You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

834 lines
21 KiB
C

/**
*
* \section COPYRIGHT
*
* Copyright 2013-2021 Software Radio Systems Limited
*
* By using this file, you agree to the terms and conditions set
* forth in the LICENSE file which can be found at the top level of
* the distribution.
*
*/
#include <complex.h>
#include <float.h>
#include <math.h>
#include <stdlib.h>
#include <string.h>
#include "srsran/phy/utils/bit.h"
#include "srsran/phy/utils/debug.h"
#include "srsran/phy/utils/simd.h"
#include "srsran/phy/utils/vector.h"
#include "srsran/phy/utils/vector_simd.h"
void srsran_vec_xor_bbb(const uint8_t* x, const uint8_t* y, uint8_t* z, const uint32_t len)
{
srsran_vec_xor_bbb_simd(x, y, z, len);
}
// Used in PRACH detector, AGC and chest_dl for noise averaging
float srsran_vec_acc_ff(const float* x, const uint32_t len)
{
return srsran_vec_acc_ff_simd(x, len);
}
cf_t srsran_vec_acc_cc(const cf_t* x, const uint32_t len)
{
return srsran_vec_acc_cc_simd(x, len);
}
void srsran_vec_sub_fff(const float* x, const float* y, float* z, const uint32_t len)
{
srsran_vec_sub_fff_simd(x, y, z, len);
}
void srsran_vec_sub_sss(const int16_t* x, const int16_t* y, int16_t* z, const uint32_t len)
{
srsran_vec_sub_sss_simd(x, y, z, len);
}
void srsran_vec_sub_bbb(const int8_t* x, const int8_t* y, int8_t* z, const uint32_t len)
{
srsran_vec_sub_bbb_simd(x, y, z, len);
}
// Noise estimation in chest_dl, interpolation
void srsran_vec_sub_ccc(const cf_t* x, const cf_t* y, cf_t* z, const uint32_t len)
{
return srsran_vec_sub_fff((const float*)x, (const float*)y, (float*)z, 2 * len);
}
// Used in PSS/SSS and sum_ccc
void srsran_vec_sum_fff(const float* x, const float* y, float* z, const uint32_t len)
{
srsran_vec_add_fff_simd(x, y, z, len);
}
void srsran_vec_sum_sss(const int16_t* x, const int16_t* y, int16_t* z, const uint32_t len)
{
srsran_vec_sum_sss_simd(x, y, z, len);
}
void srsran_vec_sum_ccc(const cf_t* x, const cf_t* y, cf_t* z, const uint32_t len)
{
srsran_vec_sum_fff((float*)x, (float*)y, (float*)z, 2 * len);
}
// PSS, PBCH, DEMOD, FFTW, etc.
void srsran_vec_sc_prod_fff(const float* x, const float h, float* z, const uint32_t len)
{
srsran_vec_sc_prod_fff_simd(x, h, z, len);
}
// Used throughout
void srsran_vec_sc_prod_cfc(const cf_t* x, const float h, cf_t* z, const uint32_t len)
{
srsran_vec_sc_prod_cfc_simd(x, h, z, len);
}
void srsran_vec_sc_prod_fcc(const float* x, const cf_t h, cf_t* z, const uint32_t len)
{
srsran_vec_sc_prod_fcc_simd(x, h, z, len);
}
// Chest UL
void srsran_vec_sc_prod_ccc(const cf_t* x, const cf_t h, cf_t* z, const uint32_t len)
{
srsran_vec_sc_prod_ccc_simd(x, h, z, len);
}
// Used in turbo decoder
void srsran_vec_convert_if(const int16_t* x, const float scale, float* z, const uint32_t len)
{
srsran_vec_convert_if_simd(x, z, scale, len);
}
void srsran_vec_convert_fi(const float* x, const float scale, int16_t* z, const uint32_t len)
