Date: Thu, 26 Oct 2006 13:39:43 -0700 From: "Kevin Oberman" <oberman@es.net> To: Nate Lawson <nate@root.org> Cc: acpi@freebsd.org Subject: Re: Odd power management on ThinkPad T43 Message-ID: <20061026203943.CB9D04504D@ptavv.es.net> In-Reply-To: Your message of "Wed, 04 Oct 2006 22:43:43 PDT." <45249B8F.2060407@root.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--==_Exmh_1161895183_68145P Content-Type: text/plain; charset=us-ascii Content-Disposition: inline > Date: Wed, 04 Oct 2006 22:43:43 -0700 > From: Nate Lawson <nate@root.org> > > Kevin Oberman wrote: > > I'm running current on an IBM ThinkPad T43 and I'm not sure I have a problem, > > but something odd seems to be going on. > > > > I have a 2.0 GHz Pentium-M which I believe is 760. I believe it's one IBM has > > not released information on the EST specs. > > > > If I do NOT have cpufreq loaded, I see: > > dev.cpu.0.freq_levels: 2000/27000 1750/23625 1600/22600 1400/19775 1333/19666 > > 1166/17207 1066/16733 932/14641 800/13800 700/12075 600/10350 500/8625 > > 400/6900 300/5175 200/3450 100/1725 > > > > If I load cpufreq I see: > > dev.cpu.0.freq_levels: 1500/-1 1312/-1 1200/-1 1050/-1 1000/-1 875/-1 800/-1 > > 700/-1 600/-1 525/-1 450/-1 375/-1 300/-1 225/-1 150/-1 75/-1 > > > > With cpufreq I report perf0, est0 and p4tcc0 in dmesg. Without loading cpufreq > > I still see acpi_perf0 and acpi_throttle0. > > > > This would lead me to believe that without cpufreq I am only seeing > > throttling, but I see my clock speed decrease (x86info) which I did not expect > > to see with pure throttling. > > > > Am I better off when on battery to use cpufreq or not? Is there something to > > tweak to get full 2GHz performance with EST? > > This sounds like a bad table match for est0. Perhaps it's detecting > your CPU as a 1.5Ghz one when it's actually 2Ghz. > > An easy way to tell is to load cpufreq but disable just est with: > hint.est.0.disabled="1" > > You should get acpi_perf and p4tcc, and the frequencies will be correct > for your system. acpi_perf is often more user-friendly anyway since it > reports the power consumed at each level instead of just "-1". > > It's also possible you were booting on battery and had lower levels > available. Easy way to tell is report output of sysctl -a | grep cpu Well, this experiment was not a success at all! With cpufreq and EST disabled, I had 8 "steps", none at the points they should have been and testing speed and performance showed almost no power reduction between steps and better performance at 496 MHz than at 1985 MHz. I get the following frequency/energy pairs for different setting: dev.cpu.0.freq_levels: 1985/-1 1736/-1 1488/-1 1240/-1 992/-1 744/-1 496/-1 248/-1 The performance measured with byte/sec of MD5 calculations is: 1985 154775012 1736 141901764 1488 140810800 1240 154728184 992 125388600 744 107718376 496 154821869 248 112606365 248 107675442 496 110848872 744 111006586 992 114505624 1240 150149431 1488 136979772 1736 152260357 1985 154612270 High numbers are good. I would describe these results as totally weird. EST vs. running without cpufreq show very consistent and linear change in both performance and power consumption with speed. Only running without cpufreq provides the energy values. I'm baffled, but I think I can live with just running without cpufreq. I'm just not real sure if there is any real difference between running with EST or without cpufreq. My tests sure don't show much. Thanks for the suggestions, even if the results were not what was expected. -- R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: oberman@es.net Phone: +1 510 486-8634 Key fingerprint:059B 2DDF 031C 9BA3 14A4 EADA 927D EBB3 987B 3751 --==_Exmh_1161895183_68145P Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (FreeBSD) Comment: Exmh version 2.5 06/03/2002 iD8DBQFFQR0Pkn3rs5h7N1ERAoehAJ0ThchF5AySmMmBSmiPSsVArxOdsACdHj5N FCHeo3UaDHP/jexzcD0Ov94= =toQo -----END PGP SIGNATURE----- --==_Exmh_1161895183_68145P--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20061026203943.CB9D04504D>