maki/openairinterface5g
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix pdsch UE variables memory allocation

Browse Source
This commit is contained in:
sfn
2021-03-12 09:19:44 +01:00
parent d2f353acf6
commit 3a83e8df03
3 changed files with 114 additions and 109 deletions
Download Patch File Download Diff File Expand all files Collapse all files
openair1/PHY/INIT/nr_init_ue.c
+19 -16
View File
@@ -71,31 +71,34 @@ void phy_init_nr_ue__PDSCH(NR_UE_PDSCH *const pdsch,
pdsch->llr128 = (int16_t **)malloc16_clear( sizeof(int16_t *) );
// FIXME! no further allocation for (int16_t*)pdsch->llr128 !!! expect SIGSEGV
// FK, 11-3-2015: this is only as a temporary pointer, no memory is stored there
pdsch->rxdataF_ext = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->rxdataF_uespec_pilots = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->rxdataF_comp0 = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->rho = (int32_t **)malloc16_clear( fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_estimates = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_estimates_ext = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_bf_ch_estimates = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_bf_ch_estimates_ext = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->rxdataF_ext = (int32_t **)malloc16_clear( fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->rxdataF_uespec_pilots = (int32_t **)malloc16_clear( fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->rxdataF_comp0 = (int32_t **)malloc16_clear( NR_MAX_NB_LAYERS*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->rho = (int32_t ***)malloc16_clear( fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_estimates = (int32_t **)malloc16_clear( NR_MAX_NB_LAYERS*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_estimates_ext = (int32_t **)malloc16_clear( NR_MAX_NB_LAYERS*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_bf_ch_estimates = (int32_t **)malloc16_clear( NR_MAX_NB_LAYERS*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_bf_ch_estimates_ext = (int32_t **)malloc16_clear( NR_MAX_NB_LAYERS*fp->nb_antennas_rx*sizeof(int32_t *) );
//pdsch->dl_ch_rho_ext = (int32_t**)malloc16_clear( 8*sizeof(int32_t*) );
//pdsch->dl_ch_rho2_ext = (int32_t**)malloc16_clear( 8*sizeof(int32_t*) );
pdsch->dl_ch_mag0 = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_magb0 = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_magr0 = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->ptrs_phase_per_slot = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->ptrs_re_per_slot = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_ptrs_estimates_ext = (int32_t **)malloc16_clear( 4*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_mag0 = (int32_t **)malloc16_clear( NR_MAX_NB_LAYERS*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_magb0 = (int32_t **)malloc16_clear( NR_MAX_NB_LAYERS*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_magr0 = (int32_t **)malloc16_clear( NR_MAX_NB_LAYERS*fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->ptrs_phase_per_slot = (int32_t **)malloc16_clear( fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->ptrs_re_per_slot = (int32_t **)malloc16_clear( fp->nb_antennas_rx*sizeof(int32_t *) );