{
srsran_vec_convert_fi_simd(x, z, scale, len);
}
void srsran_vec_convert_conj_cs(const cf_t* x, const float scale, int16_t* z, const uint32_t len)
{
srsran_vec_convert_conj_cs_simd(x, z, scale, len);
}
void srsran_vec_convert_fb(const float* x, const float scale, int8_t* z, const uint32_t len)
{
srsran_vec_convert_fb_simd(x, z, scale, len);
}
void srsran_vec_lut_sss(const short* x, const unsigned short* lut, short* y, const uint32_t len)
{
srsran_vec_lut_sss_simd(x, lut, y, len);
}
void srsran_vec_lut_bbb(const int8_t* x, const unsigned short* lut, int8_t* y, const uint32_t len)
{
srsran_vec_lut_bbb_simd(x, lut, y, len);
}
void srsran_vec_lut_sis(const short* x, const unsigned int* lut, short* y, const uint32_t len)
{
for (int i = 0; i < len; i++) {
y[lut[i]] = x[i];
}
}
void* srsran_vec_malloc(uint32_t size)
{
void* ptr;
if (posix_memalign(&ptr, SRSRAN_SIMD_BIT_ALIGN, size)) {
return NULL;
} else {
return ptr;
}
}
cf_t* srsran_vec_cf_malloc(uint32_t nsamples)
{
return SRSRAN_MEM_ALLOC(cf_t, nsamples);
}
float* srsran_vec_f_malloc(uint32_t nsamples)
{
return SRSRAN_MEM_ALLOC(float, nsamples);
}
int32_t* srsran_vec_i32_malloc(uint32_t nsamples)
{
return SRSRAN_MEM_ALLOC(int32_t, nsamples);
}
uint32_t* srsran_vec_u32_malloc(uint32_t nsamples)
{
return SRSRAN_MEM_ALLOC(uint32_t, nsamples);
}
int16_t* srsran_vec_i16_malloc(uint32_t nsamples)
{
return SRSRAN_MEM_ALLOC(int16_t, nsamples);
}
uint16_t* srsran_vec_u16_malloc(uint32_t nsamples)
{
return SRSRAN_MEM_ALLOC(uint16_t, nsamples);
}
int8_t* srsran_vec_i8_malloc(uint32_t nsamples)
{
return SRSRAN_MEM_ALLOC(int8_t, nsamples);
}
uint8_t* srsran_vec_u8_malloc(uint32_t nsamples)
{
return SRSRAN_MEM_ALLOC(uint8_t, nsamples);
}
void srsran_vec_zero(void* ptr, uint32_t nbytes)
{
memset(ptr, 0, nbytes);
}
void srsran_vec_u8_zero(uint8_t* ptr, uint32_t nsamples)
{
SRSRAN_MEM_ZERO(ptr, uint8_t, nsamples);
}
void srsran_vec_i8_zero(int8_t* ptr, uint32_t nsamples)
{
SRSRAN_MEM_ZERO(ptr, int8_t, nsamples);
}
void srsran_vec_i16_zero(int16_t* ptr, uint32_t nsamples)
{
SRSRAN_MEM_ZERO(ptr, int16_t, nsamples);
}
void srsran_vec_u32_zero(uint32_t* ptr, uint32_t nsamples)
{
SRSRAN_MEM_ZERO(ptr, uint32_t, nsamples);
}
void srsran_vec_cf_zero(cf_t* ptr, uint32_t nsamples)
{
SRSRAN_MEM_ZERO(ptr, cf_t, nsamples);
}
void srsran_vec_f_zero(float* ptr, uint32_t nsamples)
{
SRSRAN_MEM_ZERO(ptr, float, nsamples);
}
void srsran_vec_cf_copy(cf_t* dst, const cf_t* src, uint32_t len)
{
memcpy(dst, src, sizeof(cf_t) * len);
}
void srsran_vec_f_copy(float* dst, const float* src, uint32_t len)
{
memcpy(dst, src, sizeof(float) * len);
}
void srsran_vec_u8_copy(uint8_t* dst, const uint8_t* src, uint32_t len)
{
memcpy(dst, src, sizeof(uint8_t) * len);
}
void srsran_vec_i8_copy(int8_t* dst, const int8_t* src, uint32_t len)
{
memcpy(dst, src, sizeof(int8_t) * len);
}
