Re: specjbb2005 and aim7 regression with 2.6.32-rc kernels

[Mike Galbraith <efault@gmx.de>]

On Fri, 2009-11-06 at 15:38 +0800, Zhang, Yanmin wrote:
> Comparing with 2.6.31, specjbb2005 and aim7 have some regressions with
> 2.6.32-rc kernels on core2 machines.
> 
> 1) On 4*4 core tigerton: specjbb2005 has about 5% regression.
> 2) On 2*4 stoakley: aim7 has about 5% regression.
> 
> On Nehalem, specjbb2005 has about 2%~8%  improvement instead of regression.
> 
> aim7 has much dependency on schedule patameters, such like sched_latency_ns,
> sched_min_granularity_ns, and sched_wakeup_granularity_ns. 2.6.32-rc kernel
> decreases these parameter values. I restore them and retest aim7 on stoakley.
> aim7 regression becomes about 2% and specjbb2005 regression also becomes
> 2%. But on Nehalem, the improvement shrinks.

Yeah, the price of lower latency.  We may want to tweak big machine
setup a little.

Be advised that there's a locking problem which appears to be falsifying
benchmark results somewhat.  I've got a tentative fix, but I don't think
it's quite enough.  (I haven't looked yet at what protects cpus_allowed,
so aren't sure yet.)  Just wanted to let you know lest your testing time
investment may be producing somewhat skewed results, so you may want to
hold off a little bit.  (your testing time is much appreciated, don't
want to waste a single drop;)

	-Mike