pdsch->dl_ch_ptrs_estimates_ext = (int32_t **)malloc16_clear( fp->nb_antennas_rx*sizeof(int32_t *) );
// the allocated memory size is fixed:
AssertFatal( fp->nb_antennas_rx <= 4, "nb_antennas_rx > 4" );//Extend the max number of UE Rx antennas to 4
for (int i=0; i<fp->nb_antennas_rx; i++) {
pdsch->rho[i] = (int32_t *)malloc16_clear( sizeof(int32_t)*(fp->N_RB_DL*12*7*2) );
pdsch->rho[i] = (int32_t **)malloc16_clear( NR_MAX_NB_LAYERS*NR_MAX_NB_LAYERS*sizeof(int32_t) );
for (int j=0; j<4; j++) { //4: DL antenna ports
for (int j=0; j<NR_MAX_NB_LAYERS; j++) {
const int idx = (j*fp->nb_antennas_rx)+i;
const size_t num = 7*2*fp->N_RB_DL*12;
for (int k=0; k<NR_MAX_NB_LAYERS; k++) {
pdsch->rho[i][j*NR_MAX_NB_LAYERS+k] = (int32_t *)malloc16_clear( sizeof(int32_t) * num );
}
pdsch->rxdataF_ext[idx] = (int32_t *)malloc16_clear( sizeof(int32_t) * num );
pdsch->rxdataF_uespec_pilots[idx] = (int32_t *)malloc16_clear( sizeof(int32_t) * fp->N_RB_DL*12);
pdsch->rxdataF_comp0[idx] = (int32_t *)malloc16_clear( sizeof(int32_t) * num );
openair1/PHY/NR_UE_TRANSPORT/nr_dlsch_demodulation.c
+93 -91
View File
@@ -776,7 +776,7 @@ void nr_dlsch_channel_compensation(int **rxdataF_ext,
int **dl_ch_magb,
int **dl_ch_magr,
int **rxdataF_comp,
int **rho,
int ***rho,
NR_DL_FRAME_PARMS *frame_parms,
uint8_t nb_aatx,
unsigned char symbol,
@@ -791,7 +791,7 @@ void nr_dlsch_channel_compensation(int **rxdataF_ext,
#if defined(__i386) || defined(__x86_64)
unsigned short rb;
unsigned char aatx,aarx;
unsigned char aatx,aarx,atx;
__m128i *dl_ch128,*dl_ch128_2,*dl_ch_mag128,*dl_ch_mag128b,*dl_ch_mag128r,*rxdataF128,*rxdataF_comp128,*rho128;
__m128i mmtmpD0,mmtmpD1,mmtmpD2,mmtmpD3,QAM_amp128,QAM_amp128b,QAM_amp128r;
QAM_amp128b = _mm_setzero_si128();
@@ -964,124 +964,126 @@ void nr_dlsch_channel_compensation(int **rxdataF_ext,
}
}
if (rho) {
//rho[aatx*n_rx+0] = Cov(H_0,0,H*_0,aatx), aatx=0,...n_tx-1
//rho[aatx*n_rx+1] = Cov(H_1,1,H*_1,aatx), aatx=0,...n_tx-1
//rho[aatx*n_rx+aarx] = Cov(H_aarx,arrx,H*_aarx,aatx), aatx=0,...n_tx-1
//we compute the Tx correlation matrix for each Rx antenna
//As an example the 2x2 MIMO case requires
//rho[aarx][nb_aatx*nb_aatx] = [cov(H_aarx_0,H_aarx_0) cov(H_aarx_0,H_aarx_1)
// cov(H_aarx_1,H_aarx_0) cov(H_aarx_1,H_aarx_1)], aarx=0,...,nb_antennas_rx-1
int avg_rho_re[nb_aatx][frame_parms->nb_antennas_rx];
int avg_rho_im[nb_aatx][frame_parms->nb_antennas_rx];
int avg_rho_re[frame_parms->nb_antennas_rx][nb_aatx];
int avg_rho_im[frame_parms->nb_antennas_rx][nb_aatx];
for (aarx=0; aarx<frame_parms->nb_antennas_rx; aarx++) {
dl_ch128 = (__m128i *)&dl_ch_estimates_ext[aarx][symbol*nb_rb*12];//H_{aarx,0}
for (aatx=1; aatx<nb_aatx; aatx++) {
avg_rho_re[aatx][aarx] = 0;
avg_rho_im[aatx][aarx] = 0;
rho128 = (__m128i *)&rho[aatx*frame_parms->nb_antennas_rx+aarx][symbol*nb_rb*12];
