lkml.org 
[lkml]   [2002]   [Feb]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] improving O(1)-J9 in heavily threaded situations
Ingo Molnar wrote:
>
> there is one more thing in the -K2 patch that could cause your problems.
> In kernel/softirq.c, you'll find this line:
> //__initcall(spawn_ksoftirqd);
> please uncomment it - this was just a debugging thing that was left in
> the patch accidentally. I've made a -K3 patch that has this fixed. Do you
> still see the audio problems?

I did this and also tried -K3. It didn't fix the problem.

I addded lost block count printing to the SCHED_FIFO server processes. Most
of the loss (about 75%) happens at lowest level soundcard server and rest in
distributor process. Usually it looses 1 block at time but occasionally
there is peak of about 18 lost blocks.

If I make the client process read larger blocks (> 4kB) from the distributor
process number of lost blocks at soundcard server raises significantly. I
can make it a bit smaller without increased loss, but of course it means
larger overhead and eventually more lost blocks if made something like 512
bytes. 4 kB is optimal block size as it's also internal block size used by
soundcard and distributor servers.

4 kB block size means 2.9 ms in time in this case.


- Jussi Laako

--
PGP key fingerprint: 161D 6FED 6A92 39E2 EB5B 39DD A4DE 63EB C216 1E4B
Available at PGP keyservers

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:24    [W:0.074 / U:0.052 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site