss_test.c index 6f6594861a..fa2a8cffd1 100644 --- a/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/sss_test.c +++ b/openair1/SIMULATION/NR_UE_PHY/unit_tests/src/sss_test.c @@ -227,10 +227,10 @@ int main(int argc, char *argv[]) Nid2 = GET_NID2(Nid_cell[index]); Nid1 = GET_NID1(Nid_cell[index]); - + int nid_2_num = get_softmodem_params()->sl_mode == 0 ? N_ID_2_NUMBER : N_ID_2_NUMBER_SL; for (int position = 0; position < size_test_position; position++) { - PHY_vars_UE->frame_parms.Nid_cell = (3 * N_ID_1_NUMBER) + N_ID_2_NUMBER; /* set to invalid value */ + PHY_vars_UE->frame_parms.Nid_cell = (3 * N_ID_1_NUMBER) + nid_2_num; /* set to invalid value */ phase = (*p_test_synchro_pss_sss)(PHY_vars_UE, test_position[position], &frame_parms_gNB, &test); /* return phase index which gives phase error from an array */ diff --git a/openair2/COMMON/mac_messages_def.h b/openair2/COMMON/mac_messages_def.h index d4a923b90b..50e7a108c9 100644 --- a/openair2/COMMON/mac_messages_def.h +++ b/openair2/COMMON/mac_messages_def.h @@ -48,6 +48,8 @@ MESSAGE_DEF(RRC_MAC_MCCH_DATA_IND, MESSAGE_PRIORITY_MED_PLUS, RrcMacMcchDat MESSAGE_DEF(RRC_MAC_PCCH_DATA_REQ, MESSAGE_PRIORITY_MED_PLUS, RrcMacPcchDataReq, rrc_mac_pcch_data_req) +MESSAGE_DEF(NR_RRC_MAC_RA_IND, MESSAGE_PRIORITY_MED_PLUS, NRRrcMacRaInd, nr_rrc_mac_ra_ind) + /* RRC configures DRX context (MAC timers) of a UE */ MESSAGE_DEF(RRC_MAC_DRX_CONFIG_REQ, MESSAGE_PRIORITY_MED, rrc_mac_drx_config_req_t, rrc_mac_drx_config_req) diff --git a/openair2/COMMON/mac_messages_types.h b/openair2/COMMON/mac_messages_types.h index fde990b0fe..5703a886c0 100644 --- a/openair2/COMMON/mac_messages_types.h +++ b/openair2/COMMON/mac_messages_types.h @@ -54,7 +54,9 @@ #define RRC_MAC_MCCH_DATA_IND(mSGpTR) (mSGpTR)->ittiMsg.rrc_mac_mcch_data_ind #define RRC_MAC_PCCH_DATA_REQ(mSGpTR) (mSGpTR)->ittiMsg.rrc_mac_pcch_data_req -#define RRC_MAC_DRX_CONFIG_REQ(mSGpTR) (mSGpTR)->ittiMsg.rrc_mac_drx_config_req +#define NR_RRC_MAC_RA_IND(mSGpTR) (mSGpTR)->ittiMsg.nr_rrc_mac_ra_ind + +#define RRC_MAC_DRX_CONFIG_REQ(mSGpTR) (mSGpTR)->ittiMsg.rrc_mac_drx_config_req // Some constants from "LAYER2/MAC/defs.h" #define BCCH_SDU_SIZE (512) @@ -65,6 +67,12 @@ //-------------------------------------------------------------------------------------------// // Messages between RRC and MAC layers + +typedef struct NRRrcMacRaInd_s { + uint32_t frame; + bool RA_succeeded; +} NRRrcMacRaInd; + typedef struct RrcMacInSyncInd_s { uint32_t frame; uint8_t sub_frame; diff --git a/openair2/F1AP/f1ap_common.c b/openair2/F1AP/f1ap_common.c index 862e701b6d..f417a1c337 100644 --- a/openair2/F1AP/f1ap_common.c +++ b/openair2/F1AP/f1ap_common.c @@ -36,7 +36,7 @@ static f1ap_cudu_inst_t *f1_du_inst[NUMBER_OF_gNB_MAX]= {0}; static f1ap_cudu_inst_t *f1_cu_inst[NUMBER_OF_gNB_MAX]= {0}; uint8_t F1AP_get_next_transaction_identifier(instance_t mod_idP, instance_t cu_mod_idP) { - static uint8_t transaction_identifier[NUMBER_OF_gNB_MAX]; + static uint8_t transaction_identifier[NUMBER_OF_gNB_MAX] = {0}; transaction_identifier[mod_idP+cu_mod_idP] = (transaction_identifier[mod_idP+cu_mod_idP] + 1) % F1AP_TRANSACTION_IDENTIFIER_NUMBER; return transaction_identifier[mod_idP+cu_mod_idP]; diff --git a/openair2/F1AP/f1ap_cu_ue_context_management.c b/openair2/F1AP/f1ap_cu_ue_context_management.c index a38dd7c68f..8fdee52abb 100644 --- a/openair2/F1AP/f1ap_cu_ue_context_management.c +++ b/openair2/F1AP/f1ap_cu_ue_context_management.c @@ -693,7 +693,7 @@ int CU_handle_UE_CONTEXT_SETUP_RESPONSE(instance_t instance, F1AP_DLUPTNLInformation_ToBeSetup_Item_t *dl_up_tnl_info_p = (F1AP_DLUPTNLInformation_ToBeSetup_Item_t *)drbs_setup_item_p->dLUPTNLInformation_ToBeSetup_List.list.array[0]; F1AP_GTPTunnel_t *dl_up_tnl0 = dl_up_tnl_info_p->dLUPTNLInformation.choice.gTPTunnel; BIT_STRING_TO_TRANSPORT_LAYER_ADDRESS_IPv4(&dl_up_tnl0->transportLayerAddress, drb_p->up_dl_tnl[0].tl_address); - OCTET_STRING_TO_INT32(&dl_up_tnl0->gTP_TEID, drb_p->up_dl_tnl[0].teid); + OCTET_STRING_TO_UINT32(&dl_up_tnl0->gTP_TEID, drb_p->up_dl_tnl[0].teid); GtpuUpdateTunnelOutgoingAddressAndTeid(getCxt(CUtype, instance)->gtpInst, f1ap_ue_context_setup_resp->rnti, (ebi_t)drbs_setup_item_p->dRBID, @@ -1654,7 +1654,7 @@ int CU_handle_UE_CONTEXT_MODIFICATION_RESPONSE(instance_t instance, F1AP_DLUPTNLInformation_ToBeSetup_Item_t *dl_up_tnl_info_p = (F1AP_DLUPTNLInformation_ToBeSetup_Item_t *)drbs_setupmod_item_p->dLUPTNLInformation_ToBeSetup_List.list.array[0]; F1AP_GTPTunnel_t *dl_up_tnl0 = dl_up_tnl_info_p->dLUPTNLInformation.choice.gTPTunnel; BIT_STRING_TO_TRANSPORT_LAYER_ADDRESS_IPv4(&dl_up_tnl0->transportLayerAddress, drb_p->up_dl_tnl[0].tl_address); - OCTET_STRING_TO_INT32(&dl_up_tnl0->gTP_TEID, drb_p->up_dl_tnl[0].teid); + OCTET_STRING_TO_UINT32(&dl_up_tnl0->gTP_TEID, drb_p->up_dl_tnl[0].teid); GtpuUpdateTunnelOutgoingAddressAndTeid(getCxt(CUtype, instance)->gtpInst, f1ap_ue_context_modification_resp->rnti, (ebi_t)drbs_setupmod_item_p->dRBID, diff --git a/openair2/F1AP/f1ap_du_rrc_message_transfer.c b/openair2/F1AP/f1ap_du_rrc_message_transfer.c index 997105d92a..5fc8daef3f 100644 --- a/openair2/F1AP/f1ap_du_rrc_message_transfer.c +++ b/openair2/F1AP/f1ap_du_rrc_message_transfer.c @@ -219,7 +219,7 @@ int DU_send_INITIAL_UL_RRC_MESSAGE_TRANSFER(instance_t instanceP, ie6->id = F1AP_ProtocolIE_ID_id_TransactionID; ie6->criticality = F1AP_Criticality_ignore; ie6->value.present = F1AP_InitialULRRCMessageTransferIEs__value_PR_TransactionID; - ie6->value.choice.TransactionID = F1AP_get_next_transaction_identifier(f1ap_req(false, instanceP)->gNB_DU_id, f1ap_req(false, instanceP)->gNB_DU_id); + ie6->value.choice.TransactionID = F1AP_get_next_transaction_identifier(0, 0); /* encode */ if (f1ap_encode_pdu(&pdu, &buffer, &len) < 0) { diff --git a/openair2/F1AP/f1ap_du_ue_context_management.c b/openair2/F1AP/f1ap_du_ue_context_management.c index ccd3381e95..05833e3fad 100644 --- a/openair2/F1AP/f1ap_du_ue_context_management.c +++ b/openair2/F1AP/f1ap_du_ue_context_management.c @@ -161,7 +161,7 @@ int DU_handle_UE_CONTEXT_SETUP_REQUEST(instance_t instance, F1AP_ULUPTNLInformation_ToBeSetup_Item_t *ul_up_tnl_info_p = (F1AP_ULUPTNLInformation_ToBeSetup_Item_t *)drbs_tobesetup_item_p->uLUPTNLInformation_ToBeSetup_List.list.array[0]; F1AP_GTPTunnel_t *ul_up_tnl0 = ul_up_tnl_info_p->uLUPTNLInformation.choice.gTPTunnel; BIT_STRING_TO_TRANSPORT_LAYER_ADDRESS_IPv4(&ul_up_tnl0->transportLayerAddress, drb_p->up_ul_tnl[0].tl_address); - OCTET_STRING_TO_INT32(&ul_up_tnl0->gTP_TEID, drb_p->up_ul_tnl[0].teid); + OCTET_STRING_TO_UINT32(&ul_up_tnl0->gTP_TEID, drb_p->up_ul_tnl[0].teid); // 3GPP assumes GTP-U is on port 2152, but OAI is configurable drb_p->up_ul_tnl[0].port=getCxt(false,instance)->setupReq.CUport; @@ -877,7 +877,7 @@ int DU_handle_UE_CONTEXT_MODIFICATION_REQUEST(instance_t instance, uint32_t asso F1AP_ULUPTNLInformation_ToBeSetup_Item_t *ul_up_tnl_info_p = (F1AP_ULUPTNLInformation_ToBeSetup_Item_t *)drbs_tobesetupmod_item_p->uLUPTNLInformation_ToBeSetup_List.list.array[0]; F1AP_GTPTunnel_t *ul_up_tnl0 = ul_up_tnl_info_p->uLUPTNLInformation.choice.gTPTunnel; BIT_STRING_TO_TRANSPORT_LAYER_ADDRESS_IPv4(&ul_up_tnl0->transportLayerAddress, drb_p->up_ul_tnl[0].tl_address); - OCTET_STRING_TO_INT32(&ul_up_tnl0->gTP_TEID, drb_p->up_ul_tnl[0].teid); + OCTET_STRING_TO_UINT32(&ul_up_tnl0->gTP_TEID, drb_p->up_ul_tnl[0].teid); // 3GPP assumes GTP-U is on port 2152, but OAI is configurable drb_p->up_ul_tnl[0].port=getCxt(false,instance)->setupReq.CUport; diff --git a/openair2/LAYER2/NR_MAC_COMMON/nr_mac.h b/openair2/LAYER2/NR_MAC_COMMON/nr_mac.h index 9465d69e8b..c15a5cd194 100644 --- a/openair2/LAYER2/NR_MAC_COMMON/nr_mac.h +++ b/openair2/LAYER2/NR_MAC_COMMON/nr_mac.h @@ -483,19 +483,13 @@ typedef enum nr_ssb_and_cset_mux_pattern_type_e { NR_SSB_AND_CSET_MUX_PATTERN_TYPE3 } nr_ssb_and_cset_mux_pattern_type_t; -typedef enum { - SFN_C_MOD_2_EQ_0, - SFN_C_MOD_2_EQ_1, - SFN_C_IMPOSSIBLE -} SFN_C_TYPE; - typedef struct