dl_ch128_2 = (__m128i *)&dl_ch_estimates_ext[aatx*frame_parms->nb_antennas_rx+aarx][symbol*nb_rb*12];
for (aatx=0; aatx<nb_aatx; aatx++) {
dl_ch128 = (__m128i *)&dl_ch_estimates_ext[aatx*frame_parms->nb_antennas_rx+aarx][symbol*nb_rb*12];
for (rb=0; rb<nb_rb; rb++) {
// multiply by conjugated channel
mmtmpD0 = _mm_madd_epi16(dl_ch128[0],dl_ch128_2[0]);
// print_ints("re",&mmtmpD0);
// mmtmpD0 contains real part of 4 consecutive outputs (32-bit)
mmtmpD1 = _mm_shufflelo_epi16(dl_ch128[0],_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_shufflehi_epi16(mmtmpD1,_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_sign_epi16(mmtmpD1,*(__m128i*)&conjugate[0]);
// print_ints("im",&mmtmpD1);
mmtmpD1 = _mm_madd_epi16(mmtmpD1,dl_ch128_2[0]);
// mmtmpD1 contains imag part of 4 consecutive outputs (32-bit)
mmtmpD0 = _mm_srai_epi32(mmtmpD0,output_shift);
// print_ints("re(shift)",&mmtmpD0);
mmtmpD1 = _mm_srai_epi32(mmtmpD1,output_shift);
// print_ints("im(shift)",&mmtmpD1);
mmtmpD2 = _mm_unpacklo_epi32(mmtmpD0,mmtmpD1);
mmtmpD3 = _mm_unpackhi_epi32(mmtmpD0,mmtmpD1);
// print_ints("c0",&mmtmpD2);
// print_ints("c1",&mmtmpD3);
rho128[0] = _mm_packs_epi32(mmtmpD2,mmtmpD3);
//print_shorts("rx:",dl_ch128_2);
//print_shorts("ch:",dl_ch128);
//print_shorts("pack:",rho128);
for (atx=0; atx<nb_aatx; atx++) {
avg_rho_re[aarx][aatx*nb_aatx+atx] = 0;
avg_rho_im[aarx][aatx*nb_aatx+atx] = 0;
rho128 = (__m128i *)&rho[aarx][aatx*nb_aatx+atx][symbol*nb_rb*12];
dl_ch128_2 = (__m128i *)&dl_ch_estimates_ext[atx*frame_parms->nb_antennas_rx+aarx][symbol*nb_rb*12];
avg_rho_re[aatx][aarx] +=(((int16_t*)&rho128[0])[0]+
for (rb=0; rb<nb_rb; rb++) {
// multiply by conjugated channel
mmtmpD0 = _mm_madd_epi16(dl_ch128[0],dl_ch128_2[0]);
// print_ints("re",&mmtmpD0);
// mmtmpD0 contains real part of 4 consecutive outputs (32-bit)
mmtmpD1 = _mm_shufflelo_epi16(dl_ch128[0],_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_shufflehi_epi16(mmtmpD1,_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_sign_epi16(mmtmpD1,*(__m128i*)&conjugate[0]);
// print_ints("im",&mmtmpD1);
mmtmpD1 = _mm_madd_epi16(mmtmpD1,dl_ch128_2[0]);
// mmtmpD1 contains imag part of 4 consecutive outputs (32-bit)
mmtmpD0 = _mm_srai_epi32(mmtmpD0,output_shift);
// print_ints("re(shift)",&mmtmpD0);
mmtmpD1 = _mm_srai_epi32(mmtmpD1,output_shift);
// print_ints("im(shift)",&mmtmpD1);
mmtmpD2 = _mm_unpacklo_epi32(mmtmpD0,mmtmpD1);
mmtmpD3 = _mm_unpackhi_epi32(mmtmpD0,mmtmpD1);
// print_ints("c0",&mmtmpD2);
// print_ints("c1",&mmtmpD3);
rho128[0] = _mm_packs_epi32(mmtmpD2,mmtmpD3);
//print_shorts("rx:",dl_ch128_2);
//print_shorts("ch:",dl_ch128);
//print_shorts("pack:",rho128);
avg_rho_re[aarx][aatx*nb_aatx+atx] +=(((int16_t*)&rho128[0])[0]+
((int16_t*)&rho128[0])[2] +
((int16_t*)&rho128[0])[4] +
((int16_t*)&rho128[0])[6])/16;
((int16_t*)&rho128[0])[6])/16;//
avg_rho_im[aatx][aarx] +=(((int16_t*)&rho128[0])[1]+
avg_rho_im[aarx][aatx*nb_aatx+atx] +=(((int16_t*)&rho128[0])[1]+
((int16_t*)&rho128[0])[3] +
((int16_t*)&rho128[0])[5] +
((int16_t*)&rho128[0])[7])/16;
((int16_t*)&rho128[0])[7])/16;//