void srsran_vec_i16_copy(int16_t* dst, const int16_t* src, uint32_t len)
{
memcpy(dst, src, sizeof(int16_t) * len);
}
void srsran_vec_u16_copy(uint16_t* dst, const uint16_t* src, uint32_t len)
{
memcpy(dst, src, sizeof(uint16_t) * len);
}
void* srsran_vec_realloc(void* ptr, uint32_t old_size, uint32_t new_size)
{
#ifndef LV_HAVE_SSE
return realloc(ptr, new_size);
#else
void* new_ptr;
if (posix_memalign(&new_ptr, SRSRAN_SIMD_BIT_ALIGN, new_size)) {
return NULL;
} else {
memcpy(new_ptr, ptr, old_size);
free(ptr);
return new_ptr;
}
#endif
}
void srsran_vec_fprint_c(FILE* stream, const cf_t* x, const uint32_t len)
{
int i;
fprintf(stream, "[");
for (i = 0; i < len; i++) {
fprintf(stream, "%+2.5f%+2.5fi, ", __real__ x[i], __imag__ x[i]);
}
fprintf(stream, "];\n");
}
void srsran_vec_fprint_f(FILE* stream, const float* x, const uint32_t len)
{
int i;
fprintf(stream, "[");
for (i = 0; i < len; i++) {
fprintf(stream, "%+2.2f, ", x[i]);
}
fprintf(stream, "];\n");
}
void srsran_vec_fprint_b(FILE* stream, const uint8_t* x, const uint32_t len)
{
int i;
fprintf(stream, "[");
for (i = 0; i < len; i++) {
fprintf(stream, "%d, ", x[i]);
}
fprintf(stream, "];\n");
}
void srsran_vec_fprint_bs(FILE* stream, const int8_t* x, const uint32_t len)
{
int i;
fprintf(stream, "[");
for (i = 0; i < len; i++) {
fprintf(stream, "%4d, ", x[i]);
}
fprintf(stream, "];\n");
}
void srsran_vec_fprint_byte(FILE* stream, const uint8_t* x, const uint32_t len)
{
int i;
fprintf(stream, "[");
for (i = 0; i < len; i++) {
fprintf(stream, "%02x ", x[i]);
}
fprintf(stream, "];\n");
}
void srsran_vec_fprint_i(FILE* stream, const int* x, const uint32_t len)
{
int i;
fprintf(stream, "[");
for (i = 0; i < len; i++) {
fprintf(stream, "%d, ", x[i]);
}
fprintf(stream, "];\n");
}
void srsran_vec_fprint_s(FILE* stream, const int16_t* x, const uint32_t len)
{
int i;
fprintf(stream, "[");
for (i = 0; i < len; i++) {
fprintf(stream, "%4d, ", x[i]);
}
fprintf(stream, "];\n");
}
void srsran_vec_fprint_hex(FILE* stream, uint8_t* x, const uint32_t len)
{
uint32_t i, nbytes;
uint8_t byte;
nbytes = len / 8;
fprintf(stream, "[");
for (i = 0; i < nbytes; i++) {
byte = (uint8_t)srsran_bit_pack(&x, 8);
fprintf(stream, "%02x ", byte);
}
if (len % 8) {
byte = (uint8_t)srsran_bit_pack(&x, len % 8) << (8 - (len % 8));
fprintf(stream, "%02x ", byte);
}
fprintf(stream, "];\n");
}
uint32_t srsran_vec_sprint_hex(char* str, const uint32_t max_str_len, uint8_t* x, const uint32_t len)
{
uint32_t i, nbytes;
uint8_t byte;
nbytes = len / 8;
// check that hex string fits in buffer (every byte takes 3 characters, plus brackets)
if ((3 * (len / 8 + ((len % 8) ? 1 : 0))) + 2 >= max_str_len) {
ERROR("Buffer too small for printing hex string (max_str_len=%d, payload_len=%d).", max_str_len, len);
return 0;
}
int n = 0;
n += sprintf(&str[n], "[");
for (i = 0; i < nbytes; i++) {
byte = (uint8_t)srsran_bit_pack(&x, 8);
