Mercurial > libavcodec.hg
diff x86/h264dsp_mmx.c @ 12437:b242eb86ea9a libavcodec
Move H264 chroma MC from inline asm to yasm. This fixes VP3/5/6 and VC-1
fate failures on Win64.
author | rbultje |
---|---|
date | Mon, 30 Aug 2010 16:31:04 +0000 |
parents | fe78a4548d12 |
children | 33ecda76b2f2 |
line wrap: on
line diff
--- a/x86/h264dsp_mmx.c Mon Aug 30 16:25:46 2010 +0000 +++ b/x86/h264dsp_mmx.c Mon Aug 30 16:31:04 2010 +0000 @@ -2105,127 +2105,6 @@ H264_MC_816(H264_MC_HV, ssse3) #endif -/* rnd interleaved with rnd div 8, use p+1 to access rnd div 8 */ -DECLARE_ALIGNED(8, static const uint64_t, h264_rnd_reg)[4] = { - 0x0020002000200020ULL, 0x0004000400040004ULL, 0x001C001C001C001CULL, 0x0003000300030003ULL -}; - -#define H264_CHROMA_OP(S,D) -#define H264_CHROMA_OP4(S,D,T) -#define H264_CHROMA_MC8_TMPL put_h264_chroma_generic_mc8_mmx -#define H264_CHROMA_MC4_TMPL put_h264_chroma_generic_mc4_mmx -#define H264_CHROMA_MC2_TMPL put_h264_chroma_mc2_mmx2 -#define H264_CHROMA_MC8_MV0 put_pixels8_mmx -#include "dsputil_h264_template_mmx.c" - -static void put_h264_chroma_mc8_mmx_rnd(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - put_h264_chroma_generic_mc8_mmx(dst, src, stride, h, x, y, h264_rnd_reg); -} -static void put_vc1_chroma_mc8_mmx_nornd(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - put_h264_chroma_generic_mc8_mmx(dst, src, stride, h, x, y, h264_rnd_reg+2); -} -static void put_h264_chroma_mc4_mmx(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - put_h264_chroma_generic_mc4_mmx(dst, src, stride, h, x, y, h264_rnd_reg); -} - -#undef H264_CHROMA_OP -#undef H264_CHROMA_OP4 -#undef H264_CHROMA_MC8_TMPL -#undef H264_CHROMA_MC4_TMPL -#undef H264_CHROMA_MC2_TMPL -#undef H264_CHROMA_MC8_MV0 - -#define H264_CHROMA_OP(S,D) "pavgb " #S ", " #D " \n\t" -#define H264_CHROMA_OP4(S,D,T) "movd " #S ", " #T " \n\t"\ - "pavgb " #T ", " #D " \n\t" -#define H264_CHROMA_MC8_TMPL avg_h264_chroma_generic_mc8_mmx2 -#define H264_CHROMA_MC4_TMPL avg_h264_chroma_generic_mc4_mmx2 -#define H264_CHROMA_MC2_TMPL avg_h264_chroma_mc2_mmx2 -#define H264_CHROMA_MC8_MV0 avg_pixels8_mmx2 -#include "dsputil_h264_template_mmx.c" -static void avg_h264_chroma_mc8_mmx2_rnd(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - avg_h264_chroma_generic_mc8_mmx2(dst, src, stride, h, x, y, h264_rnd_reg); -} -static void avg_vc1_chroma_mc8_mmx2_nornd(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - avg_h264_chroma_generic_mc8_mmx2(dst, src, stride, h, x, y, h264_rnd_reg+2); -} -static void avg_h264_chroma_mc4_mmx2(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - avg_h264_chroma_generic_mc4_mmx2(dst, src, stride, h, x, y, h264_rnd_reg); -} -#undef H264_CHROMA_OP -#undef H264_CHROMA_OP4 -#undef H264_CHROMA_MC8_TMPL -#undef H264_CHROMA_MC4_TMPL -#undef H264_CHROMA_MC2_TMPL -#undef H264_CHROMA_MC8_MV0 - -#define H264_CHROMA_OP(S,D) "pavgusb " #S ", " #D " \n\t" -#define H264_CHROMA_OP4(S,D,T) "movd " #S ", " #T " \n\t"\ - "pavgusb " #T ", " #D " \n\t" -#define H264_CHROMA_MC8_TMPL avg_h264_chroma_generic_mc8_3dnow -#define H264_CHROMA_MC4_TMPL avg_h264_chroma_generic_mc4_3dnow -#define H264_CHROMA_MC8_MV0 avg_pixels8_3dnow -#include "dsputil_h264_template_mmx.c" -static void avg_h264_chroma_mc8_3dnow_rnd(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - avg_h264_chroma_generic_mc8_3dnow(dst, src, stride, h, x, y, h264_rnd_reg); -} -static void avg_h264_chroma_mc4_3dnow(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - avg_h264_chroma_generic_mc4_3dnow(dst, src, stride, h, x, y, h264_rnd_reg); -} -#undef H264_CHROMA_OP -#undef H264_CHROMA_OP4 -#undef H264_CHROMA_MC8_TMPL -#undef H264_CHROMA_MC4_TMPL -#undef H264_CHROMA_MC8_MV0 - -#if HAVE_SSSE3 -#define AVG_OP(X) -#undef H264_CHROMA_MC8_TMPL -#undef H264_CHROMA_MC4_TMPL -#define H264_CHROMA_MC8_TMPL put_h264_chroma_mc8_ssse3 -#define H264_CHROMA_MC4_TMPL put_h264_chroma_mc4_ssse3 -#define H264_CHROMA_MC8_MV0 put_pixels8_mmx -#include "dsputil_h264_template_ssse3.c" -static void put_h264_chroma_mc8_ssse3_rnd(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - put_h264_chroma_mc8_ssse3(dst, src, stride, h, x, y, 1); -} -static void put_vc1_chroma_mc8_ssse3_nornd(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - put_h264_chroma_mc8_ssse3(dst, src, stride, h, x, y, 0); -} - -#undef AVG_OP -#undef H264_CHROMA_MC8_TMPL -#undef H264_CHROMA_MC4_TMPL -#undef H264_CHROMA_MC8_MV0 -#define AVG_OP(X) X -#define H264_CHROMA_MC8_TMPL avg_h264_chroma_mc8_ssse3 -#define H264_CHROMA_MC4_TMPL avg_h264_chroma_mc4_ssse3 -#define H264_CHROMA_MC8_MV0 avg_pixels8_mmx2 -#include "dsputil_h264_template_ssse3.c" -static void avg_h264_chroma_mc8_ssse3_rnd(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - avg_h264_chroma_mc8_ssse3(dst, src, stride, h, x, y, 1); -} -static void avg_vc1_chroma_mc8_ssse3_nornd(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y) -{ - avg_h264_chroma_mc8_ssse3(dst, src, stride, h, x, y, 0); -} -#undef AVG_OP -#undef H264_CHROMA_MC8_TMPL -#undef H264_CHROMA_MC4_TMPL -#undef H264_CHROMA_MC8_MV0 -#endif - /***********************************/ /* weighted prediction */