// multiply by conjugated channel
mmtmpD0 = _mm_madd_epi16(dl_ch128[1],dl_ch128_2[1]);
// mmtmpD0 contains real part of 4 consecutive outputs (32-bit)
mmtmpD1 = _mm_shufflelo_epi16(dl_ch128[1],_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_shufflehi_epi16(mmtmpD1,_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_sign_epi16(mmtmpD1,*(__m128i*)conjugate);
mmtmpD1 = _mm_madd_epi16(mmtmpD1,dl_ch128_2[1]);
// mmtmpD1 contains imag part of 4 consecutive outputs (32-bit)
mmtmpD0 = _mm_srai_epi32(mmtmpD0,output_shift);
mmtmpD1 = _mm_srai_epi32(mmtmpD1,output_shift);
mmtmpD2 = _mm_unpacklo_epi32(mmtmpD0,mmtmpD1);
mmtmpD3 = _mm_unpackhi_epi32(mmtmpD0,mmtmpD1);
rho128[1] =_mm_packs_epi32(mmtmpD2,mmtmpD3);
//print_shorts("rx:",dl_ch128_2+1);
//print_shorts("ch:",dl_ch128+1);
//print_shorts("pack:",rho128+1);
// multiply by conjugated channel
mmtmpD0 = _mm_madd_epi16(dl_ch128[1],dl_ch128_2[1]);
// mmtmpD0 contains real part of 4 consecutive outputs (32-bit)
mmtmpD1 = _mm_shufflelo_epi16(dl_ch128[1],_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_shufflehi_epi16(mmtmpD1,_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_sign_epi16(mmtmpD1,*(__m128i*)conjugate);
mmtmpD1 = _mm_madd_epi16(mmtmpD1,dl_ch128_2[1]);
// mmtmpD1 contains imag part of 4 consecutive outputs (32-bit)
mmtmpD0 = _mm_srai_epi32(mmtmpD0,output_shift);
mmtmpD1 = _mm_srai_epi32(mmtmpD1,output_shift);
mmtmpD2 = _mm_unpacklo_epi32(mmtmpD0,mmtmpD1);
mmtmpD3 = _mm_unpackhi_epi32(mmtmpD0,mmtmpD1);
rho128[1] =_mm_packs_epi32(mmtmpD2,mmtmpD3);
//print_shorts("rx:",dl_ch128_2+1);
//print_shorts("ch:",dl_ch128+1);
//print_shorts("pack:",rho128+1);
// multiply by conjugated channel
avg_rho_re[aatx][aarx] +=(((int16_t*)&rho128[1])[0]+
// multiply by conjugated channel
avg_rho_re[aarx][aatx*nb_aatx+atx] +=(((int16_t*)&rho128[1])[0]+
((int16_t*)&rho128[1])[2] +
((int16_t*)&rho128[1])[4] +
((int16_t*)&rho128[1])[6])/16;
avg_rho_im[aatx][aarx] +=(((int16_t*)&rho128[1])[1]+
avg_rho_im[aarx][aatx*nb_aatx+atx] +=(((int16_t*)&rho128[1])[1]+
((int16_t*)&rho128[1])[3] +
((int16_t*)&rho128[1])[5] +
((int16_t*)&rho128[1])[7])/16;
mmtmpD0 = _mm_madd_epi16(dl_ch128[2],dl_ch128_2[2]);
// mmtmpD0 contains real part of 4 consecutive outputs (32-bit)
mmtmpD1 = _mm_shufflelo_epi16(dl_ch128[2],_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_shufflehi_epi16(mmtmpD1,_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_sign_epi16(mmtmpD1,*(__m128i*)conjugate);
mmtmpD1 = _mm_madd_epi16(mmtmpD1,dl_ch128_2[2]);
// mmtmpD1 contains imag part of 4 consecutive outputs (32-bit)
mmtmpD0 = _mm_srai_epi32(mmtmpD0,output_shift);
mmtmpD1 = _mm_srai_epi32(mmtmpD1,output_shift);
mmtmpD2 = _mm_unpacklo_epi32(mmtmpD0,mmtmpD1);
mmtmpD3 = _mm_unpackhi_epi32(mmtmpD0,mmtmpD1);
mmtmpD0 = _mm_madd_epi16(dl_ch128[2],dl_ch128_2[2]);
// mmtmpD0 contains real part of 4 consecutive outputs (32-bit)
mmtmpD1 = _mm_shufflelo_epi16(dl_ch128[2],_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_shufflehi_epi16(mmtmpD1,_MM_SHUFFLE(2,3,0,1));
mmtmpD1 = _mm_sign_epi16(mmtmpD1,*(__m128i*)conjugate);
