Subject: Re: [linux-audio-dev] SCHED_FIFO versus SCHED_RR
From: dave willis (dubson_AT_dhammanet.net)
Date: Wed Nov 28 2001 - 11:45:04 EET
On Tue, 27 Nov 2001, dave willis wrote:
> > Strangely, some people would prefer to have their audio glitch than
> > to have their computer lock them out and require a hard reset. Go
> > figure. ;-)
>
> i'd prefer that, too. but what i meant was i'd like my program to not
> glitch in sound as long as i was still able to use *it*. if i can't
> browse the web at the same time, oh well. this is assuming the
> program is temporarily hogging the cpu, and it will quit hogging soon
> enough on its own (i can't think of any examples on my own, but maybe
> i can write a test for it somehow).
okay, so i've tested this now with my various heavily-modified latencytest
programs. if my program is hogging the entire cpu, of course i'll have
overruns. however, if another program is using 20% of my cpu and my
program decides it needs to use 80% or more, i want it to work fine (as
far as audio output, etc) and the other program to just deal with it. i've
simulated this with two different latencytest programs on two soundcards
(one at sched 98 with cpu 20%, the other at sched 99 with cpu alternating
from 70% to 90%).
if anyone would like my modifications (extra options and information) to
latencytest, let me know. i can put them all into one documented program
(instead of 3 undocumented ones like i have).
and if any of this isn't clear, let me know as i'm very tired and going to
bed now.
-dave
--
perl -e'@email=split(//,".tenmhd\@nosbud");foreach$letter(@email){$
email=$letter.$email;}$email=~s/(m|net\.)/a\1\1/g;print"$email\n";'
This archive was generated by hypermail 2b28 : Wed Nov 28 2001 - 11:40:01 EET