提交 7b46add7 编写于 作者: R Rostislav Pehlivanov

opus_pvq: do not compile encoding/decoding code if the encoder/decoder is disabled

This should save quite a bit of space if either has been disabled for size reasons.
Could just check if the encoding flag is set during runtime on every single location,
however the overhead of branch misses would somewhat decrease performance.
Signed-off-by: NRostislav Pehlivanov <atomnuker@gmail.com>
上级 ce87e630
......@@ -876,64 +876,22 @@ static av_always_inline uint32_t quant_band_template(CeltPVQ *pvq, CeltFrame *f,
static QUANT_FN(pvq_decode_band)
{
#if CONFIG_OPUS_DECODER
return quant_band_template(pvq, f, rc, band, X, Y, N, b, blocks, lowband, duration,
lowband_out, level, gain, lowband_scratch, fill, 0);
#else
return 0;
#endif
}
static QUANT_FN(pvq_encode_band)
{
#if CONFIG_OPUS_ENCODER
return quant_band_template(pvq, f, rc, band, X, Y, N, b, blocks, lowband, duration,
lowband_out, level, gain, lowband_scratch, fill, 1);
}
static float pvq_band_cost(CeltPVQ *pvq, CeltFrame *f, OpusRangeCoder *rc, int band,
float *bits, float lambda)
{
int i, b = 0;
uint32_t cm[2] = { (1 << f->blocks) - 1, (1 << f->blocks) - 1 };
const int band_size = ff_celt_freq_range[band] << f->size;
float buf[176 * 2], lowband_scratch[176], norm1[176], norm2[176];
float dist, cost, err_x = 0.0f, err_y = 0.0f;
float *X = buf;
float *X_orig = f->block[0].coeffs + (ff_celt_freq_bands[band] << f->size);
float *Y = (f->channels == 2) ? &buf[176] : NULL;
float *Y_orig = f->block[1].coeffs + (ff_celt_freq_bands[band] << f->size);
OPUS_RC_CHECKPOINT_SPAWN(rc);
memcpy(X, X_orig, band_size*sizeof(float));
if (Y)
memcpy(Y, Y_orig, band_size*sizeof(float));
f->remaining2 = ((f->framebits << 3) - f->anticollapse_needed) - opus_rc_tell_frac(rc) - 1;
if (band <= f->coded_bands - 1) {
int curr_balance = f->remaining / FFMIN(3, f->coded_bands - band);
b = av_clip_uintp2(FFMIN(f->remaining2 + 1, f->pulses[band] + curr_balance), 14);
}
if (f->dual_stereo) {
pvq->quant_band(pvq, f, rc, band, X, NULL, band_size, b / 2, f->blocks, NULL,
f->size, norm1, 0, 1.0f, lowband_scratch, cm[0]);
pvq->quant_band(pvq, f, rc, band, Y, NULL, band_size, b / 2, f->blocks, NULL,
f->size, norm2, 0, 1.0f, lowband_scratch, cm[1]);
} else {
pvq->quant_band(pvq, f, rc, band, X, Y, band_size, b, f->blocks, NULL, f->size,
norm1, 0, 1.0f, lowband_scratch, cm[0] | cm[1]);
}
for (i = 0; i < band_size; i++) {
err_x += (X[i] - X_orig[i])*(X[i] - X_orig[i]);
if (Y)
err_y += (Y[i] - Y_orig[i])*(Y[i] - Y_orig[i]);
}
dist = sqrtf(err_x) + sqrtf(err_y);
cost = OPUS_RC_CHECKPOINT_BITS(rc)/8.0f;
*bits += cost;
OPUS_RC_CHECKPOINT_ROLLBACK(rc);
return lambda*dist*cost;
#else
return 0;
#endif
}
int av_cold ff_celt_pvq_init(CeltPVQ **pvq, int encode)
......@@ -944,7 +902,6 @@ int av_cold ff_celt_pvq_init(CeltPVQ **pvq, int encode)
s->pvq_search = ppp_pvq_search_c;
s->quant_band = encode ? pvq_encode_band : pvq_decode_band;
s->band_cost = pvq_band_cost;
if (ARCH_X86)
ff_opus_dsp_init_x86(s);
......
