[FFmpeg-devel] [PATCH + RFC] Faster ff_celp_lp_synthesis_filterf() (and failed SSE SIMD version)

Jason Garrett-Glaser darkshikari
Sun Dec 13 22:20:36 CET 2009


On Sun, Dec 13, 2009 at 11:55 AM, Vitor Sessak <vitor1001 at gmail.com> wrote:
> Hi,
>
> ff_celp_lp_synthesis_filterf() is used for QCELP and RA288, and in the
> future ARMNB, SIPR and WMAVoice. It always shows up as one of the most
> costly functions when profiling, so here is my try to optimize it.

Don't have time to look over the SSE in full now, but your primary
problem is that you have way too many shuffles and moves.  If you want
to make it worthwhile, you'll have to find a way to avoid them.

Dark Shikari



More information about the ffmpeg-devel mailing list