diff options
| author | Christophe Gisquet <christophe.gisquet@gmail.com> | 2014-04-11 21:02:08 +0200 |
|---|---|---|
| committer | Michael Niedermayer <michaelni@gmx.at> | 2014-04-20 18:25:36 +0200 |
| commit | 319235c67c59d6abaa78d5af57121ab9816f937d (patch) | |
| tree | 4025d69a59c3e70b8cad858d486983b747671236 /libavcodec/vc1dsp.h | |
| parent | de9cd5884822375d492ff4dcc98e55317a66c196 (diff) | |
| download | ffmpeg-319235c67c59d6abaa78d5af57121ab9816f937d.tar.gz | |
vc1dsp: introduce cases for 8x8 and 16x16
This allows further unrolling the DSP implementation where possible.
x86 and ARM DSP modified by simply moving the multiple calls from vc1dec
to the DSP code. Decoding improvements should only occurs because of the
compiler actually able to unroll more.
Decoding time: ~8.80s -> 8.64s (ie around 2%)
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Diffstat (limited to 'libavcodec/vc1dsp.h')
| -rw-r--r-- | libavcodec/vc1dsp.h | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/libavcodec/vc1dsp.h b/libavcodec/vc1dsp.h index 1aef32a306..7cf613597c 100644 --- a/libavcodec/vc1dsp.h +++ b/libavcodec/vc1dsp.h @@ -57,8 +57,8 @@ typedef struct VC1DSPContext { /* put 8x8 block with bicubic interpolation and quarterpel precision * last argument is actually round value instead of height */ - vc1op_pixels_func put_vc1_mspel_pixels_tab[16]; - vc1op_pixels_func avg_vc1_mspel_pixels_tab[16]; + vc1op_pixels_func put_vc1_mspel_pixels_tab[2][16]; + vc1op_pixels_func avg_vc1_mspel_pixels_tab[2][16]; /* This is really one func used in VC-1 decoding */ h264_chroma_mc_func put_no_rnd_vc1_chroma_pixels_tab[3]; |