......@@ -37,10 +37,7 @@ struct CeltPVQ {
DECLARE_ALIGNED(32, float, hadamard_tmp)[256];
float (*pvq_search)(float *X, int *y, int K, int N);
QUANT_FN(*quant_band);
float (*band_cost)(struct CeltPVQ *pvq, CeltFrame *f, OpusRangeCoder *rc,
int band, float *bits, float lambda);
};
void ff_opus_dsp_init_x86(struct CeltPVQ *s);
......
......@@ -25,6 +25,56 @@
#include "mdct15.h"
#include "libavutil/qsort.h"
static float pvq_band_cost(CeltPVQ *pvq, CeltFrame *f, OpusRangeCoder *rc, int band,
float *bits, float lambda)
{
int i, b = 0;
uint32_t cm[2] = { (1 << f->blocks) - 1, (1 << f->blocks) - 1 };
const int band_size = ff_celt_freq_range[band] << f->size;
float buf[176 * 2], lowband_scratch[176], norm1[176], norm2[176];
float dist, cost, err_x = 0.0f, err_y = 0.0f;
float *X = buf;
float *X_orig = f->block[0].coeffs + (ff_celt_freq_bands[band] << f->size);
float *Y = (f->channels == 2) ? &buf[176] : NULL;
float *Y_orig = f->block[1].coeffs + (ff_celt_freq_bands[band] << f->size);
OPUS_RC_CHECKPOINT_SPAWN(rc);
memcpy(X, X_orig, band_size*sizeof(float));
if (Y)
memcpy(Y, Y_orig, band_size*sizeof(float));
f->remaining2 = ((f->framebits << 3) - f->anticollapse_needed) - opus_rc_tell_frac(rc) - 1;
if (band <= f->coded_bands - 1) {
int curr_balance = f->remaining / FFMIN(3, f->coded_bands - band);
b = av_clip_uintp2(FFMIN(f->remaining2 + 1, f->pulses[band] + curr_balance), 14);
}
if (f->dual_stereo) {
pvq->quant_band(pvq, f, rc, band, X, NULL, band_size, b / 2, f->blocks, NULL,
f->size, norm1, 0, 1.0f, lowband_scratch, cm[0]);
pvq->quant_band(pvq, f, rc, band, Y, NULL, band_size, b / 2, f->blocks, NULL,
f->size, norm2, 0, 1.0f, lowband_scratch, cm[1]);
} else {
pvq->quant_band(pvq, f, rc, band, X, Y, band_size, b, f->blocks, NULL, f->size,
norm1, 0, 1.0f, lowband_scratch, cm[0] | cm[1]);
}
for (i = 0; i < band_size; i++) {
err_x += (X[i] - X_orig[i])*(X[i] - X_orig[i]);
if (Y)
err_y += (Y[i] - Y_orig[i])*(Y[i] - Y_orig[i]);
}
dist = sqrtf(err_x) + sqrtf(err_y);
cost = OPUS_RC_CHECKPOINT_BITS(rc)/8.0f;
*bits += cost;
OPUS_RC_CHECKPOINT_ROLLBACK(rc);
return lambda*dist*cost;
}
/* Populate metrics without taking into consideration neighbouring steps */
static void step_collect_psy_metrics(OpusPsyContext *s, int index)
{
......@@ -320,7 +370,7 @@ static int bands_dist(OpusPsyContext *s, CeltFrame *f, float *total_dist)
for (i = 0; i < CELT_MAX_BANDS; i++) {
float bits = 0.0f;
float dist = f->pvq->band_cost(f->pvq, f, &dump, i, &bits, s->lambda);
float dist = pvq_band_cost(f->pvq, f, &dump, i, &bits, s->lambda);
tdist += dist;
}
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册