[FFmpeg-devel] [PATCH 2/5] fate: avoid framemd5, use framecrc its faster

Michael Niedermayer michaelni at gmx.at
Thu May 16 22:10:04 CEST 2013

On Thu, May 16, 2013 at 08:35:32PM +0200, Nicolas George wrote:
> Le decadi 20 floréal, an CCXXI, Michael Niedermayer a écrit :
> > hmm, i get:
> > ffmpeg -f rawvideo md5:  user    0m2.212s    sys 0m0.428s
> > ffmpeg -f md5 -          user    0m2.192s    sys 0m0.408s
> > ffmpeg -f crc -          user    0m0.356s    sys 0m0.408s
> > ffmpeg -f null -         user    0m0.008s    sys 0m0.084s
> > aviocat md5:             user    0m2.004s    sys 0m0.108s
> > gnu md5sum               user    0m1.272s    sys 0m0.116s
> > ffadler                  user    0m0.164s    sys 0m0.060s
> > 
> > Note, my ffmpeg tests did read the random file as if it was
> > rawvideo (which it was not) no real demuxing thus done
> > test done with a random 900mb file on sandybridge i7 4ghz
> It is hard to reproduce your runs, and the framework overhead may make
> comparison less relevant.
> I have written a small benchmark for various hash functions, I can send the
> program itself if wanted, and the results are below (the time is in cycles
> per byte).
> You will notice that OpenSSL is always faster than lavu; the difference is
> negligible for SHA-2 on x86_32, but elsewhere it is significant, and it is
> even huge for MD5 on Atom, for some reason.

under which license is the openssl code ? if its lgpl we should
consider replacing ours by the openssl code. If not borrowing its
tricks would make sense


Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Everything should be made as simple as possible, but not simpler.
-- Albert Einstein
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20130516/c2ebdee6/attachment.asc>

More information about the ffmpeg-devel mailing list