powerd adaptive mode latching

Igor Mozolevsky igor at hybrid-lab.co.uk
Wed Jan 16 07:27:21 PST 2008


On 16/01/2008, John Baldwin <jhb at freebsd.org> wrote:
> On Saturday 12 January 2008 04:27:46 pm Maxim Sobolev wrote:
> > I wonder if somebody did measurement of power consumption with powerd
> > and without it on typical tasks. There is very interesting idea in the
> > last issue of ACM Queue that it might be much more beneficial to run CPU
> > at the full speed and then switch it to low-power mode as soon as
> > possible in the idle loop than to run longer at reduced speed for a
> > longer period of time.
>
> I haven't done hard tests, but even with C3 enabled on my laptop I find that
> anecdotally I get longer battey life with powerd enabled.

That could be something to do with the rate of throttling of powerd; I
can't remember how long it was exactly, but under the default setting
it would take powerd 9s to go from 100MHz to full clock speed on my
T43p, by which time, I would suspect most of hard tasks would be done
and the clock would be dropped back to something minuscule. Baring in
mind that the power consumption vs clock rate is not a linear
relationship in most cases.

I think the only way we could get a significant benefit from low CPU
clock in idle loop without the powerd would be if/when the kernel
becomes tick-less (which I think is a pretty cool idea)...

Igor


More information about the freebsd-current mailing list