aacenc: add SIMD optimizations for abs_pow34 and quantization
Performance improvements:
quant_bands:
with: 681 decicycles in quant_bands, 8388453 runs, 155 skips
without: 1190 decicycles in quant_bands, 8388386 runs, 222 skips
Around 42% for the function
Twoloop coder:
abs_pow34:
with/without: 7.82s/8.17s
Around 4% for the entire encoder
Both:
with/without: 7.15s/8.17s
Around 12% for the entire encoder
Fast coder:
abs_pow34:
with/without: 3.40s/3.77s
Around 10% for the entire encoder
Both:
with/without: 3.02s/3.77s
Around 20% faster for the entire encoder
Signed-off-by: Rostislav Pehlivanov <[email protected]>
Tested-by: Michael Niedermayer <[email protected]>
Reviewed-by: James Almer <[email protected]>
diff --git a/libavcodec/aaccoder.c b/libavcodec/aaccoder.c
index 35787e8..9f3b4ed 100644
--- a/libavcodec/aaccoder.c
+++ b/libavcodec/aaccoder.c
@@ -88,7 +88,7 @@
float next_minrd = INFINITY;
int next_mincb = 0;
- abs_pow34_v(s->scoefs, sce->coeffs, 1024);
+ s->abs_pow34(s->scoefs, sce->coeffs, 1024);
start = win*128;
for (cb = 0; cb < CB_TOT_ALL; cb++) {
path[0][cb].cost = 0.0f;
@@ -299,7 +299,7 @@
}
}
idx = 1;
- abs_pow34_v(s->scoefs, sce->coeffs, 1024);
+ s->abs_pow34(s->scoefs, sce->coeffs, 1024);
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
start = w*128;
for (g = 0; g < sce->ics.num_swb; g++) {
@@ -446,7 +446,7 @@
if (!allz)
return;
- abs_pow34_v(s->scoefs, sce->coeffs, 1024);
+ s->abs_pow34(s->scoefs, sce->coeffs, 1024);
ff_quantize_band_cost_cache_init(s);
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
@@ -652,8 +652,8 @@
s->fdsp->vector_fmul_scalar(PNS, PNS, scale, sce->ics.swb_sizes[g]);
pns_senergy = s->fdsp->scalarproduct_float(PNS, PNS, sce->ics.swb_sizes[g]);
pns_energy += pns_senergy;
- abs_pow34_v(NOR34, &sce->coeffs[start_c], sce->ics.swb_sizes[g]);
- abs_pow34_v(PNS34, PNS, sce->ics.swb_sizes[g]);
+ s->abs_pow34(NOR34, &sce->coeffs[start_c], sce->ics.swb_sizes[g]);
+ s->abs_pow34(PNS34, PNS, sce->ics.swb_sizes[g]);
dist1 += quantize_band_cost(s, &sce->coeffs[start_c],
NOR34,
sce->ics.swb_sizes[g],
@@ -757,8 +757,9 @@
{
int start = 0, i, w, w2, g, sid_sf_boost, prev_mid, prev_side;
uint8_t nextband0[128], nextband1[128];
- float M[128], S[128];
- float *L34 = s->scoefs, *R34 = s->scoefs + 128, *M34 = s->scoefs + 128*2, *S34 = s->scoefs + 128*3;
+ float *M = s->scoefs + 128*0, *S = s->scoefs + 128*1;
+ float *L34 = s->scoefs + 128*2, *R34 = s->scoefs + 128*3;
+ float *M34 = s->scoefs + 128*4, *S34 = s->scoefs + 128*5;
const float lambda = s->lambda;
const float mslambda = FFMIN(1.0f, lambda / 120.f);
SingleChannelElement *sce0 = &cpe->ch[0];
@@ -789,8 +790,8 @@
S[i] = M[i]
- sce1->coeffs[start+(w+w2)*128+i];
}
- abs_pow34_v(M34, M, sce0->ics.swb_sizes[g]);
- abs_pow34_v(S34, S, sce0->ics.swb_sizes[g]);
+ s->abs_pow34(M34, M, sce0->ics.swb_sizes[g]);
+ s->abs_pow34(S34, S, sce0->ics.swb_sizes[g]);
for (i = 0; i < sce0->ics.swb_sizes[g]; i++ ) {
Mmax = FFMAX(Mmax, M34[i]);
Smax = FFMAX(Smax, S34[i]);
@@ -833,10 +834,10 @@
- sce1->coeffs[start+(w+w2)*128+i];
}
- abs_pow34_v(L34, sce0->coeffs+start+(w+w2)*128, sce0->ics.swb_sizes[g]);
- abs_pow34_v(R34, sce1->coeffs+start+(w+w2)*128, sce0->ics.swb_sizes[g]);
- abs_pow34_v(M34, M, sce0->ics.swb_sizes[g]);
- abs_pow34_v(S34, S, sce0->ics.swb_sizes[g]);
+ s->abs_pow34(L34, sce0->coeffs+start+(w+w2)*128, sce0->ics.swb_sizes[g]);
+ s->abs_pow34(R34, sce1->coeffs+start+(w+w2)*128, sce0->ics.swb_sizes[g]);
+ s->abs_pow34(M34, M, sce0->ics.swb_sizes[g]);
+ s->abs_pow34(S34, S, sce0->ics.swb_sizes[g]);
dist1 += quantize_band_cost(s, &sce0->coeffs[start + (w+w2)*128],
L34,
sce0->ics.swb_sizes[g],
