Bug#547207: Is still reproduceable? [ogg123: widely different CPU usage between invocations]

[Maximiliano Curia]

Hola Florian Weimer!

El 29/01/2012 a las 18:39 escribiste:
> > Please tell us if you can reproduce it in a squeeze, wheezy, or sid
> > installation.

> On squeeze, I still see this:

> Cpu0  : 19.6%us, 27.2%sy,  0.0%ni, 53.2%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st

Which tool are you using to gather this data for a particular process?

> This is on a Core i7 X980.  Decoding Ogg Vorbis shouldn't need 50% of
> one core on this CPU.

Ok, I'm testing this in a quad core with squeeze and it uses ~10% cpu (kernel
2.6.32-5-amd64), which is 40% of one of the processors.

In a similar machine (core 2 duo with squeeze) with kernel
3.2.0-0.bpo.1-686-pae it uses ~12% cpu, which is 24% of one of the processors.

My machine is an i5 running sid in amd64 and it uses ~1.5% cpu which is 6% of
one of the processors.

Just for fun, I also tested ogg123 in a atom single core machine running
squeeze, kernel 2.6.32-5-686, eating 2% of the cpu. :)

The vorbis-tools package is the same in the four machines, so I think the
problem is not in the vorbis-tools package. Seeing this results I think the
problem might be a issue with threads migrating between cpus, which shouldn't
happend.

It would be great if you could test it with the kernel 3.2.0 from
backports.debian.org.

Thanks,
-- 
"If you optimize everything, you will always be unhappy."
-- Donald Knuth
Saludos /\/\ /\ >< `/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.alioth.debian.org/pipermail/pkg-xiph-maint/attachments/20120129/49fa319a/attachment.pgp>