Type0_PDCCH_CSS_config_s { int32_t num_rbs; int32_t num_symbols; int32_t rb_offset; // Offset from SSB RB0 uint32_t type0_pdcch_ss_mux_pattern; uint16_t frame; - SFN_C_TYPE sfn_c; + int sfn_c; uint32_t n_c; uint32_t n_0; uint32_t number_of_search_space_per_slot; diff --git a/openair2/LAYER2/NR_MAC_COMMON/nr_mac_common.c b/openair2/LAYER2/NR_MAC_COMMON/nr_mac_common.c index 67d1e4fc5f..e9fa641a07 100644 --- a/openair2/LAYER2/NR_MAC_COMMON/nr_mac_common.c +++ b/openair2/LAYER2/NR_MAC_COMMON/nr_mac_common.c @@ -164,12 +164,12 @@ const int32_t table_38213_13_10_c4[16] = { 0, 8, 0, 8,-41, 25,-41, 49, reserv const float table_38213_13_11_c1[16] = { 0, 0, 2, 2, 5, 5, 7, 7, 0, 5, 0, 0, 2, 2, 5, 5}; // O const int32_t table_38213_13_11_c2[16] = { 1, 2, 1, 2, 1, 2, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1}; -const float table_38213_13_11_c3[16] = { 1, 0.5f, 1, 0.5f, 1, 0.5f, 1, 0.5f, 1, 1, 1, 1, 1, 1, 1, 1}; // M +const float table_38213_13_11_c3[16] = { 1, 0.5f, 1, 0.5f, 1, 0.5f, 1, 0.5f, 2, 2, 1, 1, 1, 1, 1, 1}; // M const int32_t table_38213_13_11_c4[16] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 1, 2, 1, 2}; // i is even as default const float table_38213_13_12_c1[16] = { 0, 0, 2.5f, 2.5f, 5, 5, 0, 2.5f, 5, 7.5f, 7.5f, 7.5f, 0, 5, reserved, reserved}; // O, index 14-15 reserved const int32_t table_38213_13_12_c2[16] = { 1, 2, 1, 2, 1, 2, 2, 2, 2, 1, 2, 2, 1, 1, reserved, reserved}; // index 14-15 reserved -const float table_38213_13_12_c3[16] = { 1, 0.5f, 1, 0.5f, 1, 0.5f, 0.5f, 0.5f, 0.5f, 1, 0.5f, 0.5f, 1, 1, reserved, reserved}; // M, index 14-15 reserved +const float table_38213_13_12_c3[16] = { 1, 0.5f, 1, 0.5f, 1, 0.5f, 0.5f, 0.5f, 0.5f, 1, 0.5f, 0.5f, 2, 2, reserved, reserved}; // M, index 14-15 reserved const int32_t table_38213_10_1_1_c2[5] = { 0, 0, 4, 2, 1 }; @@ -4301,7 +4301,7 @@ void get_type0_PDCCH_CSS_config_parameters(NR_Type0_PDCCH_CSS_config_t *type0_PD // type0-pdcch search space float big_o = 0.0f; float big_m = 0.0f; - type0_PDCCH_CSS_config->sfn_c = SFN_C_IMPOSSIBLE; // only valid for mux=1 + type0_PDCCH_CSS_config->sfn_c = -1; // only valid for mux=1 type0_PDCCH_CSS_config->n_c = UINT_MAX; type0_PDCCH_CSS_config->number_of_search_space_per_slot = UINT_MAX; type0_PDCCH_CSS_config->first_symbol_index = UINT_MAX; @@ -4316,11 +4316,7 @@ void get_type0_PDCCH_CSS_config_parameters(NR_Type0_PDCCH_CSS_config_t *type0_PD uint32_t temp = (uint32_t)(big_o*(1<ssb_index*big_m); type0_PDCCH_CSS_config->n_c = temp / num_slot_per_frame; - if((temp/num_slot_per_frame) & 0x1){ - type0_PDCCH_CSS_config->sfn_c = SFN_C_MOD_2_EQ_1; - }else{ - type0_PDCCH_CSS_config->sfn_c = SFN_C_MOD_2_EQ_0; - } + type0_PDCCH_CSS_config->sfn_c = type0_PDCCH_CSS_config->n_c % 2; if((index_4lsb == 1 || index_4lsb == 3 || index_4lsb == 5 || index_4lsb == 7) && (type0_PDCCH_CSS_config->ssb_index&1)){ type0_PDCCH_CSS_config->first_symbol_index = type0_PDCCH_CSS_config->num_symbols; @@ -4335,9 +4331,13 @@ void get_type0_PDCCH_CSS_config_parameters(NR_Type0_PDCCH_CSS_config_t *type0_PD if(type0_PDCCH_CSS_config->type0_pdcch_ss_mux_pattern == 1 && frequency_range == FR2){ big_o = table_38213_13_12_c1[index_4lsb]; - type0_PDCCH_CSS_config->number_of_search_space_per_slot = table_38213_13_11_c2[index_4lsb]; + type0_PDCCH_CSS_config->number_of_search_space_per_slot = table_38213_13_12_c2[index_4lsb]; big_m = table_38213_13_12_c3[index_4lsb]; + uint32_t temp = (uint32_t)(big_o*(1<ssb_index*big_m); + type0_PDCCH_CSS_config->n_c = temp / num_slot_per_frame; + type0_PDCCH_CSS_config->sfn_c = type0_PDCCH_CSS_config->n_c % 2; + if((index_4lsb == 1 || index_4lsb == 3 || index_4lsb == 5 || index_4lsb == 10) && (type0_PDCCH_CSS_config->ssb_index&1)){ type0_PDCCH_CSS_config->first_symbol_index = 7; }else if((index_4lsb == 6 || index_4lsb == 7 || index_4lsb == 8 || index_4lsb == 11) && (type0_PDCCH_CSS_config->ssb_index&1)){ @@ -4461,8 +4461,8 @@ void get_type0_PDCCH_CSS_config_parameters(NR_Type0_PDCCH_CSS_config_t *type0_PD // AssertFatal(first_symbol_index!