On Sun, 28 Jul 2002, Ingo Molnar wrote:
> On Sun, 28 Jul 2002, Andrew Morton wrote:
>
> > Ingo, could you please review the use of inlines in the
> > scheduler sometime? They seem to be excessive.
> >
> > For example, this patch reduces the sched.d icache footprint
> > by 1.5 kilobytes.
>
> the patch also hurts context-switch latencies - it went
> from 1.35 usecs to 1.42 usecs - a 5% drop.
That's 140 cpu cycles on a 2 GHz CPU, probably less time than
what a single cache miss would cost for non-scheduler-intensive
workloads.
This would mean that for a workload which is doing real work,
and pushes the scheduler out of the L1 cache, the slowdown
would be a lot more simply due to the extra cache misses the
extra inlines generate.
regards,
Rik
-- Bravely reimplemented by the knights who say "NIH".http://www.surriel.com/ http://distro.conectiva.com/
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Jul 30 2002 - 14:00:30 EST