2011-08-09 11:00:09 +02:00
|
|
|
/*
|
|
|
|
* RV40 decoder motion compensation functions x86-optimised
|
|
|
|
* Copyright (c) 2008 Konstantin Shishkov
|
|
|
|
*
|
|
|
|
* This file is part of Libav.
|
|
|
|
*
|
|
|
|
* Libav is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* Libav is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
|
|
* License along with Libav; if not, write to the Free Software
|
|
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
|
|
|
*/
|
|
|
|
|
|
|
|
/**
|
|
|
|
* @file
|
|
|
|
* RV40 decoder motion compensation functions x86-optimised
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "libavcodec/rv34dsp.h"
|
|
|
|
|
|
|
|
void ff_put_rv40_chroma_mc8_mmx (uint8_t *dst, uint8_t *src,
|
|
|
|
int stride, int h, int x, int y);
|
|
|
|
void ff_avg_rv40_chroma_mc8_mmx2 (uint8_t *dst, uint8_t *src,
|
|
|
|
int stride, int h, int x, int y);
|
|
|
|
void ff_avg_rv40_chroma_mc8_3dnow(uint8_t *dst, uint8_t *src,
|
|
|
|
int stride, int h, int x, int y);
|
|
|
|
|
|
|
|
void ff_put_rv40_chroma_mc4_mmx (uint8_t *dst, uint8_t *src,
|
|
|
|
int stride, int h, int x, int y);
|
|
|
|
void ff_avg_rv40_chroma_mc4_mmx2 (uint8_t *dst, uint8_t *src,
|
|
|
|
int stride, int h, int x, int y);
|
|
|
|
void ff_avg_rv40_chroma_mc4_3dnow(uint8_t *dst, uint8_t *src,
|
|
|
|
int stride, int h, int x, int y);
|
|
|
|
|
rv40: x86 SIMD for biweight
Provide MMX, SSE2 and SSSE3 versions, with a fast-path when the weights are
multiples of 512 (which is often the case when the values round up nicely).
*_TIMER report for the 16x16 and 8x8 cases:
C:
9015 decicycles in 16, 524257 runs, 31 skips
2656 decicycles in 8, 524271 runs, 17 skips
MMX:
4156 decicycles in 16, 262090 runs, 54 skips
1206 decicycles in 8, 262131 runs, 13 skips
MMX on fast-path:
2760 decicycles in 16, 524222 runs, 66 skips
995 decicycles in 8, 524252 runs, 36 skips
SSE2:
2163 decicycles in 16, 262131 runs, 13 skips
832 decicycles in 8, 262137 runs, 7 skips
SSE2 with fast path:
1783 decicycles in 16, 524276 runs, 12 skips
711 decicycles in 8, 524283 runs, 5 skips
SSSE3:
2117 decicycles in 16, 262136 runs, 8 skips
814 decicycles in 8, 262143 runs, 1 skips
SSSE3 with fast path:
1315 decicycles in 16, 524285 runs, 3 skips
578 decicycles in 8, 524286 runs, 2 skips
This means around a 4% speedup for some sequences.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
2012-01-12 00:11:15 +01:00
|
|
|
#define DECLARE_WEIGHT(opt) \
|
2012-03-19 22:46:28 +01:00
|
|
|
void ff_rv40_weight_func_rnd_16_##opt(uint8_t *dst, uint8_t *src1, uint8_t *src2, \
|
|
|
|
int w1, int w2, ptrdiff_t stride); \
|
|
|
|
void ff_rv40_weight_func_rnd_8_##opt (uint8_t *dst, uint8_t *src1, uint8_t *src2, \
|
|
|
|
int w1, int w2, ptrdiff_t stride); \
|
|
|
|
void ff_rv40_weight_func_nornd_16_##opt(uint8_t *dst, uint8_t *src1, uint8_t *src2, \
|
|
|
|
int w1, int w2, ptrdiff_t stride); \
|
|
|
|
void ff_rv40_weight_func_nornd_8_##opt (uint8_t *dst, uint8_t *src1, uint8_t *src2, \
|
|
|
|
int w1, int w2, ptrdiff_t stride);
|
rv40: x86 SIMD for biweight
Provide MMX, SSE2 and SSSE3 versions, with a fast-path when the weights are
multiples of 512 (which is often the case when the values round up nicely).