n += sprintf(&str[n], "%02x ", byte);
}
if (len % 8) {
byte = (uint8_t)srsran_bit_pack(&x, len % 8) << (8 - (len % 8));
n += sprintf(&str[n], "%02x ", byte);
}
n += sprintf(&str[n], "]");
str[n] = 0;
str[max_str_len - 1] = 0;
return n;
}
void srsran_vec_sprint_bin(char* str, const uint32_t max_str_len, const uint8_t* x, const uint32_t len)
{
// Trim maximum size
uint32_t N = SRSRAN_MIN(max_str_len - 1, len);
// If the number of bits does not fit in the string, leave space for "..." if possible
if (N < len) {
if (N >= 3) {
N -= 3;
} else {
N = 0;
}
}
// Write 1s and 0s
for (uint32_t i = 0; i < N; i++) {
str[i] = x[i] == 0 ? '0' : '1';
}
// Write "..." if all 1s and 0s did not fit
if (N < len) {
for (uint32_t i = N; i < max_str_len - 1; i++) {
str[i] = '.';
}
str[max_str_len - 1] = 0;
} else {
str[N] = 0;
}
}
void srsran_vec_save_file(char* filename, const void* buffer, const uint32_t len)
{
FILE* f;
f = fopen(filename, "w");
if (f) {
fwrite(buffer, len, 1, f);
fclose(f);
} else {
perror("fopen");
}
}
#define SAFE_READ(PTR, SIZE, N, FILE) \
do { \
size_t nbytes = SIZE * N; \
if (nbytes != fread(PTR, SIZE, N, FILE)) { \
perror("read"); \
fclose(FILE); \
exit(1); \
} \
} while (false)
void srsran_vec_load_file(char* filename, void* buffer, const uint32_t len)
{
FILE* f;
f = fopen(filename, "r");
if (f) {
SAFE_READ(buffer, len, 1, f);
fclose(f);
} else {
perror("fopen");
}
}
// Used in PSS
void srsran_vec_conj_cc(const cf_t* x, cf_t* y, const uint32_t len)
{
/* This function is used in initialisation only, then no optimisation is required */
int i;
for (i = 0; i < len; i++) {
y[i] = conjf(x[i]);
}
}
// Used in scrambling complex
void srsran_vec_prod_cfc(const cf_t* x, const float* y, cf_t* z, const uint32_t len)
{
srsran_vec_prod_cfc_simd(x, y, z, len);
}
// Used in scrambling float
void srsran_vec_prod_fff(const float* x, const float* y, float* z, const uint32_t len)
{
srsran_vec_prod_fff_simd(x, y, z, len);
}
void srsran_vec_prod_sss(const int16_t* x, const int16_t* y, int16_t* z, const uint32_t len)
{
srsran_vec_prod_sss_simd(x, y, z, len);
}
// Scrambling
void srsran_vec_neg_sss(const int16_t* x, const int16_t* y, int16_t* z, const uint32_t len)
{
srsran_vec_neg_sss_simd(x, y, z, len);
}
void srsran_vec_neg_bbb(const int8_t* x, const int8_t* y, int8_t* z, const uint32_t len)
{
srsran_vec_neg_bbb_simd(x, y, z, len);
}
void srsran_vec_neg_bb(const int8_t* x, int8_t* z, const uint32_t len)
{
for (uint32_t i = 0; i < len; i++) {
z[i] = -x[i];
}
}
// CFO and OFDM processing
void srsran_vec_prod_ccc(const cf_t* x, const cf_t* y, cf_t* z, const uint32_t len)
{
srsran_vec_prod_ccc_simd(x, y, z, len);
}
void srsran_vec_prod_ccc_split(const float* x_re,
const float* x_im,
const float* y_re,
const float* y_im,
float* z_re,
float* z_im,
const uint32_t len)
{
srsran_vec_prod_ccc_split_simd(x_re, x_im, y_re, y_im, z_re, z_im, len);
}
// PRACH, CHEST UL, etc.
