[LAU] A surprisingly stupid RT priority question

[Paul Davis]

On Sat, Dec 8, 2012 at 2:19 PM, Paul Coccoli <pcoccoli at gmail.com> wrote:

> On Sun, Dec 2, 2012 at 8:01 PM, Paul Davis <paul at linuxaudiosystems.com>
> wrote:
> > nice has absolutely nothing to do with this, and if it has any effect,
> it is
> > accidental and should not be relied on.
>
> I know that's your stock answer whenever someone mentions nice, but if
> the OP is talking about SCHED_OTHER processes, nice does play a role.
>

nice alters the behaviour of scheduler with respect to SCHED_OTHER tasks
using an algorithm that is (almost) completely irrelevant for programs that
do very little interaction with the user and use most of their CPU time
streaming media.

applications that stream media should either (a) use enough buffering that
they do not run into xruns with respect to the delivery endpoint or (b) use
SCHED_FIFO/SCHED_RR (c) both. using nice is a bandaid that simply masks
design problems, if in fact it has the right effect at all.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxaudio.org/pipermail/linux-audio-user/attachments/20121208/2f369ab5/attachment.html>