Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 06 Nov 2005 11:20:55 -0800
From:      Nate Lawson <nate@root.org>
To:        =?ISO-8859-1?Q?Dag-Erling_Sm=F8rgrav?= <des@des.no>
Cc:        cvs-src@FreeBSD.org, Mike Silbersack <silby@silby.com>, src-committers@FreeBSD.org, cvs-all@FreeBSD.org
Subject:   Re: cvs commit: src/sys/conf files src/sys/modules/acpi/acpi Makefile src/sys/dev/acpica acpi_battery.c acpi_smbat.c acpi_smbus.h acpiio.h
Message-ID:  <436E5797.7090605@root.org>
In-Reply-To: <861x1u55qg.fsf@xps.des.no>
References:  <200511052355.jA5NtuPg026403@repoman.freebsd.org>	<20051105191616.M870@odysseus.silby.com> <861x1u55qg.fsf@xps.des.no>

next in thread | previous in thread | raw e-mail | index | archive | help
Dag-Erling Smørgrav wrote:
> Mike Silbersack <silby@silby.com> writes:
> 
>>2.  Now, powerd seems to be causing ~30% system cpu load - top shows
>>it switching between the "nanslp" and "ecpoll" wait states.  This may
>>be due to some other recent change to acpi, I'm not sure how to best
>>test.
> 
> 
> Try running powerd with a 5000 ms polling interval.  With the default
> of 500 ms, it never seems to stabilize, but keeps oscillating wildly
> in the 75-300 MHz range on my Dell Latitude D600.

That is bad for performance.  It can then take up to 10-15 seconds to 
promote back to 100% CPU when your system becomes busy.

I've said it many times: we need a real predictive algorithm.  Taking a 
single sample will always have hysteresis problems.  See the acpi@ list 
archives for much more detail and a good summary, including linked 
papers, here.

http://wikitest.freebsd.org/moin.cgi/powerd

-- 
Nate



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?436E5797.7090605>