diff options
| author | Jason Garrett-Glaser <darkshikari@gmail.com> | 2010-07-02 05:27:41 +0000 |
|---|---|---|
| committer | Jason Garrett-Glaser <darkshikari@gmail.com> | 2010-07-02 05:27:41 +0000 |
| commit | dcc602d8022aac4d419038d6020e07584011aa8a (patch) | |
| tree | be5dc8007fbe6f763317f5b7fa6014e63ae82a7a /libavcodec/x86/vp8dsp-init.c | |
| parent | 301ab19dd97deb30a073e93d3c8a969bb7f6966f (diff) | |
| download | ffmpeg-dcc602d8022aac4d419038d6020e07584011aa8a.tar.gz | |
SSSE3 versions of width4 VP8 6-tap MC functions
Also make some small changes to saturation order of 4-tap SSSE3 MC to fix a
non-bitexactness bug.
Patch mostly by Eli Friedman <eli.friedman AT gmail DOT com>.
Originally committed as revision 23965 to svn://svn.ffmpeg.org/ffmpeg/trunk
Diffstat (limited to 'libavcodec/x86/vp8dsp-init.c')
| -rw-r--r-- | libavcodec/x86/vp8dsp-init.c | 18 |
1 files changed, 18 insertions, 0 deletions
diff --git a/libavcodec/x86/vp8dsp-init.c b/libavcodec/x86/vp8dsp-init.c index fbdc2c70c8..6247da9dcc 100644 --- a/libavcodec/x86/vp8dsp-init.c +++ b/libavcodec/x86/vp8dsp-init.c @@ -54,6 +54,18 @@ extern void ff_put_vp8_epel8_v6_sse2 (uint8_t *dst, int dststride, uint8_t *src, int srcstride, int height, int mx, int my); +extern void ff_put_vp8_epel4_h4_ssse3 (uint8_t *dst, int dststride, + uint8_t *src, int srcstride, + int height, int mx, int my); +extern void ff_put_vp8_epel4_h6_ssse3 (uint8_t *dst, int dststride, + uint8_t *src, int srcstride, + int height, int mx, int my); +extern void ff_put_vp8_epel4_v4_ssse3 (uint8_t *dst, int dststride, + uint8_t *src, int srcstride, + int height, int mx, int my); +extern void ff_put_vp8_epel4_v6_ssse3 (uint8_t *dst, int dststride, + uint8_t *src, int srcstride, + int height, int mx, int my); extern void ff_put_vp8_epel8_h4_ssse3 (uint8_t *dst, int dststride, uint8_t *src, int srcstride, int height, int mx, int my); @@ -173,6 +185,11 @@ HVTAPSSE2(6, 4, 8) HVTAPSSE2(6, 6, 8) HVTAPSSE2(6, 6, 16) +HVTAP(ssse3, 16, 4, 4, 4, 8) +HVTAP(ssse3, 16, 4, 6, 4, 8) +HVTAP(ssse3, 16, 6, 4, 4, 8) +HVTAP(ssse3, 16, 6, 6, 4, 8) + #define HVBILIN(OPT, ALIGN, SIZE, MAXHEIGHT) \ static void ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT( \ uint8_t *dst, int dststride, uint8_t *src, \ @@ -264,6 +281,7 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c) if (mm_flags & FF_MM_SSSE3) { VP8_LUMA_MC_FUNC(0, 16, ssse3); VP8_MC_FUNC(1, 8, ssse3); + VP8_MC_FUNC(2, 4, ssse3); VP8_BILINEAR_MC_FUNC(0, 16, ssse3); VP8_BILINEAR_MC_FUNC(1, 8, ssse3); } |