[FFmpeg-devel] threads regression tests seem to always fail on core 2 duo

[Michael Niedermayer]

On Mon, Jan 21, 2008 at 02:56:23PM +0100, Baptiste Coudurier wrote:
> Hi guys,
> 
> Baptiste Coudurier wrote:
> > Hi all,
> > 
> > On my core 2 duo box, threads regression tests always fail:
> > 
> > processor       : 0
> > vendor_id       : GenuineIntel
> > cpu family      : 6
> > model           : 15
> > model name      : Intel(R) Core(TM)2 Duo CPU     E6850  @ 3.00GHz
> > stepping        : 11
> > cpu MHz         : 2992.719
> > cache size      : 4096 KB
> > physical id     : 0
> > siblings        : 2
> > core id         : 0
> > cpu cores       : 2
> > fdiv_bug        : no
> > hlt_bug         : no
> > f00f_bug        : no
> > coma_bug        : no
> > fpu             : yes
> > fpu_exception   : yes
> > cpuid level     : 10
> > wp              : yes
> > flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge
> > mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm
> > constant_tsc pni monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr lahf_lm
> > bogomips        : 5988.93
> > clflush size    : 64
> > 
> > processor       : 1
> > vendor_id       : GenuineIntel
> > cpu family      : 6
> > model           : 15
> > model name      : Intel(R) Core(TM)2 Duo CPU     E6850  @ 3.00GHz
> > stepping        : 11
> > cpu MHz         : 2992.719
> > cache size      : 4096 KB
> > physical id     : 0
> > siblings        : 2
> > core id         : 1
> > cpu cores       : 2
> > fdiv_bug        : no
> > hlt_bug         : no
> > f00f_bug        : no
> > coma_bug        : no
> > fpu             : yes
> > fpu_exception   : yes
> > cpuid level     : 10
> > wp              : yes
> > flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge
> > mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm
> > constant_tsc pni monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr lahf_lm
> > bogomips        : 5984.98
> > clflush size    : 64
> > 
> > FFmpeg version SVN-r11000, Copyright (c) 2000-2007 Fabrice Bellard, et al.
> >   configuration: --enable-gpl --enable-libmp3lame --enable-libfaac
> > --enable-liba52 --enable-libx264 --enable-pthreads --enable-libxvid
> > --enable-libfaad
> >   libavutil version: 49.5.0
> >   libavcodec version: 51.48.0
> >   libavformat version: 51.19.0
> >   built on Nov 11 2007 20:14:35, gcc: 4.2.3 20071014 (prerelease)
> > (Debian 4.2.2-3)
> > usage: ffmpeg [[infile options] -i infile]... {[outfile options] outfile}...
> > 
> > --- ./tests/data/ffmpeg.regression      2007-11-11 20:17:25.000000000 +0100
> > +++ /usr/local/src/ffmpeg/svn2/tests/ffmpeg.regression.ref
> > 2007-11-07 13:00:31.000000000 +0100
> > @@ -27,9 +27,9 @@
> >  803833 ./tests/data/a-mpeg2threadivlc.mpg
> >  c0e8cbde76ff8377494fe6843fd03e11 *./tests/data/out.yuv
> >  stddev:  7.63 PSNR:30.47 bytes:7602176
> > -11cd11a9a797b38679203c21b1c467f5 *./tests/data/a-mpeg2reuse.mpg
> > -2104591 ./tests/data/a-mpeg2reuse.mpg
> > -e1e9c68b0940da9ea6ca86f8c114a7e3 *./tests/data/out.yuv
> > +c8d8b07b5fa97e0affb4a52d4d30a2b6 *./tests/data/a-mpeg2reuse.mpg
> > +2102956 ./tests/data/a-mpeg2reuse.mpg
> > +75d3dfc8133f0122cb9e272a21bc8c5c *./tests/data/out.yuv
> >  stddev:  7.67 PSNR:30.42 bytes:7602176
> >  c83ae8d8f3e2b4506df58e6a2f7e3b2a *./tests/data/a-msmpeg4v2.avi
> >  636512 ./tests/data/a-msmpeg4v2.avi
> > @@ -87,9 +87,9 @@
> >  878264 ./tests/data/a-mpeg4-Q.avi
> >  8995abbcc97ed4767fcbc0bf46accd01 *./tests/data/out.yuv
> >  stddev:  5.61 PSNR:33.13 bytes:7602176
> > -05f6426f5dd4755f5cd72855f25c2837 *./tests/data/a-mpeg4-thread.avi
> > -763972 ./tests/data/a-mpeg4-thread.avi
> > -1bc53964082a0e73d00f57ab383e7268 *./tests/data/out.yuv
> > +a20b1d73bb0671dbd2012494c31719a6 *./tests/data/a-mpeg4-thread.avi
> > +761506 ./tests/data/a-mpeg4-thread.avi
> > +a9835b5b6a34a3fd70c168dc60fb9687 *./tests/data/out.yuv
> >  stddev: 12.31 PSNR:26.31 bytes:7602176
> >  743cf18c1e8941591f8ab41350247c42 *./tests/data/a-mpeg4-PSP.mp4
> >  406441 ./tests/data/a-mpeg4-PSP.mp4
> > 
> > I always get the same diff.
> > 
> 
> Im just pinging the thread in case someone with threads codepath
> knowledge would be kind to have a look, not being able to run full
> regression tests (libavtest pass though) is becoming problematic.

iam aware of a random threads regression failure for these 2 tests
sadly last time i tried (long time ago)
i wasnt able to reliably reproduce it, it worked most of the time ...
so i had not much luck with debugging it. IIRC the problem even
happens with intra only ...

If its 100% reproduceable for you, please try old versions to see
if its a regression but possibly it was always there ...
If its a regression -> find out when we broke it
if not try to find a minimal test case, that is

ffmpeg-nothreads <cmdline>
ffmpeg-threads <cmdline>

with minimal cmdline and different output

of course it could be a seperate issue (maybe specific to core2 ...)
then trying to disble all optims to check if they are the cause
might be worth it as well

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

In a rich man's house there is no place to spit but his face.
-- Diogenes of Sinope
-------------- 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/20080121/b9ff0e8a/attachment.pgp>