354 lines
9.4 KiB
C
354 lines
9.4 KiB
C
/* Adapted Phil Karn's r=1/3 k=9 viterbi decoder to r=1/3 k=7
|
|
*
|
|
* K=15 r=1/6 Viterbi decoder for x86 SSE2
|
|
* Copyright Mar 2004, Phil Karn, KA9Q
|
|
* May be used under the terms of the GNU Lesser General Public License (LGPL)
|
|
*/
|
|
|
|
#include "parity.h"
|
|
#include <limits.h>
|
|
#include <memory.h>
|
|
#include <stdint.h>
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
|
|
//#define DEBUG
|
|
|
|
#ifdef LV_HAVE_AVX2
|
|
|
|
#include <emmintrin.h>
|
|
#include <immintrin.h>
|
|
#include <tmmintrin.h>
|
|
|
|
typedef union {
|
|
// unsigned char c[64];
|
|
//__m128i v[4];
|
|
unsigned short c[64];
|
|
__m256i v[4];
|
|
} metric_t;
|
|
|
|
typedef union {
|
|
unsigned int w[2];
|
|
unsigned char c[8];
|
|
unsigned short s[4];
|
|
__m64 v[1];
|
|
} decision_t;
|
|
|
|
union branchtab27 {
|
|
|
|
// unsigned char c[32];
|
|
//__m128i v[2];
|
|
unsigned short c[32];
|
|
__m256i v[2];
|
|
|
|
} Branchtab37_sse2[3];
|
|
|
|
int firstGo;
|
|
/* State info for instance of Viterbi decoder */
|
|
struct v37 {
|
|
metric_t metrics1; /* path metric buffer 1 */
|
|
metric_t metrics2; /* path metric buffer 2 */
|
|
decision_t* dp; /* Pointer to current decision */
|
|
metric_t * old_metrics, *new_metrics; /* Pointers to path metrics, swapped on every bit */
|
|
decision_t* decisions; /* Beginning of decisions for block */
|
|
uint32_t len;
|
|
};
|
|
|
|
void set_viterbi37_polynomial_avx2_16bit(int polys[3])
|
|
{
|
|
int state;
|
|
for (state = 0; state < 32; state++) {
|
|
Branchtab37_sse2[0].c[state] = (polys[0] < 0) ^ parity((2 * state) & polys[0]) ? 65535 : 0;
|
|
Branchtab37_sse2[1].c[state] = (polys[1] < 0) ^ parity((2 * state) & polys[1]) ? 65535 : 0;
|
|
Branchtab37_sse2[2].c[state] = (polys[2] < 0) ^ parity((2 * state) & polys[2]) ? 65535 : 0;
|
|
}
|
|
}
|
|
|
|
void clear_v37_avx2_16bit(struct v37* vp)
|
|
{
|
|
bzero(vp->decisions, sizeof(decision_t) * vp->len);
|
|
vp->dp = NULL;
|
|
bzero(&vp->metrics1, sizeof(metric_t));
|
|
bzero(&vp->metrics2, sizeof(metric_t));
|
|
vp->old_metrics = NULL;
|
|
vp->new_metrics = NULL;
|
|
}
|
|
|
|
/* Initialize Viterbi decoder for start of new frame */
|
|
int init_viterbi37_avx2_16bit(void* p, int starting_state)
|
|
{
|
|
|
|
struct v37* vp = p;
|
|
uint32_t i;
|
|
|
|
for (i = 0; i < 64; i++)
|
|
vp->metrics1.c[i] = 63;
|
|
|
|
clear_v37_avx2_16bit(vp);
|
|
firstGo = 1;
|
|
vp->old_metrics = &vp->metrics1;
|
|
vp->new_metrics = &vp->metrics2;
|
|
vp->dp = vp->decisions;
|
|
if (starting_state != -1) {
|
|
vp->old_metrics->c[starting_state & 63] = 0; /* Bias known start state */
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
/* Create a new instance of a Viterbi decoder */
|
|
void* create_viterbi37_avx2_16bit(int polys[3], uint32_t len)
|
|
{
|
|
void* p;
|
|
struct v37* vp;
|
|
|
|
set_viterbi37_polynomial_avx2_16bit(polys);
|
|
|
|
/* Ordinary malloc() only returns 8-byte alignment, we need 16 */
|
|
if (posix_memalign(&p, sizeof(__m256i), sizeof(struct v37)))
|
|
return NULL;
|
|
|
|
vp = (struct v37*)p;
|
|
if (posix_memalign(&p, sizeof(__m256i), (len + 6) * sizeof(decision_t))) {
|
|
free(vp);
|
|
return NULL;
|
|
}
|
|
vp->decisions = (decision_t*)p;
|
|
vp->len = len + 6;
|
|
return vp;
|
|
}
|
|
|
|
/* Viterbi chainback */
|
|
int chainback_viterbi37_avx2_16bit(void* p,
|
|
uint8_t* data, /* Decoded output data */
|
|
uint32_t nbits, /* Number of data bits */
|
|
uint32_t endstate)
|
|
{ /* Terminal encoder state */
|
|
struct v37* vp = p;
|
|
|
|
if (p == NULL)
|
|
return -1;
|
|
|
|
decision_t* d = (decision_t*)vp->decisions;
|
|
|
|
/* Make room beyond the end of the encoder register so we can
|
|
* accumulate a full byte of decoded data
|
|
*/
|
|
endstate %= 64;
|
|
endstate <<= 2;
|
|
|
|
/* The store into data[] only needs to be done every 8 bits.