*_TIMER report for the 16x16 and 8x8 cases:
C:
9015 decicycles in 16, 524257 runs, 31 skips
2656 decicycles in 8, 524271 runs, 17 skips
MMX:
4156 decicycles in 16, 262090 runs, 54 skips
1206 decicycles in 8, 262131 runs, 13 skips
MMX on fast-path:
2760 decicycles in 16, 524222 runs, 66 skips
995 decicycles in 8, 524252 runs, 36 skips
SSE2:
2163 decicycles in 16, 262131 runs, 13 skips
832 decicycles in 8, 262137 runs, 7 skips
SSE2 with fast path:
1783 decicycles in 16, 524276 runs, 12 skips
711 decicycles in 8, 524283 runs, 5 skips
SSSE3:
2117 decicycles in 16, 262136 runs, 8 skips
814 decicycles in 8, 262143 runs, 1 skips
SSSE3 with fast path:
1315 decicycles in 16, 524285 runs, 3 skips
578 decicycles in 8, 524286 runs, 2 skips
This means around a 4% speedup for some sequences.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
2012-01-12 00:11:15 +01:00
|
|
|
DECLARE_WEIGHT(mmx)
|
|
|
|
DECLARE_WEIGHT(sse2)
|
|
|
|
DECLARE_WEIGHT(ssse3)
|
|
|
|
|
2011-08-09 11:00:09 +02:00
|
|
|
void ff_rv40dsp_init_x86(RV34DSPContext *c, DSPContext *dsp)
|
|
|
|
{
|
|
|
|
#if HAVE_YASM
|
2012-01-30 09:39:16 +01:00
|
|
|
int mm_flags = av_get_cpu_flags();
|
|
|
|
|
2011-08-09 11:00:09 +02:00
|
|
|
if (mm_flags & AV_CPU_FLAG_MMX) {
|
|
|
|
c->put_chroma_pixels_tab[0] = ff_put_rv40_chroma_mc8_mmx;
|
|
|
|
c->put_chroma_pixels_tab[1] = ff_put_rv40_chroma_mc4_mmx;
|
2012-03-19 22:46:28 +01:00
|
|
|
c->rv40_weight_pixels_tab[0][0] = ff_rv40_weight_func_rnd_16_mmx;
|
|
|
|
c->rv40_weight_pixels_tab[0][1] = ff_rv40_weight_func_rnd_8_mmx;
|
|
|
|
c->rv40_weight_pixels_tab[1][0] = ff_rv40_weight_func_nornd_16_mmx;
|
|
|
|
c->rv40_weight_pixels_tab[1][1] = ff_rv40_weight_func_nornd_8_mmx;
|
2011-08-09 11:00:09 +02:00
|
|
|
}
|
|
|
|
if (mm_flags & AV_CPU_FLAG_MMX2) {
|
|
|
|
c->avg_chroma_pixels_tab[0] = ff_avg_rv40_chroma_mc8_mmx2;
|
|
|
|
c->avg_chroma_pixels_tab[1] = ff_avg_rv40_chroma_mc4_mmx2;
|
|
|
|
} else if (mm_flags & AV_CPU_FLAG_3DNOW) {
|
|
|
|
c->avg_chroma_pixels_tab[0] = ff_avg_rv40_chroma_mc8_3dnow;
|
|
|
|
c->avg_chroma_pixels_tab[1] = ff_avg_rv40_chroma_mc4_3dnow;
|
|
|
|
}
|
rv40: x86 SIMD for biweight
Provide MMX, SSE2 and SSSE3 versions, with a fast-path when the weights are
multiples of 512 (which is often the case when the values round up nicely).
