Re: CFS scheduler OLTP perforamnce

[Andi Kleen]

On Fri, Dec 12, 2008 at 03:22:15PM +0100, Peter Zijlstra wrote:
> On Fri, 2008-12-12 at 15:15 +0100, Andi Kleen wrote:
> > Peter Zijlstra <peterz@xxxxxxxxxxxxx> writes:
> > >> 
> > >> It seems that in this case renice to higher priority with CFS did not
> > >> reduce scheduling latency as well as SCHED_RR.
> > >
> > > Is there a question in this email?
> > 
> > The question is how to make nice perform as well as SCHED_RR.
> 
> Depending on the circumstances, you can't - SCHED_RR doesn't bother with
> fairness.

When the spread between nice levels (negative/positive) is large enough
at least the log writer should be able to schedule soon most of the
time, no?

At least that doesn't seem to work.

Also in general there seems to be a starvation issue here between
producer and consumer.

-Andi
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/