[PATCH] smartsched: Tweak the default scheduler intervals

Adam Jackson ajax at nwnk.net
Tue Nov 5 12:12:12 PST 2013


On Tue, 2013-11-05 at 13:03 -0500, Mouse wrote:
> > But then I read BSD's setitimer man page, and was suddenly very happy
> > I run something else.  (Honestly, you're on the runqueue, you know
> > how to set timers more finely than 100Hz and you already expect to be
> > awake for the next 10ms, maybe do the trivial amount of integer math
> > required to kick the timer interrupt both when the user asked for it
> > and on the next scheduler tick.)
> 
> You don't say which BSD, nor which version; there is no single "BSD's
> setitimer manpage".  But at least some of the BSDs run on hardware that
> simply cannot do what you suggest.

Which?  Genuinely curious.

I checked Free and Open, which were nearly identical.  Net has similar
verbiage.  And, if I'm being entirely fair, 'man 7 time' on Linux says
something similar about rounding being possible, but also mentions
jiffies becoming 1kHz and that timers aren't forced to jiffies on (what
I consider) modern common architectures since, wow, bit over six years
ago.

I still wouldn't recommend running with 100Hz jiffies if I wanted a
machine that could do realtime graphics, but the patch should be an
improvement to fairness even if one does do that.  Obviously there's a
tradeoff if you're trying to run on something in the tens-of-MHz-or-less
domain since 1000 IRQs per second starts to become a significant
overhead, but I mean, VMS had dynamic ticks since the early 90s...

- ajax



More information about the xorg-devel mailing list