mmtmpD1 = _mm_madd_epi16(mmtmpD1,dl_ch128_2[2]);
// mmtmpD1 contains imag part of 4 consecutive outputs (32-bit)
mmtmpD0 = _mm_srai_epi32(mmtmpD0,output_shift);
mmtmpD1 = _mm_srai_epi32(mmtmpD1,output_shift);
mmtmpD2 = _mm_unpacklo_epi32(mmtmpD0,mmtmpD1);
mmtmpD3 = _mm_unpackhi_epi32(mmtmpD0,mmtmpD1);
rho128[2] = _mm_packs_epi32(mmtmpD2,mmtmpD3);
//print_shorts("rx:",dl_ch128_2+2);
//print_shorts("ch:",dl_ch128+2);
//print_shorts("pack:",rho128+2);
avg_rho_re[aatx][aarx] +=(((int16_t*)&rho128[2])[0]+
rho128[2] = _mm_packs_epi32(mmtmpD2,mmtmpD3);
//print_shorts("rx:",dl_ch128_2+2);
//print_shorts("ch:",dl_ch128+2);
//print_shorts("pack:",rho128+2);
avg_rho_re[aarx][aatx*nb_aatx+atx] +=(((int16_t*)&rho128[2])[0]+
((int16_t*)&rho128[2])[2] +
((int16_t*)&rho128[2])[4] +
((int16_t*)&rho128[2])[6])/16;
avg_rho_im[aatx][aarx] +=(((int16_t*)&rho128[2])[1]+
avg_rho_im[aarx][aatx*nb_aatx+atx] +=(((int16_t*)&rho128[2])[1]+
((int16_t*)&rho128[2])[3] +
((int16_t*)&rho128[2])[5] +
((int16_t*)&rho128[2])[7])/16;
dl_ch128+=3;
dl_ch128_2+=3;
rho128+=3;
dl_ch128+=3;
dl_ch128_2+=3;
rho128+=3;
}
if (first_symbol_flag==1) {
//measurements->rx_correlation[0][0][aarx] = signal_energy(&rho[aarx][aatx*nb_aatx+atx][symbol*nb_rb*12],rb*12);
avg_rho_re[aarx][aatx*nb_aatx+atx] = 16*avg_rho_re[aarx][aatx*nb_aatx+atx]/(nb_rb*12);
avg_rho_im[aarx][aatx*nb_aatx+atx] = 16*avg_rho_im[aarx][aatx*nb_aatx+atx]/(nb_rb*12);
printf("rho[rx]%d tx%d tx%d = Re: %d Im: %d\n",aarx, aatx,atx, avg_rho_re[aatx*nb_aatx+atx], avg_rho_im[aatx*nb_aatx+atx]);
}
}
/*if (first_symbol_flag==1) {
* //measurements->rx_correlation[0][0][aarx] = signal_energy(&rho[aarx][symbol*nb_rb*12],rb*12);
* avg_rho_re[aatx][aarx] = 16*avg_rho_re[aatx][aarx]/(nb_rb*12);
* avg_rho_im[aatx][aarx] = 16*avg_rho_im[aatx][aarx]/(nb_rb*12);
* printf("rho_rx%d tx%d= Re: %d Im: %d\n",aarx, aatx, avg_rho_re[aatx][aarx], avg_rho_im[aatx][aarx]);
* }*/
}
}
}
}
}
_mm_empty();
_m_empty();
@@ -1292,7 +1294,7 @@ void nr_dlsch_channel_compensation_core(int **rxdataF_ext,
int **dl_ch_mag,
int **dl_ch_magb,
int **rxdataF_comp,
int **rho,
int ***rho,
unsigned char n_tx,
unsigned char n_rx,
unsigned char mod_order,
@@ -1429,7 +1431,7 @@ void nr_dlsch_channel_compensation_core(int **rxdataF_ext,
for (aarx=0; aarx<n_rx; aarx++) {
rho128 = (__m128i *)&rho[aarx][start_point];
rho128 = (__m128i *)&rho[aarx][0][start_point];
dl_ch128 = (__m128i *)&dl_ch_estimates_ext[aarx][start_point];
dl_ch128_2 = (__m128i *)&dl_ch_estimates_ext[2+aarx][start_point];
openair1/PHY/NR_UE_TRANSPORT/nr_transport_proto_ue.h
+2 -2
View File
@@ -803,7 +803,7 @@ void nr_dlsch_channel_compensation(int32_t **rxdataF_ext,
int32_t **dl_ch_magb,
int32_t **dl_ch_magr,
int32_t **rxdataF_comp,
int32_t **rho,
int32_t ***rho,
NR_DL_FRAME_PARMS *frame_parms,
uint8_t nb_aatx,
uint8_t symbol,
@@ -819,7 +819,7 @@ void nr_dlsch_channel_compensation_core(int **rxdataF_ext,
int **dl_ch_mag,
int **dl_ch_magb,
int **rxdataF_comp,
int **rho,
int ***rho,
unsigned char n_tx,
unsigned char n_rx,
unsigned char mod_order,