Re: [patch] scheduler: active_load_balance fixes

From: Darren Hart
Date: Tue Oct 26 2004 - 00:00:07 EST


On Sun, 2004-10-24 at 02:37 -0700, Andrew Morton wrote:
> Nick Piggin <piggin@xxxxxxxxxxxxxxx> wrote:
> >
> >
> >
> > Darren Hart wrote:
> >
> > >The following patch against the latest mm fixes several problems with
> > >active_load_balance().
> > >
> > >
> >
> > This seems much better. Andrew can you put this into -mm please.
> >
>
> Whenever we touch the load balancing we get sad little reports about
> performance regressions two months later. How do we gain confidence in
> this change?
>

I ran kernbench and specjbb on a 16 way xeon ht numa machine (32 total
sibling CPUs) and an 8 way ppc64 machine against 2.6.9-mm1 w/ and w/o my
active_load_balance() patch. Kernbench was marginally faster on each
machine, and specjbb performed better on 64% of the tests. SpecJBB is a
bit erratic anyway, so I feel good about these numbers.


Kernbench results below. (2.6.9-mm1-ab is the run with the
active_load_balance patch).

32 way xeon
2.6.9-mm1
Elapsed: 81.444s User: 1044.06s System: 138.008s CPU: 1451.2%
2.6.9-mm1-ab
Elapsed: 81.372s User: 1037.842s System: 139.134s CPU: 1446%

8 way ppc64
2.6.9-mm1
Elapsed: 53.336s User: 352.932s System: 45.302s CPU: 746%
2.6.9-mm1-ab
Elapsed: 53.24s User: 353.096s System: 44.98s CPU: 747%

--
Darren Hart
IBM, Linux Technology Center
503 578 3185
dvhltc@xxxxxxxxxx
-
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/