*_TIMER report for the 16x16 and 8x8 cases:
C:
9015 decicycles in 16, 524257 runs, 31 skips
2656 decicycles in 8, 524271 runs, 17 skips
MMX:
4156 decicycles in 16, 262090 runs, 54 skips
1206 decicycles in 8, 262131 runs, 13 skips
MMX on fast-path:
2760 decicycles in 16, 524222 runs, 66 skips
995 decicycles in 8, 524252 runs, 36 skips
SSE2:
2163 decicycles in 16, 262131 runs, 13 skips
832 decicycles in 8, 262137 runs, 7 skips
SSE2 with fast path:
1783 decicycles in 16, 524276 runs, 12 skips
711 decicycles in 8, 524283 runs, 5 skips
SSSE3:
2117 decicycles in 16, 262136 runs, 8 skips
814 decicycles in 8, 262143 runs, 1 skips
SSSE3 with fast path:
1315 decicycles in 16, 524285 runs, 3 skips
578 decicycles in 8, 524286 runs, 2 skips
This means around a 4% speedup for some sequences.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
2012-01-12 00:11:15 +01:00
|
|
|
if (mm_flags & AV_CPU_FLAG_SSE2) {
|
2012-03-19 22:46:28 +01:00
|
|
|
c->rv40_weight_pixels_tab[0][0] = ff_rv40_weight_func_rnd_16_sse2;
|
|
|
|
c->rv40_weight_pixels_tab[0][1] = ff_rv40_weight_func_rnd_8_sse2;
|
|
|
|
c->rv40_weight_pixels_tab[1][0] = ff_rv40_weight_func_nornd_16_sse2;
|
|
|
|
c->rv40_weight_pixels_tab[1][1] = ff_rv40_weight_func_nornd_8_sse2;
|
rv40: x86 SIMD for biweight
Provide MMX, SSE2 and SSSE3 versions, with a fast-path when the weights are
multiples of 512 (which is often the case when the values round up nicely).
*_TIMER report for the 16x16 and 8x8 cases:
C:
9015 decicycles in 16, 524257 runs, 31 skips
2656 decicycles in 8, 524271 runs, 17 skips
MMX:
4156 decicycles in 16, 262090 runs, 54 skips
1206 decicycles in 8, 262131 runs, 13 skips
MMX on fast-path:
2760 decicycles in 16, 524222 runs, 66 skips
995 decicycles in 8, 524252 runs, 36 skips
SSE2:
2163 decicycles in 16, 262131 runs, 13 skips
832 decicycles in 8, 262137 runs, 7 skips
SSE2 with fast path:
1783 decicycles in 16, 524276 runs, 12 skips
711 decicycles in 8, 524283 runs, 5 skips
SSSE3:
2117 decicycles in 16, 262136 runs, 8 skips
814 decicycles in 8, 262143 runs, 1 skips
SSSE3 with fast path:
1315 decicycles in 16, 524285 runs, 3 skips
578 decicycles in 8, 524286 runs, 2 skips
This means around a 4% speedup for some sequences.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
2012-01-12 00:11:15 +01:00
|
|
|
}
|
|
|
|
if (mm_flags & AV_CPU_FLAG_SSSE3) {
|
2012-03-19 22:46:28 +01:00
|
|
|
c->rv40_weight_pixels_tab[0][0] = ff_rv40_weight_func_rnd_16_ssse3;
|
|
|
|
c->rv40_weight_pixels_tab[0][1] = ff_rv40_weight_func_rnd_8_ssse3;
|
|
|
|
c->rv40_weight_pixels_tab[1][0] = ff_rv40_weight_func_nornd_16_ssse3;
|
|
|
|
c->rv40_weight_pixels_tab[1][1] = ff_rv40_weight_func_nornd_8_ssse3;
|
rv40: x86 SIMD for biweight
Provide MMX, SSE2 and SSSE3 versions, with a fast-path when the weights are
multiples of 512 (which is often the case when the values round up nicely).
*_TIMER report for the 16x16 and 8x8 cases:
C:
9015 decicycles in 16, 524257 runs, 31 skips
2656 decicycles in 8, 524271 runs, 17 skips
MMX:
4156 decicycles in 16, 262090 runs, 54 skips
1206 decicycles in 8, 262131 runs, 13 skips
MMX on fast-path:
2760 decicycles in 16, 524222 runs, 66 skips
995 decicycles in 8, 524252 runs, 36 skips
SSE2:
2163 decicycles in 16, 262131 runs, 13 skips
832 decicycles in 8, 262137 runs, 7 skips
SSE2 with fast path:
1783 decicycles in 16, 524276 runs, 12 skips
711 decicycles in 8, 524283 runs, 5 skips
SSSE3:
2117 decicycles in 16, 262136 runs, 8 skips
814 decicycles in 8, 262143 runs, 1 skips
SSSE3 with fast path:
1315 decicycles in 16, 524285 runs, 3 skips
578 decicycles in 8, 524286 runs, 2 skips
This means around a 4% speedup for some sequences.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
2012-01-12 00:11:15 +01:00
|
|
|
}
|
2011-08-09 11:00:09 +02:00
|
|
|
#endif
|
|
|
|
}
|