[linux-audio-user] Which kernel for low latency and kernel 2.6.1-mm# problems

Florian Schmidt mista.tapas at gmx.net
Wed Jan 21 06:46:00 EST 2004


On Wed, 21 Jan 2004 11:15:11 +0000
Steve Harris <S.W.Harris at ecs.soton.ac.uk> wrote:

> > I don't know what these numbers mean...
> > 
> > What does this transalate to in interval-in-milliseconds, and after
> > how many milliseconds delay will an underrun occur?
> 
> The default sampe rate is 48kHz, so 1024 samples/block gives 21ms
> between interupts. Jack easts a bit of that, but you'd have ~20ms
> before an underrun occured.
> 
> > And what else is happening on the system at the time?
> 
> The original poster would have to anwer that. Jack itsself is just
> messing with SYSV SHM and occasioanly touching fifos.
> 
> As someone else said, jack runs reliably on most 2.4+ll+o(1)sched
> systems with 5ms (256 sample) interupt fequencies and usually lower.

I usually use 4 or 8 pretty large periods of 512 frames [my soundcard
doesn't allow larger one]. On linux 2.4.22 [LL+preempt+cap patch] i can
get very stable performance. I can even compile a kernel in the
background and get no xruns [except for an occasional one]. I tested the
2.6.0 kernel and it was much more touchy. Start using the disk and xruns
occur right and left. I will try the 2.6.1 today though..

I don't know though if i configured the 2.6 kernel right [preemptible,
etc..]

I once took a look at the latency measuring tools, but i could not
really make sense out of them [not being a kernel hacker]. So i suppose
testing would be much easier, if there was a little tool, which everyone
could just run and it would do some testing which we could then report
back to andrew..

Flo


-- 
http://www.soundclick.com/bands/9/florianschmidt.htm
http://www.arted.biz/mista.tapas



More information about the Linux-audio-user mailing list