[Ffmpeg-devel] [PATCH] fix mpeg4 lowres chroma bug and increase h264/mpeg4 MC speed

Michael Niedermayer michaelni
Mon Feb 19 14:50:15 CET 2007


On Sun, Feb 18, 2007 at 05:24:25PM -0800, Trent Piepho wrote:
> > also the amount of time (repeating benchmarks) and inconvinience (stoping all
> > applications and deamons) must be considered
> It would even easier to just flip a coin...

yes but that wouldnt give you an estimate of the speed of the code, anyway the
disscussion apparently has drifted away quite a bit from the original subject
which is a bug in a motion compensation function (which i do like to fix ...),
you dont dispute that your code is slower you rather argue that
statistics based on random normal distributed errors would be universally
supperior even when none of its axioms are true, not to mention that your
benchmarking will almost certainly give the same awnser on the same hw

also you dont seem to like reality, effort and time needed for benchmarks does
matter a lot, discarding outliers is common statistics practice

i like to benchmark my asm code frequently, that is after every change, 
spending 10min or more each time for a perfect benchmark is not realistic

Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

No great genius has ever existed without some touch of madness. -- Aristotle
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20070219/89f1c65f/attachment.pgp>

More information about the ffmpeg-devel mailing list