Re: Horrendous Audio Stutter - current git

["Parag Warudkar" <parag.warudkar@gmail.com>]

On Fri, May 2, 2008 at 4:34 AM, Peter Zijlstra <a.p.zijlstra@chello.nl> wrote:
> On Thu, 2008-05-01 at 20:14 -0400, Parag Warudkar wrote:

>
>  Parag, would you also test with !NEW_FAIR_SLEEPERS to see if that solves
>  your problem?
>
>  The easiest way to disable it is (assumes you have debugfs mounted
>  at /debug):
>
>   # echo NO_NEW_FAIR_SLEEPERS > /debug/sched_features

Seems to make it stutter a little less - but still not usable.

>
>  You can also disable NORMALIZED_SLEEPER that way (of course you would
>  first have to enable NEW_FAIR_SLEEPERS again):
>
>   # echo NO_NORMALIZED_SLEEPER > /debug/sched_features
>
No significant difference from NO_NEW_FAIR_SLEEPERS that I can tell.

>  So by default we have both enabled; could you report if either
>
>  NO_NEW_FAIR_SLEEPERS
>
>  NEW_FAIR_SLEEPERS + NO_NORMALIZED_SLEEPERS
>
>  works for you?

I would say no - the audio still is unusable - slightly better with
NO_NEW_FAIR_SLEEPERS, no difference with the other.

But per Mike's suggestion if I disable CONFIG_GROUP_SCHED ,
CONFIG_FAIR_GROUP_SCHED and CONFIG_USER_SCHED, even with
NEW_FAIR_SLEEPERS audio is good again, which does not change with
NO_NEW_FAIR_SLEEPERS. (No skips for long time under -j8 - still skips
a _very_ tiny bit - noticeable if I listen too carefully - but we can
ignore that for the time being.)

Mike -  the GROUP_SCHED stuff is set to y in the Ubuntu kernel and I
have no audio skips - may be a regression introduced after 2.6.24?

But why is the GROUP_SCHED stuff default y if it so unusable?

>
>  Also, could you apply this patch, and report the bonus_max value for
>  your music player under all three scenarios?
>

I see Frans already reported that and there was some conclusion - let
me know if more data will help.

Thanks!

Parag