[FFmpeg-devel] [PATCH 2/2] configure: Enable GCC vectorization on ≥4.9

[Clément Bœsch]

On Sun, May 08, 2016 at 11:15:22PM -0300, James Almer wrote:
[...]
> >>> there where failures with some gcc versions recetly on fate that
> >>> disappesred when the gcc version used on these clients changed.
> >>> I dont know if anyone identified what was the cause of these issues
> >>
> >> Do you know what clients? You can look at the history to see what the failed test
> >> was and the compiler version used.
> > 
> > i think they where some clients from ubitux IIRC
> 
> Ubitux's clients haven't changed the gcc version for some months now. Afaik the
> current one is the same as it was before tree vectorization was enabled.

> His clients also every now and then fail for some weird reason unrelated to gcc or
> even the latest ffmpeg snapshot. They seem to make changes to the source files,
> adding random characters that generate errors. See
> 
> http://fate.ffmpeg.org/log.cgi?time=20160503100201&log=compile&slot=x86_64-archlinux-gcc-threads
> http://fate.ffmpeg.org/log.cgi?time=20160424045346&log=compile&slot=x86_64-archlinux-gcc-threads

Yeah, I got all kind of freaking random bitflip everywhere, like 2 or 3x a
week (and that's only accounting the one i detected). I recently disabled
a disk from the raid and it doesn't seem to happen anymore so I'm guessing
it's a bug in the HD firmware or something. I'll try to swap the disk in
use for the raid to see if the bit rot happens again starting next week or
so.

Anyway, sorry about that.

And yeah, I haven't updated GCC for a while now, but will do when I figure
out the bitrot shitstorm properly.

-- 
Clément B.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 473 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20160509/7c65a111/attachment.sig>