diff options
| author | Jason Garrett-Glaser <darkshikari@gmail.com> | 2010-07-23 06:02:52 +0000 |
|---|---|---|
| committer | Jason Garrett-Glaser <darkshikari@gmail.com> | 2010-07-23 06:02:52 +0000 |
| commit | 3ae079a3c8be54cc3c9c7e68c8b2324f3a4e4253 (patch) | |
| tree | 40ed03ae012a9e053643a2b1da000be0e1a32aa6 /libavcodec/x86/vp8dsp-init.c | |
| parent | 3df56f411810c665704518e0e9e083b02eaca573 (diff) | |
| download | ffmpeg-3ae079a3c8be54cc3c9c7e68c8b2324f3a4e4253.tar.gz | |
VP8: optimize DC-only chroma case in the same way as luma.
Add MMX idct_dc_add4uv function for this case.
~40% faster chroma idct.
Originally committed as revision 24455 to svn://svn.ffmpeg.org/ffmpeg/trunk
Diffstat (limited to 'libavcodec/x86/vp8dsp-init.c')
| -rw-r--r-- | libavcodec/x86/vp8dsp-init.c | 16 |
1 files changed, 9 insertions, 7 deletions
diff --git a/libavcodec/x86/vp8dsp-init.c b/libavcodec/x86/vp8dsp-init.c index 5da70824fc..4f40bdd11f 100644 --- a/libavcodec/x86/vp8dsp-init.c +++ b/libavcodec/x86/vp8dsp-init.c @@ -220,8 +220,9 @@ HVBILIN(ssse3, 8, 16, 16) extern void ff_vp8_idct_dc_add_mmx(uint8_t *dst, DCTELEM block[16], int stride); extern void ff_vp8_idct_dc_add_sse4(uint8_t *dst, DCTELEM block[16], int stride); -extern void ff_vp8_idct_dc_add4_mmx(uint8_t *dst, DCTELEM block[4][16], int stride); -extern void ff_vp8_idct_dc_add4_sse2(uint8_t *dst, DCTELEM block[4][16], int stride); +extern void ff_vp8_idct_dc_add4y_mmx(uint8_t *dst, DCTELEM block[4][16], int stride); +extern void ff_vp8_idct_dc_add4y_sse2(uint8_t *dst, DCTELEM block[4][16], int stride); +extern void ff_vp8_idct_dc_add4uv_mmx(uint8_t *dst, DCTELEM block[2][16], int stride); extern void ff_vp8_luma_dc_wht_mmx(DCTELEM block[4][4][16], DCTELEM dc[16]); extern void ff_vp8_idct_add_mmx(uint8_t *dst, DCTELEM block[16], int stride); extern void ff_vp8_idct_add_sse(uint8_t *dst, DCTELEM block[16], int stride); @@ -284,10 +285,11 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c) #if HAVE_YASM if (mm_flags & FF_MM_MMX) { - c->vp8_idct_dc_add = ff_vp8_idct_dc_add_mmx; - c->vp8_idct_dc_add4 = ff_vp8_idct_dc_add4_mmx; - c->vp8_idct_add = ff_vp8_idct_add_mmx; - c->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_mmx; + c->vp8_idct_dc_add = ff_vp8_idct_dc_add_mmx; + c->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_mmx; + c->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_mmx; + c->vp8_idct_add = ff_vp8_idct_add_mmx; + c->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_mmx; c->put_vp8_epel_pixels_tab[0][0][0] = c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_mmx; c->put_vp8_epel_pixels_tab[1][0][0] = @@ -354,7 +356,7 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c) } if (mm_flags & FF_MM_SSE2) { - c->vp8_idct_dc_add4 = ff_vp8_idct_dc_add4_sse2; + c->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_sse2; c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_sse2; c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_sse2; |