[FFmpeg-devel] [PATCH] Patch cleanup for MPEG 1 & 2 optimizations

[Michael Niedermayer]

On Sun, Apr 13, 2008 at 10:22:59PM -0400, Jeff Downs wrote:
> On Sun, 13 Apr 2008, Michael Niedermayer wrote:
> 
> > [...]
> > 
> > Can you try with the original patch as well
> > (http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/2007-August/053114.html)
> 
> I saw similar benchmarks to those previously posted (OP's patches had 
> little/no impact on speed).
> 
> The original patch cited here won't apply clean...
> Instead of benchmarking that, I took the OP's patches and hardcoded the 
> values for is_mpeg12.  Using that on a short MPEG-1 file I get:
> 
> Current SVN:
> User: avg: 0.201  stddev: 0.005  med: 0.202
> Real: avg: 0.207  stddev: 0.005  med: 0.206
> 
> OP's patches w/ is_mpeg12 hardcoded true:
> User: avg: 0.197  stddev: 0.003  med: 0.198
> Real: avg: 0.200  stddev: 0.002  med: 0.200
> 
> These are 10 runs each with one cache bake throwaway run at start. So, 
> about 2% improvement (on core2 duo).

The question remaining is where does gcc mess up this time?
Any obvious things it inlines/doesnt inline
anything visible with -Wdisabled-optimization -Winline ?

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Many that live deserve death. And some that die deserve life. Can you give
it to them? Then do not be too eager to deal out death in judgement. For
even the very wise cannot see all ends. -- Gandalf
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20080414/8d10049d/attachment.pgp>