Re: + restore-missing-sysfs-max_cstate-attr.patch added to -mm tree

[Arjan van de Ven]

On Sat, 1 Dec 2007 02:17:40 -0800
Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:

> On Fri, 30 Nov 2007 22:14:08 -0500 Mark Lord <lkml@xxxxxx> wrote:
> 
> > > latency. If your app cant take any latency, you should set
> > > those... and the side effect is that the kernel will not do
> > > long-latency C-states or P-state transitions..
> > ..
> > 
> > I don't mind the cpufreq changing (actually, I want it to drop in
> > cpugfreq to save power and keep the fan off), but the C-states just
> > kill this app.
> 
> semi-OT: I was finding that disabling cpufreq altogether on the Vaio
> speeds up `quilt push 1000' by a lot - around 30% iirc.

I assume this is using the ondemand governor and not userspace?
(some older distros mistakingly used "userspace" and yes, that will
suck)

# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor 
ondemand

(to make it easily cut-n-pastable to check)

> 
> There do seem to be some unsophisticated decisions in there and we're
> losing quite a bit of performance as a result.

if you can give a simple recipe for one of these, we can add it to our
workload testsuite that at least some of us use every time we change
either the C states or the cpufreq stuff....

(and yes we want to improve things)


-- 
If you want to reach me at my work email, use arjan@xxxxxxxxxxxxxxx
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
--
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/