void srsran_vec_prod_conj_ccc(const cf_t* x, const cf_t* y, cf_t* z, const uint32_t len)
{
srsran_vec_prod_conj_ccc_simd(x, y, z, len);
}
//#define DIV_USE_VEC
// Used in SSS
void srsran_vec_div_ccc(const cf_t* x, const cf_t* y, cf_t* z, const uint32_t len)
{
srsran_vec_div_ccc_simd(x, y, z, len);
}
/* Complex division by float z=x/y */
void srsran_vec_div_cfc(const cf_t* x, const float* y, cf_t* z, const uint32_t len)
{
srsran_vec_div_cfc_simd(x, y, z, len);
}
void srsran_vec_div_fff(const float* x, const float* y, float* z, const uint32_t len)
{
srsran_vec_div_fff_simd(x, y, z, len);
}
// PSS. convolution
cf_t srsran_vec_dot_prod_ccc(const cf_t* x, const cf_t* y, const uint32_t len)
{
return srsran_vec_dot_prod_ccc_simd(x, y, len);
}
// Convolution filter and in SSS search
cf_t srsran_vec_dot_prod_cfc(const cf_t* x, const float* y, const uint32_t len)
{
uint32_t i;
cf_t res = 0;
for (i = 0; i < len; i++) {
res += x[i] * y[i];
}
return res;
}
// SYNC
cf_t srsran_vec_dot_prod_conj_ccc(const cf_t* x, const cf_t* y, const uint32_t len)
{
return srsran_vec_dot_prod_conj_ccc_simd(x, y, len);
}
// PHICH
float srsran_vec_dot_prod_fff(const float* x, const float* y, const uint32_t len)
{
uint32_t i;
float res = 0;
for (i = 0; i < len; i++) {
res += x[i] * y[i];
}
return res;
}
int32_t srsran_vec_dot_prod_sss(const int16_t* x, const int16_t* y, const uint32_t len)
{
return srsran_vec_dot_prod_sss_simd(x, y, len);
}
float srsran_vec_avg_power_cf(const cf_t* x, const uint32_t len)
{
return crealf(srsran_vec_dot_prod_conj_ccc(x, x, len)) / len;
}
float srsran_vec_avg_power_sf(const int16_t* x, const uint32_t len)
{
// Accumulator
float acc = 0.0f;
for (uint32_t i = 0; i < len; i++) {
// Read value and typecast to float
float t = (float)x[i];
// Square value
acc += t * t;
}
// Do average
if (len) {
acc /= len;
}
// Return accumulated value
return acc;
}
float srsran_vec_avg_power_bf(const int8_t* x, const uint32_t len)
{
// Accumulator
float acc = 0.0f;
for (uint32_t i = 0; i < len; i++) {
// Read value and typecast to float
float t = (float)x[i];
// Square value
acc += t * t;
}
// Do average
if (len) {
acc /= len;
}
// Return accumulated value
return acc;
}
// Correlation assumes zero-mean x and y
float srsran_vec_corr_ccc(const cf_t* x, cf_t* y, const uint32_t len)
{
// return crealf(srsran_vec_dot_prod_conj_ccc(x,y,len)) / len;
float s_x = crealf(srsran_vec_dot_prod_conj_ccc(x, x, len)) / len;
float s_y = crealf(srsran_vec_dot_prod_conj_ccc(y, y, len)) / len;
float cov = crealf(srsran_vec_dot_prod_conj_ccc(x, y, len)) / len;
return cov / (sqrtf(s_x * s_y));
}
// PSS (disabled and using abs_square )
void srsran_vec_abs_cf(const cf_t* x, float* abs, const uint32_t len)
{
srsran_vec_abs_cf_simd(x, abs, len);
}
void srsran_vec_abs_dB_cf(const cf_t* x, float default_value, float* abs, const uint32_t len)
{
// Convert complex input to absplute value
srsran_vec_abs_cf(x, abs, len);
// Convert absolute value to dB
for (int i = 0; i < len; i++) {
// Check boundaries
if (isnormal(abs[i])) {
// Avoid infinites and zeros
abs[i] = srsran_convert_amplitude_to_dB(abs[i]);
} else {
// Set to default value instead
abs[i] = default_value;
}
}
}
void srsran_vec_arg_deg_cf(const cf_t* x, float default_value, float* arg, const uint32_t len)
{
for (int i = 0; i < len; i++) {
// Convert complex value to argument in degrees
arg[i] = cargf(x[i]) * (180.0f / M_PI);