=UINT_MAX,""); // mac->type0_pdcch_dci_config.monitoring_symbols_within_slot = (0x3fff << first_symbol_index) & (0x3fff >> (14-coreset_duration-first_symbol_index)) & 0x3fff; - AssertFatal(type0_PDCCH_CSS_config->sfn_c!=SFN_C_IMPOSSIBLE,""); - AssertFatal(type0_PDCCH_CSS_config->n_c!=UINT_MAX,""); + AssertFatal(type0_PDCCH_CSS_config->sfn_c >= 0, ""); + AssertFatal(type0_PDCCH_CSS_config->n_c != UINT_MAX, ""); type0_PDCCH_CSS_config->n_0 = ((uint32_t)(big_o*(1<ssb_index*big_m))%num_slot_per_frame; type0_PDCCH_CSS_config->cset_start_rb = ssb_offset_point_a - type0_PDCCH_CSS_config->rb_offset; diff --git a/openair2/LAYER2/NR_MAC_UE/mac_defs.h b/openair2/LAYER2/NR_MAC_UE/mac_defs.h index 99c7ba6dcd..4eee1bb3fe 100644 --- a/openair2/LAYER2/NR_MAC_UE/mac_defs.h +++ b/openair2/LAYER2/NR_MAC_UE/mac_defs.h @@ -420,7 +420,7 @@ typedef struct { /// Type0-PDCCH seach space fapi_nr_dl_config_dci_dl_pdu_rel15_t type0_pdcch_dci_config; uint32_t type0_pdcch_ss_mux_pattern; - SFN_C_TYPE type0_pdcch_ss_sfn_c; + int type0_pdcch_ss_sfn_c; uint32_t type0_pdcch_ss_n_c; uint32_t type0_pdcch_consecutive_slots; diff --git a/openair2/LAYER2/NR_MAC_UE/nr_ra_procedures.c b/openair2/LAYER2/NR_MAC_UE/nr_ra_procedures.c index 130b46143b..1c09ec574d 100644 --- a/openair2/LAYER2/NR_MAC_UE/nr_ra_procedures.c +++ b/openair2/LAYER2/NR_MAC_UE/nr_ra_procedures.c @@ -896,7 +896,6 @@ void nr_ra_succeeded(const module_id_t mod_id, const uint8_t gNB_index, const fr if (ra->cfra) { LOG_I(MAC, "[UE %d][%d.%d][RAPROC] RA procedure succeeded. CF-RA: RAR successfully received.\n", mod_id, frame, slot); - nr_rrc_RA_succeeded(mod_id, gNB_index); mac->state = UE_CONNECTED; ra->RA_window_cnt = -1; } else { @@ -911,6 +910,7 @@ void nr_ra_succeeded(const module_id_t mod_id, const uint8_t gNB_index, const fr LOG_D(MAC, "In %s: [UE %d] clearing RA_active flag...\n", __FUNCTION__, mod_id); ra->RA_active = 0; ra->ra_state = RA_SUCCEEDED; + nr_mac_rrc_ra_ind(mod_id, frame, true); } // Handling failure of RA procedure @ MAC layer diff --git a/openair2/LAYER2/NR_MAC_UE/nr_ue_dci_configuration.c b/openair2/LAYER2/NR_MAC_UE/nr_ue_dci_configuration.c index b3389df86a..8b99818508 100644 --- a/openair2/LAYER2/NR_MAC_UE/nr_ue_dci_configuration.c +++ b/openair2/LAYER2/NR_MAC_UE/nr_ue_dci_configuration.c @@ -96,9 +96,6 @@ void config_dci_pdu(NR_UE_MAC_INST_t *mac, const NR_SearchSpace_t *ss) { - uint16_t monitoringSymbolsWithinSlot = 0; - int sps = 0; - const NR_UE_DL_BWP_t *current_DL_BWP = &mac->current_DL_BWP; const NR_UE_UL_BWP_t *current_UL_BWP = &mac->current_UL_BWP; NR_BWP_Id_t dl_bwp_id = current_DL_BWP ? current_DL_BWP->bwp_id : 0; @@ -187,70 +184,72 @@ void config_dci_pdu(NR_UE_MAC_INST_t *mac, mac->cg, &mac->def_dci_pdu_rel15[dl_config->slot][dci_format], dci_format, - NR_RNTI_TC, + rnti_type, coreset, dl_bwp_id, ss->searchSpaceType->present, mac->type0_PDCCH_CSS_config.num_rbs, alt_size); - - rel15->BWPStart = coreset_id == 0 ? mac->type0_PDCCH_CSS_config.cset_start_rb : current_DL_BWP->BWPStart; - rel15->BWPSize = coreset_id == 0 ? mac->type0_PDCCH_CSS_config.num_rbs : current_DL_BWP->BWPSize; - - switch(rnti_type) { - case NR_RNTI_C: - // we use DL BWP dedicated - sps = current_DL_BWP->cyclicprefix ? 12 : 14; - // for SPS=14 8 MSBs in positions 13 down to 6 - monitoringSymbolsWithinSlot = (ss->monitoringSymbolsWithinSlot->buf[0]<<(sps-8)) | (ss->monitoringSymbolsWithinSlot->buf[1]>>(16-sps)); - rel15->rnti = mac->crnti; - rel15->SubcarrierSpacing = current_DL_BWP->scs; - break; - case NR_RNTI_RA: - // we use the initial DL BWP - sps = current_DL_BWP->cyclicprefix == NULL ? 14 : 12; - monitoringSymbolsWithinSlot = (ss->monitoringSymbolsWithinSlot->buf[0]<<(sps-8)) | (ss->monitoringSymbolsWithinSlot->buf[1]>>(16-sps)); - rel15->rnti = mac->ra.ra_rnti; - rel15->SubcarrierSpacing = current_DL_BWP->scs; - break; - case NR_RNTI_P: - break; - case NR_RNTI_CS: - break; - case NR_RNTI_TC: - // we use the initial DL BWP - sps = current_DL_BWP->cyclicprefix == NULL ? 