|
|
* But this avoids a conditional branch, and the writes will
|
|
* combine in the cache anyway
|
|
*/
|
|
d += 6; /* Look past tail */
|
|
while (nbits--) {
|
|
int k;
|
|
|
|
k = (d[nbits].c[(endstate >> 2) / 8] >> ((endstate >> 2) % 8)) & 1;
|
|
endstate = (endstate >> 1) | (k << 7);
|
|
data[nbits] = k;
|
|
// printf("nbits=%d, endstate=%3d, k=%d, w[0]=%d, w[1]=%d, c=%d\n", nbits, endstate, k, d[nbits].s[1]&1,
|
|
// d[nbits].s[2]&1, d[nbits].c[(endstate>>2)/8]&1);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
/* Delete instance of a Viterbi decoder */
|
|
void delete_viterbi37_avx2_16bit(void* p)
|
|
{
|
|
struct v37* vp = p;
|
|
|
|
if (vp != NULL) {
|
|
free(vp->decisions);
|
|
free(vp);
|
|
}
|
|
}
|
|
|
|
void print_256i(char* s, __m256i val)
|
|
{
|
|
|
|
printf("%s: ", s);
|
|
|
|
uint16_t* x = (uint16_t*)&val;
|
|
for (int i = 0; i < 16; i++) {
|
|
printf("%.5f, ", (float)x[i] / 65535);
|
|
}
|
|
printf("\n");
|
|
}
|
|
void print_256i_char(char* s, __m256i val)
|
|
{
|
|
|
|
printf("%s: ", s);
|
|
|
|
uint8_t* x = (uint8_t*)&val;
|
|
for (int i = 0; i < 32; i++) {
|
|
printf("%d, ", x[31 - i]);
|
|
}
|
|
printf("\n");
|
|
}
|
|
|
|
inline unsigned short my_mm256_movemask_epi16(__m256i x)
|
|
{
|
|
uint32_t x1 = _mm256_movemask_epi8(x);
|
|
uint16_t tmp = 0;
|
|
for (int i = 0; i < 16; i++) {
|
|
tmp |= ((x1 >> ((i * 2) + 1)) & 0x01) << i;
|
|
}
|
|
|
|
return (tmp);
|
|
}
|
|
|
|
void update_viterbi37_blk_avx2_16bit(void* p, unsigned short* syms, int nbits, uint32_t* best_state)
|
|
{
|
|
struct v37* vp = p;
|
|
decision_t* d;
|
|
|
|
if (p == NULL)
|
|
return;
|
|
|
|
#ifdef DEBUG
|
|
printf("[");
|
|
#endif
|
|
|
|
d = (decision_t*)vp->dp;
|
|
|
|
for (int s = 0; s < nbits; s++) {
|
|
memset(d + s, 0, sizeof(decision_t));
|
|
}
|
|
|
|
while (nbits--) {
|
|
__m256i sym0v, sym1v, sym2v;
|
|
void* tmp;
|
|
int i;
|
|
|
|
// printf("nbits=%d, syms=%d,%d,%d\n", nbits, syms[0], syms[1], syms[2]);fflush(stdout);
|
|
|
|
/* Splat the 0th symbol across sym0v, the 1st symbol across sym1v, etc */
|
|
|
|
sym0v = _mm256_set1_epi16(syms[0]);
|
|
sym1v = _mm256_set1_epi16(syms[1]);
|
|
sym2v = _mm256_set1_epi16(syms[2]);
|
|
|
|
syms += 3;
|
|
|
|
for (i = 0; i < 2; i++) {
|
|
|
|
__m256i decision0, decision1, metric, m_metric, m0, m1, m2, m3, survivor0, survivor1;
|
|
|
|
/* Form branch metrics */
|
|
m0 = _mm256_avg_epu16(_mm256_xor_si256(Branchtab37_sse2[0].v[i], sym0v),
|
|
_mm256_xor_si256(Branchtab37_sse2[1].v[i], sym1v));
|
|
metric = _mm256_avg_epu16(_mm256_xor_si256(Branchtab37_sse2[2].v[i], sym2v), m0);
|
|
|
|
#ifdef DEBUG
|
|
print_128i("metric_initial", metric);
|
|
#endif
|
|
/* There's no packed bytes right shift in SSE2, so we use the word version and mask
|
|
*/
|
|
|
|
metric = _mm256_srli_epi16(metric, 3);
|
|
m_metric = _mm256_sub_epi16(_mm256_set1_epi16(8191), metric);
|
|
|
|
#ifdef DEBUG
|
|
print_128i("metric ", metric);
|
|
print_128i("m_metric ", m_metric);
|
|
#endif
|
|
|
|
/* Add branch metrics to path metrics */
|
|
|
|
