On Thu, Apr 16, 2009 at 07:47:14AM +0200, Henning Brauer wrote:
> * Jason Dixon <[email protected]> [2009-04-16 07:18]:
> > We had a spare set of servers available, so I went back to the lab and
> > reproduced the traffic profile. I then tested the same load with the MP
> > kernel. My tests revealed that even though the kernel is not threaded,
> > we benefit from equal distribution of interrupts across all cores. Our
> > interrupt load effectively decreased by a factor of 4; since we aren't
> > performing any userland activity, the other 3 cores are otherwise
> > unused.
>
> was this 4.5 or earlier? If earlier what you saw could be pic vs apic.
> since 4.5 we have apic usually on UP too.
> if it isn't that, I am stunned. could speculate about better cache
> usage, but that would be about the only idea i'd have.
I think wrong statistics collection in the MP case should also be
considered as a possible cause.
-Otto