14 : 12; - monitoringSymbolsWithinSlot = (ss->monitoringSymbolsWithinSlot->buf[0]<<(sps-8)) | (ss->monitoringSymbolsWithinSlot->buf[1]>>(16-sps)); - rel15->rnti = mac->ra.t_crnti; - rel15->SubcarrierSpacing = current_DL_BWP->scs; - break; - case NR_RNTI_SP_CSI: - break; - case NR_RNTI_SI: - sps=14; - // for SPS=14 8 MSBs in positions 13 down to 6 - monitoringSymbolsWithinSlot = (ss->monitoringSymbolsWithinSlot->buf[0]<<(sps-8)) | (ss->monitoringSymbolsWithinSlot->buf[1]>>(16-sps)); - - rel15->rnti = SI_RNTI; // SI-RNTI - 3GPP TS 38.321 Table 7.1-1: RNTI values - - rel15->SubcarrierSpacing = mac->mib->subCarrierSpacingCommon; - if(mac->frequency_range == FR2) - rel15->SubcarrierSpacing = mac->mib->subCarrierSpacingCommon + 2; - break; - case NR_RNTI_SFI: - break; - case NR_RNTI_INT: - break; - case NR_RNTI_TPC_PUSCH: - break; - case NR_RNTI_TPC_PUCCH: - break; - case NR_RNTI_TPC_SRS: - break; - default: - break; - } } + + rel15->BWPStart = coreset_id == 0 ? mac->type0_PDCCH_CSS_config.cset_start_rb : current_DL_BWP->BWPStart; + rel15->BWPSize = coreset_id == 0 ? mac->type0_PDCCH_CSS_config.num_rbs : current_DL_BWP->BWPSize; + + uint16_t monitoringSymbolsWithinSlot = 0; + int sps = 0; + + switch(rnti_type) { + case NR_RNTI_C: + // we use DL BWP dedicated + sps = current_DL_BWP->cyclicprefix ? 12 : 14; + // for SPS=14 8 MSBs in positions 13 down to 6 + monitoringSymbolsWithinSlot = (ss->monitoringSymbolsWithinSlot->buf[0]<<(sps-8)) | (ss->monitoringSymbolsWithinSlot->buf[1]>>(16-sps)); + rel15->rnti = mac->crnti; + rel15->SubcarrierSpacing = current_DL_BWP->scs; + break; + case NR_RNTI_RA: + // we use the initial DL BWP + sps = current_DL_BWP->cyclicprefix == NULL ? 14 : 12; + monitoringSymbolsWithinSlot = (ss->monitoringSymbolsWithinSlot->buf[0]<<(sps-8)) | (ss->monitoringSymbolsWithinSlot->buf[1]>>(16-sps)); + rel15->rnti = mac->ra.ra_rnti; + rel15->SubcarrierSpacing = current_DL_BWP->scs; + break; + case NR_RNTI_P: + break; + case NR_RNTI_CS: + break; + case NR_RNTI_TC: + // we use the initial DL BWP + sps = current_DL_BWP->cyclicprefix == NULL ? 14 : 12; + monitoringSymbolsWithinSlot = (ss->monitoringSymbolsWithinSlot->buf[0]<<(sps-8)) | (ss->monitoringSymbolsWithinSlot->buf[1]>>(16-sps)); + rel15->rnti = mac->ra.t_crnti; + rel15->SubcarrierSpacing = current_DL_BWP->scs; + break; + case NR_RNTI_SP_CSI: + break; + case NR_RNTI_SI: + sps=14; + // for SPS=14 8 MSBs in positions 13 down to 6 + monitoringSymbolsWithinSlot = (ss->monitoringSymbolsWithinSlot->buf[0]<<(sps-8)) | (ss->monitoringSymbolsWithinSlot->buf[1]>>(16-sps)); + rel15->rnti = SI_RNTI; // SI-RNTI - 3GPP TS 38.321 Table 7.1-1: RNTI values + rel15->SubcarrierSpacing = mac->mib->subCarrierSpacingCommon; + if(mac->frequency_range == FR2) + rel15->SubcarrierSpacing = mac->mib->subCarrierSpacingCommon + 2; + break; + case NR_RNTI_SFI: + break; + case NR_RNTI_INT: + break; + case NR_RNTI_TPC_PUSCH: + break; + case NR_RNTI_TPC_PUCCH: + break; + case NR_RNTI_TPC_SRS: + break; + default: + break; + } + for (int i = 0; i < sps; i++) { if ((monitoringSymbolsWithinSlot >> (sps - 1 - i)) & 1) { rel15->coreset.StartSymbolIndex = i; @@ -263,15 +262,17 @@ void config_dci_pdu(NR_UE_MAC_INST_t *mac, fill_dci_search_candidates(ss, rel15, Y); #ifdef DEBUG_DCI - LOG_D(MAC, "[DCI_CONFIG] Configure DCI PDU: rnti_type %d BWPSize %d BWPStart %d rel15->SubcarrierSpacing %d rel15->dci_format %d rel15->dci_length %d sps %d monitoringSymbolsWithinSlot %d \n", - rnti_type, - rel15->BWPSize, - rel15->BWPStart, - rel15->SubcarrierSpacing, - rel15->dci_format_options[0], - rel15->dci_length_options[0], - sps, - monitoringSymbolsWithinSlot); + for (int i = 0; i < rel15->num_dci_options; i++) { + LOG_D(MAC, "[DCI_CONFIG] Configure DCI PDU: rnti_type %d BWPSize %d BWPStart %d rel15->SubcarrierSpacing %d rel15->dci_format %d rel15->dci_length %d sps %d monitoringSymbolsWithinSlot %d \n", + rnti_type, + rel15->BWPSize, + rel15->BWPStart, + rel15->SubcarrierSpacing, + rel15->dci_format_options[i], + rel15->dci_length_options[i], + sps, + monitoringSymbolsWithinSlot); + } #endif // add DCI dl_config->dl_config_list[dl_config->number_pdus].pdu_type = FAPI_NR_DL_CONFIG_TYPE_DCI; diff --git a/openair2/LAYER2/NR_MAC_UE/nr_ue_procedures.c b/openair2/LAYER2/NR_MAC_UE/nr_ue_procedures.c index 6c8a0f337f..b1e0c17cc8 100644 --- a/openair2/LAYER2/NR_MAC_UE/nr_ue_procedures.c +++ b/openair2/LAYER2/NR_MAC_UE/nr_ue_procedures.c @@ -443,14 +443,14 @@ int nr_ue_process_dci_indication_pdu(module_id_t module_id,int cc_id, int gNB_in LOG_D(MAC,"Received dci indication (rnti %x,dci format %d,n_CCE %d,payloadSize %d,payload %llx)\n", dci->rnti,dci->dci_format,dci->n_CCE,dci->payloadSize,*(unsigned long long*)dci->payloadBits); - int8_t ret = nr_extract_dci_info(mac, dci->dci_format, dci->payloadSize, dci->rnti, dci->ss_type, (uint64_t *)dci->payloadBits, def_dci_pdu_rel15, slot); - if ((ret&1) == 1) return -1; + const int ret = nr_extract_dci_info(mac, dci->dci_format, dci->payloadSize, dci->rnti, dci->ss_type, (uint64_t *)dci->payloadBits, def_dci_pdu_rel15, slot); + if ((ret & 1) == 1) + return -1; else if (ret == 2) { - dci->dci_format = NR_UL_DCI_FORMAT_0_0; + dci->dci_format = (dci->dci_format == NR_UL_DCI_FORMAT_0_0) ? NR_DL_DCI_FORMAT_1_0 : NR_UL_DCI_FORMAT_0_0; def_dci_pdu_rel15 = &mac->def_dci_pdu_rel15[slot][dci->dci_format]; } - int8_t ret_proc = nr_ue_process_dci(module_id, cc_id, gNB_index, frame, slot, def_dci_pdu_rel15, dci); - return ret_proc; + return nr_ue_process_dci(module_id, cc_id, gNB_index, frame, slot, def_dci_pdu_rel15, dci); } int8_t nr_ue_process_dci(module_id_t module_id, int cc_id, uint8_t gNB_index, frame_t frame, int slot, dci_pdu_rel15_t *dci, fapi_nr_dci_indication_pdu_t *dci_ind) { @@ -653,6 +653,8 @@ int8_t nr_ue_process_dci(module_id_t module_id, int cc_id, uint8_t gNB_index, fr dl_config->dl_config_list[dl_config->number_pdus].dlsch_config_pdu.rnti = rnti; fapi_nr_dl_config_dlsch_pdu_rel15_t *dlsch_config_pdu_1_0 = &dl_config->dl_config_list[dl_config->number_pdus].dlsch_config_pdu.dlsch_config_rel15; + dlsch_config_pdu_1_0->pduBitmap = 0; + NR_PDSCH_Config_t *pdsch_config = current_DL_BWP ? current_DL_BWP->pdsch_Config : NULL; if (dci_ind->ss_type == NR_SearchSpace__searchSpaceType_PR_common) { dlsch_config_pdu_1_0->BWPSize = mac->type0_PDCCH_CSS_config.num_rbs ? mac->type0_PDCCH_CSS_config.num_rbs : current_DL_BWP->initial_BWPSize; @@ -2847,7 +2849,7 @@ static uint8_t nr_extract_dci_info(NR_UE_MAC_INST_t *mac, dci_pdu_rel15_t *dci_pdu_rel15, int slot) { - + LOG_D(MAC,"nr_extract_dci_info : dci_pdu %lx, size %d, format %d\n", *dci_pdu, dci_size, dci_format); int pos = 0; int fsize = 0; int rnti_type = get_rnti_type(mac, rnti); @@ -2864,7 +2866,7 @@ static uint8_t nr_extract_dci_info(NR_UE_MAC_INST_t *mac, current_DL_BWP->initial_BWPSize); else N_RB = mac->type0_PDCCH_CSS_config.num_rbs; - LOG_D(MAC,"nr_extract_dci_info : dci_pdu %lx, size %d\n",*dci_pdu,dci_size); + switch(dci_format) { case NR_DL_DCI_FORMAT_1_0: @@ -2914,7 +2916,7 @@ static uint8_t nr_extract_dci_info(NR_UE_MAC_INST_t *mac, //switch to DCI_0_0 if (dci_pdu_rel15->format_indicator == 0) { dci_pdu_rel15 = &mac->def_dci_pdu_rel15[slot][NR_UL_DCI_FORMAT_0_0]; - return 2+nr_extract_dci_info(mac, NR_UL_DCI_FORMAT_0_0, dci_size, rnti, ss_type, dci_pdu, dci_pdu_rel15, slot); + return 2 + nr_extract_dci_info(mac, NR_UL_DCI_FORMAT_0_0, dci_size, rnti, ss_type, dci_pdu, dci_pdu_rel15, slot); } #ifdef DEBUG_EXTRACT_DCI LOG_D(MAC,"Format indicator %d (%d bits) N_RB_BWP %d => %d (0x%lx)\n",dci_pdu_rel15->format_indicator,1,N_RB,dci_size-pos,*dci_pdu); @@ -3105,7 +3107,7 @@ static uint8_t nr_extract_dci_info(NR_UE_MAC_INST_t *mac, //switch to DCI_0_0 if (dci_pdu_rel15->format_indicator == 0) { dci_pdu_rel15 = &mac->def_dci_pdu_rel15[slot][NR_UL_DCI_FORMAT_0_0]; - return 2+nr_extract_dci_info(mac, NR_UL_DCI_FORMAT_0_0, dci_size, rnti, ss_type, dci_pdu, dci_pdu_rel15, slot); + return 2 + nr_extract_dci_info(mac, NR_UL_DCI_FORMAT_0_0, dci_size, rnti, ss_type, dci_pdu, dci_pdu_rel15, slot); } // Freq domain assignment 0-16 bit @@ -3246,8 +3248,13 @@ static uint8_t nr_extract_dci_info(NR_UE_MAC_INST_t *mac, #ifdef DEBUG_EXTRACT_DCI LOG_I(MAC,"Format indicator %d (%d bits)=> %d (0x%lx)\n",dci_pdu_rel15->format_indicator,1,dci_size-pos,*dci_pdu); #endif - if (dci_pdu_rel15->format_indicator == 1) - return 1; // discard dci, format indicator not corresponding to dci_format + + //switch to DCI_1_0 + if (dci_pdu_rel15->format_indicator == 1) { + dci_pdu_rel15 = &mac->def_dci_pdu_rel15[slot][NR_DL_DCI_FORMAT_1_0]; + return 2 + nr_extract_dci_info(mac, NR_DL_DCI_FORMAT_1_0, dci_size, rnti, ss_type, dci_pdu, dci_pdu_rel15, slot); + } + fsize = dci_pdu_rel15->frequency_domain_assignment.nbits; pos+=fsize; dci_pdu_rel15->frequency_domain_assignment.val = (*dci_pdu>>(dci_size-pos))&((1<state == UE_CONNECTED) { + if(mac->state > UE_NOT_SYNC) { dcireq.module_id = mod_id; dcireq.gNB_index = gNB_index; @@ -964,8 +964,10 @@ void nr_ue_dl_scheduler(nr_downlink_indication_t *dl_info) if(mac->ul_time_alignment.ta_apply) schedule_ta_command(dl_config, &mac->ul_time_alignment); - nr_schedule_csirs_reception(mac, rx_frame, rx_slot); - nr_schedule_csi_for_im(mac, rx_frame, rx_slot); + if(mac->state == UE_CONNECTED) { + nr_schedule_csirs_reception(mac, rx_frame, rx_slot); + nr_schedule_csi_for_im(mac, rx_frame, rx_slot); + } dcireq.dl_config_req = *dl_config; fill_scheduled_response(&scheduled_response, &dcireq.dl_config_req, NULL, NULL, mod_id, cc_id, rx_frame, rx_slot, dl_info->phy_data); @@ -974,19 +976,6 @@ void nr_ue_dl_scheduler(nr_downlink_indication_t *dl_info) mac->if_module->scheduled_response(&scheduled_response); } } - else if (mac->state == UE_PERFORMING_RA) { - // this is for Msg2/Msg4 - if (mac->ra.ra_state >= WAIT_RAR) { - if(mac->ul_time_alignment.ta_apply) - schedule_ta_command(dl_config, &mac->ul_time_alignment); - config_dci_pdu(mac, dl_config, mac->ra.ra_state == WAIT_RAR ? NR_RNTI_RA : NR_RNTI_TC , rx_slot, mac->ra_SS); - dl_config->number_pdus = 1; - LOG_D(MAC,"mac->cg %p: Calling fill_scheduled_response for type0_pdcch, num_pdus %d\n", mac->cg, dl_config->number_pdus); - fill_scheduled_response(&scheduled_response, dl_config, NULL, NULL, mod_id, cc_id, rx_frame, rx_slot, dl_info->phy_data); - if(mac->if_module != NULL && mac->if_module->scheduled_response != NULL) - mac->if_module->scheduled_response(&scheduled_response); - } - } else dl_config->number_pdus = 0; } diff --git a/openair2/RRC/NR/cucp_cuup_direct.c b/openair2/RRC/NR/cucp_cuup_direct.c index 4591537304..347e9fbff5 100644 --- a/openair2/RRC/NR/cucp_cuup_direct.c +++ b/openair2/RRC/NR/cucp_cuup_direct.c @@ -85,16 +85,13 @@ void CU_update_UP_DL_tunnel(e1ap_bearer_setup_req_t *const req, instance_t insta for (int j=0; j < req->pduSessionMod[i].numDRB2Modify; j++) { DRB_nGRAN_to_setup_t *drb_p = req->pduSessionMod[i].DRBnGRanModList + j; - transport_layer_addr_t newRemoteAddr; - newRemoteAddr.length = 32; // IPv4 - memcpy(newRemoteAddr.buffer, - &drb_p->DlUpParamList[0].tlAddress, - sizeof(in_addr_t)); + in_addr_t addr = {0}; + memcpy(&addr, &drb_p->DlUpParamList[0].tlAddress, sizeof(in_addr_t)); GtpuUpdateTunnelOutgoingAddressAndTeid(instance, (ue_id & 0xFFFF), (ebi_t)drb_p->id, - *(in_addr_t*)&newRemoteAddr.buffer, + addr, drb_p->DlUpParamList[0].teId); } } diff --git a/openair2/RRC/NR_UE/L2_interface_ue.c b/openair2/RRC/NR_UE/L2_interface_ue.c index 7222defbe4..06fb26536e 100644 --- a/openair2/RRC/NR_UE/L2_interface_ue.c +++ b/openair2/RRC/NR_UE/L2_interface_ue.c @@ -155,12 +155,10 @@ int8_t nr_mac_rrc_data_req_ue(const module_id_t Mod_idP, return 0; } -int8_t nr_rrc_RA_succeeded(const module_id_t mod_id, const uint8_t gNB_index) +void nr_mac_rrc_ra_ind(const module_id_t mod_id, int frame, bool success) { - if (NR_UE_rrc_inst[mod_id].timers_and_constants.T304_active == true) { - LOG_W(NR_RRC, "T304 was stoped with value %i\n", NR_UE_rrc_inst[mod_id].timers_and_constants.T304_cnt); - NR_UE_rrc_inst[mod_id].timers_and_constants.T304_active = false; - NR_UE_rrc_inst[mod_id].timers_and_constants.T304_cnt = 0; - } - return 0; + MessageDef *message_p = itti_alloc_new_message(TASK_MAC_UE, 0, NR_RRC_MAC_RA_IND); + NR_RRC_MAC_RA_IND (message_p).frame = frame; + NR_RRC_MAC_RA_IND (message_p).RA_succeeded = success; + itti_send_msg_to_task(TASK_RRC_NRUE, GNB_MODULE_ID_TO_INSTANCE(mod_id), message_p); } diff --git a/openair2/RRC/NR_UE/rrc_UE.c b/openair2/RRC/NR_UE/rrc_UE.c index 69efec94ce..704ea605e4 100644 --- a/openair2/RRC/NR_UE/rrc_UE.c +++ b/openair2/RRC/NR_UE/rrc_UE.c @@ -1208,7 +1208,9 @@ void nr_rrc_ue_process_masterCellGroup(const protocol_ctxt_t *const ctxt_pP, set_default_timers_and_constants(&rrc->timers_and_constants); LOG_A(NR_RRC, "Received the reconfigurationWithSync in %s\n", __FUNCTION__); NR_ReconfigurationWithSync_t *reconfigurationWithSync = cellGroupConfig->spCellConfig->reconfigurationWithSync; + rrc->timers_and_constants.T304_active = true; nr_rrc_set_T304(&rrc->timers_and_constants, reconfigurationWithSync); + // TODO: Implementation of the remaining procedures regarding the reception of the reconfigurationWithSync, TS 38.331 - Section 5.3.5.5.2 } if (rrc->cell_group_config && @@ -2249,6 +2251,18 @@ int32_t nr_rrc_ue_establish_drb(module_id_t ue_mod_idP, return 0; } +void nr_rrc_handle_ra_indication(unsigned int mod_id, bool ra_succeeded) +{ + NR_UE_Timers_Constants_t *timers = &NR_UE_rrc_inst[mod_id].timers_and_constants; + if (ra_succeeded && timers->T304_active == true) { + // successful Random Access procedure triggered by reconfigurationWithSync + timers->T304_active = false; + timers->T304_cnt = 0; + // TODO handle the rest of procedures as described in 5.3.5.3 for when + // reconfigurationWithSync is included in spCellConfig + } +} + void *rrc_nrue_task(void *args_p) { MessageDef *msg_p; @@ -2287,12 +2301,21 @@ void *rrc_nrue_task(void *args_p) break; case NRRRC_SLOT_PROCESS: - LOG_D(NR_RRC, "[UE %d] Receided %s: frame %d slot %d\n", + LOG_D(NR_RRC, "[UE %d] Received %s: frame %d slot %d\n", ue_mod_id, ITTI_MSG_NAME (msg_p), NRRRC_SLOT_PROCESS (msg_p).frame, NRRRC_SLOT_PROCESS (msg_p).slot); NR_UE_Timers_Constants_t *timers = &NR_UE_rrc_inst[ue_mod_id].timers_and_constants; nr_rrc_handle_timers(timers); break; + case NR_RRC_MAC_RA_IND: + LOG_D(NR_RRC, "[UE %d] Received %s: frame %d\n RA %s", + ue_mod_id, + ITTI_MSG_NAME (msg_p), + NR_RRC_MAC_RA_IND (msg_p).frame, + NR_RRC_MAC_RA_IND (msg_p).RA_succeeded ? "successful" : "failed"); + nr_rrc_handle_ra_indication(ue_mod_id, NR_RRC_MAC_RA_IND (msg_p).RA_succeeded); + break; + case NR_RRC_MAC_BCCH_DATA_IND: LOG_D(NR_RRC, "[UE %d] Received %s: frameP %d, gNB %d\n", ue_mod_id, ITTI_MSG_NAME (msg_p), NR_RRC_MAC_BCCH_DATA_IND (msg_p).frame, NR_RRC_MAC_BCCH_DATA_IND (msg_p).gnb_index); diff --git a/openair2/RRC/NR_UE/rrc_proto.h b/openair2/RRC/NR_UE/rrc_proto.h index f303d809fe..67369a8488 100644 --- a/openair2/RRC/NR_UE/rrc_proto.h +++ b/openair2/RRC/NR_UE/rrc_proto.h @@ -116,6 +116,7 @@ int8_t nr_mac_rrc_data_ind_ue(const module_id_t module_id, void nr_mac_rrc_sync_ind(const module_id_t module_id, const frame_t frame, const bool in_sync); +void nr_mac_rrc_ra_ind(const module_id_t mod_id, int frame, bool success); /**\brief \param module_id module id diff --git a/openair3/UTILS/conversions.h b/openair3/UTILS/conversions.h index 2c544c94e0..bf24c9f4ee 100644 --- a/openair3/UTILS/conversions.h +++ b/openair3/UTILS/conversions.h @@ -119,10 +119,10 @@ do { \ /* Convert an array of char containing vALUE to x */ #define BUFFER_TO_UINT32(buf, x) \ do { \ - x = (((uint32_t)(buf)[0] << 24) | \ - (uint32_t)((buf)[1] << 16) | \ - (uint32_t)((buf)[2] << 8) | \ - (uint32_t)((buf)[3]); \ + x = (((uint32_t)((buf)[0])) << 24) | \ + (((uint32_t)((buf)[1])) << 16) | \ + (((uint32_t)((buf)[2])) << 8) | \ + (((uint32_t)((buf)[3]))); \ } while (0) /* Convert an integer on 32 bits to an octet string from aSN1c tool */