m0 = _mm256_add_epi16(vp->old_metrics->v[i], metric);
|
|
m3 = _mm256_add_epi16(vp->old_metrics->v[2 + i], metric);
|
|
m1 = _mm256_add_epi16(vp->old_metrics->v[2 + i], m_metric);
|
|
m2 = _mm256_add_epi16(vp->old_metrics->v[i], m_metric);
|
|
|
|
/* Compare and select, using modulo arithmetic */
|
|
|
|
decision0 = _mm256_cmpgt_epi16(_mm256_sub_epi16(m0, m1), _mm256_setzero_si256());
|
|
decision1 = _mm256_cmpgt_epi16(_mm256_sub_epi16(m2, m3), _mm256_setzero_si256());
|
|
survivor0 = _mm256_or_si256(_mm256_and_si256(decision0, m1), _mm256_andnot_si256(decision0, m0));
|
|
survivor1 = _mm256_or_si256(_mm256_and_si256(decision1, m3), _mm256_andnot_si256(decision1, m2));
|
|
|
|
/* Pack each set of decisions into 16 bits */
|
|
|
|
decision0 = _mm256_permute4x64_epi64(decision0, 216);
|
|
decision1 = _mm256_permute4x64_epi64(decision1, 216);
|
|
|
|
__m256i packed = _mm256_packus_epi16(_mm256_srli_epi16(_mm256_unpacklo_epi16(decision0, decision1), 8),
|
|
_mm256_srli_epi16(_mm256_unpackhi_epi16(decision0, decision1), 8));
|
|
|
|
d->w[i] = _mm256_movemask_epi8(packed);
|
|
|
|
unsigned char temp_char1 = d->c[4 * i + 1];
|
|
unsigned char temp_char2 = d->c[4 * i + 2];
|
|
|
|
d->c[4 * i + 1] = temp_char2;
|
|
d->c[4 * i + 2] = temp_char1;
|
|
|
|
/* Store surviving metrics */
|
|
survivor0 = _mm256_permute4x64_epi64(survivor0, 216);
|
|
survivor1 = _mm256_permute4x64_epi64(survivor1, 216);
|
|
|
|
vp->new_metrics->v[2 * i] = _mm256_unpacklo_epi16(survivor0, survivor1);
|
|
vp->new_metrics->v[2 * i + 1] = _mm256_unpackhi_epi16(survivor0, survivor1);
|
|
}
|
|
|
|
// See if we need to normalize
|
|
if (vp->new_metrics->c[0] > 12288) {
|
|
int i;
|
|
|
|
uint16_t adjust;
|
|
__m256i adjustv;
|
|
union {
|
|
__m256i v;
|
|
signed short w[8];
|
|
} t;
|
|
|
|
adjustv = vp->new_metrics->v[0];
|
|
for (i = 1; i < 4; i++) {
|
|
adjustv = _mm256_min_epu16(adjustv, vp->new_metrics->v[i]);
|
|
}
|
|
|
|
adjustv = _mm256_min_epu16(adjustv, _mm256_srli_si256(adjustv, 16));
|
|
adjustv = _mm256_min_epu16(adjustv, _mm256_srli_si256(adjustv, 8));
|
|
adjustv = _mm256_min_epu16(adjustv, _mm256_srli_si256(adjustv, 4));
|
|
|
|
t.v = adjustv;
|
|
adjust = t.w[0];
|
|
adjustv = _mm256_set1_epi16(adjust);
|
|
|
|
/* We cannot use a saturated subtract, because we often have to adjust by more than SHRT_MAX
|
|
* This is okay since it can't overflow anyway
|
|
*/
|
|
for (i = 0; i < 4; i++)
|
|
vp->new_metrics->v[i] = _mm256_sub_epi16(vp->new_metrics->v[i], adjustv);
|
|
}
|
|
|
|
d++;
|
|
/* Swap pointers to old and new metrics */
|
|
tmp = vp->old_metrics;
|
|
vp->old_metrics = vp->new_metrics;
|
|
vp->new_metrics = tmp;
|
|
}
|
|
|
|
if (best_state) {
|
|
uint32_t i, bst = 0;
|
|
|
|
uint16_t minmetric = UINT16_MAX;
|
|
for (i = 0; i < 64; i++) {
|
|
if (vp->old_metrics->c[i] <= minmetric) {
|
|
bst = i;
|
|
minmetric = vp->old_metrics->c[i];
|
|
}
|
|
}
|
|
*best_state = bst;
|
|
}
|
|
|
|
#ifdef DEBUG
|
|
printf("];\n===========================================\n");
|
|
#endif
|
|
|
|
vp->dp = d;
|
|
}
|
|
|
|
#endif
|