2.6.11 low latency audio test results

2.6.11 low latency audio test results

[Lee Revell]

OK, I have run some simple tests with JACK, Hydrogen, and 2.6.11.

2.6.11 does not seem to be much of an improvement over 2.6.10.  It may
in fact be slightly worse.  This was what I expected, as it appears that
a number of latency fixes in the VM got preempted by the 4-level page
tables merge.

At 32 frames (0.667 ms latency) I get an xrun about every 10-20 seconds,
just running JACK and Hydrogen.

At 64 frames (1.33 ms latency) it's better, but I can easily cause
massive xruns with "dbench 32".

At 128 frames (2.66 ms) it seems to work pretty well.

Overall, this puts us about even with Windows XP, and somewhat worse
than Mac OS X.

Of course all of the above settings provide flawless xrun-free
performance with 2.6.11-rc4 + PREEMPT_RT.

Until Ingo releases the RT preempt patch for 2.6.11, I can't provide
details, because the vanilla kernel lacks sufficient instrumentation.
But the above results should help us move in the right direction.

Given the above results, and the performance of the RT patched kernel,
I don't see why 2.6.12 should not be able to solidly outperform Windows
and Mac in this area.

See the "Latency regressions" thread for some areas that might need
attention.

Lee

Re: 2.6.11 low latency audio test results

[K.R. Foley]

Lee Revell wrote:
> OK, I have run some simple tests with JACK, Hydrogen, and 2.6.11.
> 
> 2.6.11 does not seem to be much of an improvement over 2.6.10.  It may
> in fact be slightly worse.  This was what I expected, as it appears that
> a number of latency fixes in the VM got preempted by the 4-level page
> tables merge.
> 
> At 32 frames (0.667 ms latency) I get an xrun about every 10-20 seconds,
> just running JACK and Hydrogen.
> 
> At 64 frames (1.33 ms latency) it's better, but I can easily cause
> massive xruns with "dbench 32".
> 
> At 128 frames (2.66 ms) it seems to work pretty well.
> 
> Overall, this puts us about even with Windows XP, and somewhat worse
> than Mac OS X.
> 
> Of course all of the above settings provide flawless xrun-free
> performance with 2.6.11-rc4 + PREEMPT_RT.
> 

The above mentioned patch will apply (and build and run) just fine to 
2.6.11 if you fix the EXTRAVERSION portion of the patch to not expect -rc4.

> Until Ingo releases the RT preempt patch for 2.6.11, I can't provide
> details, because the vanilla kernel lacks sufficient instrumentation.
> But the above results should help us move in the right direction.
> 
> Given the above results, and the performance of the RT patched kernel,
> I don't see why 2.6.12 should not be able to solidly outperform Windows
> and Mac in this area.
> 
> See the "Latency regressions" thread for some areas that might need
> attention.
> 
> Lee
> 
> 
> -
> 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/
> 

Re: 2.6.11 low latency audio test results

[Lee Revell]

On Wed, 2005-03-09 at 08:27 -0600, K.R. Foley wrote:
> Lee Revell wrote:
>  
> > Of course all of the above settings provide flawless xrun-free
> > performance with 2.6.11-rc4 + PREEMPT_RT.
> > 
> 
> The above mentioned patch will apply (and build and run) just fine to 
> 2.6.11 if you fix the EXTRAVERSION portion of the patch to not expect -rc4.
> 

Right, it sure does.  No rejects except the Makefile.

Looks like the release candidate process is getting tighter.

Lee

Re: 2.6.11 low latency audio test results

[Gene Heskett]

On Thursday 10 March 2005 16:52, Lee Revell wrote:
>On Wed, 2005-03-09 at 08:27 -0600, K.R. Foley wrote:
>> Lee Revell wrote:
>> > Of course all of the above settings provide flawless xrun-free
>> > performance with 2.6.11-rc4 + PREEMPT_RT.
>>
>> The above mentioned patch will apply (and build and run) just fine
>> to 2.6.11 if you fix the EXTRAVERSION portion of the patch to not
>> expect -rc4.
>
>Right, it sure does.  No rejects except the Makefile.
>
>Looks like the release candidate process is getting tighter.
>
>Lee
>
Maybe Lee, maybe.  It kills tvtime, I built that into 2.6.11.2 
yesterday to test.  So I'm back in 2.6.11.2 with only the 
bk-ieee1394.patch applied over the .1 and .2 patches.
>-
>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/

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.34% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com attorneys please note, additions to this message
by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.