Mercurial > libavcodec.hg
changeset 12366:09a31ef6ed58 libavcodec
H.264: SSE2/SSSE3 weighted prediction asm
Patch by Eli Friedman <eli.friedman at gmail dot com>
author | darkshikari |
---|---|
date | Thu, 05 Aug 2010 00:13:38 +0000 |
parents | 7c54834209f6 |
children | 06bdd447f4f7 |
files | x86/Makefile x86/dsputil_mmx.c x86/h264dsp_mmx.c |
diffstat | 3 files changed, 23 insertions(+), 0 deletions(-) [+] |
line wrap: on
line diff
--- a/x86/Makefile Wed Aug 04 23:04:05 2010 +0000 +++ b/x86/Makefile Thu Aug 05 00:13:38 2010 +0000 @@ -10,6 +10,7 @@ YASM-OBJS-$(CONFIG_H264DSP) += x86/h264_deblock_sse2.o \ x86/h264_intrapred.o \ + x86/h264_weight_sse2.o \ YASM-OBJS-$(CONFIG_VC1_DECODER) += x86/vc1dsp_yasm.o
--- a/x86/dsputil_mmx.c Wed Aug 04 23:04:05 2010 +0000 +++ b/x86/dsputil_mmx.c Thu Aug 05 00:13:38 2010 +0000 @@ -3000,6 +3000,8 @@ c->h264_h_loop_filter_luma_intra = ff_x264_deblock_h_luma_intra_mmxext; #endif if( mm_flags&FF_MM_SSE2 ){ + c->biweight_h264_pixels_tab[0]= ff_h264_biweight_16x16_sse2; + c->biweight_h264_pixels_tab[3]= ff_h264_biweight_8x8_sse2; #if ARCH_X86_64 || !defined(__ICC) || __ICC > 1110 c->h264_v_loop_filter_luma = ff_x264_deblock_v_luma_sse2; c->h264_h_loop_filter_luma = ff_x264_deblock_h_luma_sse2; @@ -3012,6 +3014,10 @@ c->h264_idct_add16intra = ff_h264_idct_add16intra_sse2; #endif } + if ( mm_flags&FF_MM_SSSE3 ){ + c->biweight_h264_pixels_tab[0]= ff_h264_biweight_16x16_ssse3; + c->biweight_h264_pixels_tab[3]= ff_h264_biweight_8x8_ssse3; + } } #endif }
--- a/x86/h264dsp_mmx.c Wed Aug 04 23:04:05 2010 +0000 +++ b/x86/h264dsp_mmx.c Thu Aug 05 00:13:38 2010 +0000 @@ -2323,6 +2323,22 @@ H264_WEIGHT( 4, 4) H264_WEIGHT( 4, 2) +void ff_h264_biweight_8x8_sse2(uint8_t *dst, uint8_t *src, int stride, + int log2_denom, int weightd, int weights, + int offset); + +void ff_h264_biweight_16x16_sse2(uint8_t *dst, uint8_t *src, int stride, + int log2_denom, int weightd, int weights, + int offset); + +void ff_h264_biweight_8x8_ssse3(uint8_t *dst, uint8_t *src, int stride, + int log2_denom, int weightd, int weights, + int offset); + +void ff_h264_biweight_16x16_ssse3(uint8_t *dst, uint8_t *src, int stride, + int log2_denom, int weightd, int weights, + int offset); + void ff_pred16x16_vertical_mmx (uint8_t *src, int stride); void ff_pred16x16_vertical_sse (uint8_t *src, int stride); void ff_pred16x16_horizontal_mmx (uint8_t *src, int stride);