diff options
| author | Christophe Gisquet <christophe.gisquet@gmail.com> | 2013-04-09 22:16:36 +0200 |
|---|---|---|
| committer | Michael Niedermayer <michaelni@gmx.at> | 2013-04-19 13:19:45 +0200 |
| commit | 76c7277385120741914168d02b57a7a1acf87a9b (patch) | |
| tree | 1a0821b6a1b2aae976515ed3423a05ac9cd245f5 /libavcodec/jfdctint_template.c | |
| parent | 380cfce2b2138a0513f7c054134458b3b4c92fd4 (diff) | |
| download | ffmpeg-76c7277385120741914168d02b57a7a1acf87a9b.tar.gz | |
x86: sbrdsp: implement SSE2 hf_apply_noise
233 to 105 cycles on Arrandale and Win64.
Replacing the multiplication by s_m[m] by a pand and a pxor with
appropriate vectors is slower. Unrolling is a 15 cycles win.
A SSE version was 4 cycles slower.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Diffstat (limited to 'libavcodec/jfdctint_template.c')
0 files changed, 0 insertions, 0 deletions