Subject: Re: [linux-audio-dev] 2.4.0-test8 low-latency
From: Paul Winkler (slinkp23_AT_yahoo.com)
Date: Thu Sep 14 2000 - 01:11:30 EEST
Benno Senoner wrote:
> The latency benchmark has priority (SCHED_FIFO) over the outside stress code,
> thus the stress code gets only what's left in terms of CPU cycles.
> So it should not matter much if you run with CPU load=80% or 60% in terms
> of the latency graph, but it will make alot of difference in terms of disk IO
> speed, especially during the write tests, since it involves the processing of
> large
> buffers in RAM which eat CPU too.
What seems to be happening is that when I go over 77%, the stress
code gets few if any cpu cycles until the annoying sound has run its
course which is happening slower than it should. And then when the
sound stops the stress code suddenly starts running. For instance
the x11perf graph doesn't do anything for ten minutes or so; it just
sits there, blank and white until the sound stops. The result is
that the latency graphs look almost perfect, which is rather
misleading because the sound is broken up and the system is
completely useless until the test finishes (the only way I've been
able to abort a test is with the magic sysrq keys - Alt-SysRq-e
works. But then I've killed the sound driver and almost everything
else!)
> The next step would be to create a latencytest tool which varies the CPU
> load and measures when things begin to get ugly (eg big disk performance
> degradation). This such a tool one could tune the machine and limit the
> CPU usage by the audio apps in order to get optimal performance.
That would be pretty cool!
--
................. paul winkler ..................
slinkP arts: music, sound, illustration, design, etc.
web page: http://www.slinkp.com
A member of ARMS: http://www.reacharms.com
This archive was generated by hypermail 2b28 : Thu Sep 14 2000 - 02:05:52 EEST