// Check boundaries
if (arg[i] != 0.0f && !isnormal(arg[i])) {
// different than zero and not normal
arg[i] = default_value;
}
}
}
// PRACH
void srsran_vec_abs_square_cf(const cf_t* x, float* abs_square, const uint32_t len)
{
srsran_vec_abs_square_cf_simd(x, abs_square, len);
}
uint32_t srsran_vec_max_fi(const float* x, const uint32_t len)
{
return srsran_vec_max_fi_simd(x, len);
}
uint32_t srsran_vec_max_abs_fi(const float* x, const uint32_t len)
{
return srsran_vec_max_abs_fi_simd(x, len);
}
// CP autocorr
uint32_t srsran_vec_max_abs_ci(const cf_t* x, const uint32_t len)
{
return srsran_vec_max_ci_simd(x, len);
}
void srsran_vec_quant_fs(const float* in,
int16_t* out,
const float gain,
const float offset,
const float clip,
const uint32_t len)
{
int i = 0;
long tmp = 0;
const int16_t inf = (1U << 15U) - 1;
for (i = 0; i < len; i++) {
if (isinf(in[i])) {
tmp = inf * (-2 * (in[i] < 0) + 1);
} else {
tmp = (long)(offset + gain * in[i] + INT16_MAX + 0.5) - INT16_MAX;
if (tmp < -clip) {
tmp = -clip;
}
if (tmp > clip) {
tmp = clip;
}
}
out[i] = (int16_t)tmp;
}
}
void srsran_vec_quant_fc(const float* in,
int8_t* out,
const float gain,
const float offset,
const float clip,
const uint32_t len)
{
int i = 0;
long tmp = 0;
for (i = 0; i < len; i++) {
if (isinf(in[i])) {
tmp = 127 * (-2 * (in[i] < 0) + 1);
} else {
tmp = (long)(offset + gain * in[i] + INT8_MAX + 0.5) - INT8_MAX;
if (tmp < -clip) {
tmp = -clip;
}
if (tmp > clip) {
tmp = clip;
}
}
out[i] = (int8_t)tmp;
}
}
void srsran_vec_quant_fus(const float* in,
uint16_t* out,
const float gain,
const float offset,
const uint16_t clip,
const uint32_t len)
{
for (uint32_t i = 0; i < len; i++) {
int32_t tmp = (int32_t)(offset + gain * in[i]);
tmp = SRSRAN_MAX(tmp, 0);
tmp = SRSRAN_MIN(tmp, (int32_t)clip);
out[i] = (uint16_t)tmp;
}
}
void srsran_vec_quant_fuc(const float* in,
uint8_t* out,
const float gain,
const float offset,
const uint8_t clip,
const uint32_t len)
{
for (uint32_t i = 0; i < len; i++) {
int32_t tmp = (int32_t)(offset + gain * in[i]);
tmp = SRSRAN_MAX(tmp, 0);
tmp = SRSRAN_MIN(tmp, (int32_t)clip);
out[i] = (uint8_t)tmp;
}
}
void srsran_vec_quant_suc(const int16_t* in,
uint8_t* out,
const float gain,
const float offset,
const uint8_t clip,
const uint32_t len)
{
for (uint32_t i = 0; i < len; i++) {
int32_t tmp = (int32_t)(offset + (float)in[i] * gain);
tmp = SRSRAN_MAX(tmp, 0);
tmp = SRSRAN_MIN(tmp, (int32_t)clip);
out[i] = (uint8_t)tmp;
}
}
void srsran_vec_quant_sus(const int16_t* in,
uint16_t* out,
const float gain,
const float offset,
const uint16_t clip,
const uint32_t len)
{
for (uint32_t i = 0; i < len; i++) {
int32_t tmp = (int32_t)(offset + gain * (float)in[i]);
tmp = SRSRAN_MAX(tmp, 0);
tmp = SRSRAN_MIN(tmp, (int32_t)clip);
out[i] = (uint16_t)tmp;
}
}
void srsran_vec_interleave(const cf_t* x, const cf_t* y, cf_t* z, const int len)
{
srsran_vec_interleave_simd(x, y, z, len);
}
void srsran_vec_interleave_add(const cf_t* x, const cf_t* y, cf_t* z, const int len)
{
srsran_vec_interleave_add_simd(x, y, z, len);
}
void srsran_vec_gen_sine(cf_t amplitude, float freq, cf_t* z, int len)
{
srsran_vec_gen_sine_simd(amplitude, freq, z, len);
}
void srsran_vec_apply_cfo(const cf_t* x, float cfo, cf_t* z, int len)
{
srsran_vec_apply_cfo_simd(x, cfo, z, len);
}
float srsran_vec_estimate_frequency(const cf_t* x, int len)
{
return srsran_vec_estimate_frequency_simd(x, len);
}