On 2011-10-20 03.53, STeve Andre' wrote:
> On 10/18/11 23:58, Amit Kulkarni wrote:
>> On Tue, Oct 18, 2011 at 6:36 PM, STeve Andre'<and...@msu.edu> wrote:
>>>
>> http://www.openbsd.org/cgi-bin/cvsweb/~checkout~/src/sys/kern/sched_bsd.c
>>
>> I think looking at that file, maybe the CPU's cores are all awake
>> every rrticks_init. Is that a big reason for low power savings, each
>> core waking up?
>>
> That is a good question. I've not looked at that code so I can't
> comment. But going from 1.3G to 800M and only seeing a .5w
> drop seems wrong to me, with the power supply not being
> very good.
As someone else also pointed out, in an idle system perhaps you won't
see any significant power savings when reducing the clock frequency.
I mean, idling somewhat slower (eh...) more or less only affects the
interrupt handlers.
I'd measure power under full CPU load with hw.setperf set to 100 and 0
respectively and see what happens.
Also, even if what Amit points out is true, that all cores wake up at
each rrticks_init (and assuming that there's an improvement to be made
there), that would be true even in high performance mode so would most
likely not affect the relative difference in power consumption between
the two modes of operation.
Regards,
/Benny
--
internetlabbet.se / work: +46 8 551 124 80 / "Words must
Benny Lofgren / mobile: +46 70 718 11 90 / be weighed,
/ fax: +46 8 551 124 89 / not counted."
/ email: benny -at- internetlabbet.se
