[FFmpeg-devel] h264 speed regression after PAFF

[Andreas Öman]

Michael Niedermayer wrote:
> Hi
> 
> On Fri, Oct 12, 2007 at 03:04:42PM -0400, Jeff Downs wrote:
>> Suggested commit msg: "Simplify tests for field macroblocks"
> 
> missing benchmark (though the patch is ok in principle ...)

Applied

> how much do we gain from this relative to the 2 slowdowns?

with patch:
User: avg: 9.995  stddev: 0.012  med: 9.995

svnhead:
User: avg: 10.033  stddev: 0.027  med: 10.033

~0.5%