From mboxrd@z Thu Jan 1 00:00:00 1970 From: Dominik Brodowski Subject: Re: [PATCH 2.6] update passive cooling algorithm Date: Mon, 12 Jan 2004 18:39:22 +0100 Sender: cpufreq-bounces+glkc-cpufreq=gmane.org@www.linux.org.uk Message-ID: <20040112173922.GA6154@dominikbrodowski.de> References: <20040111211255.GA31105@dominikbrodowski.de> <20040112154611.GH14031@poupinou.org> Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="===============57847486923559766==" Return-path: In-Reply-To: <20040112154611.GH14031@poupinou.org> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: cpufreq-bounces+glkc-cpufreq=gmane.org@www.linux.org.uk To: Ducrot Bruno Cc: acpi-devel@lists.sourceforge.net, cpufreq@www.linux.org.uk List-Id: linux-acpi@vger.kernel.org --===============57847486923559766== Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="J/dobhs11T7y2rNN" Content-Disposition: inline --J/dobhs11T7y2rNN Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Mon, Jan 12, 2004 at 04:46:11PM +0100, Ducrot Bruno wrote: > On Sun, Jan 11, 2004 at 10:12:55PM +0100, Dominik Brodowski wrote: > > [Len, could you test and verify this patch, and push it to Linus, pleas= e?] > >=20 > > The current algorithm used by Linux ACPI for passive thermal management= has > > two shortcomings: >=20 > ... >=20 > > +/* If a passive cooling situation is detected, primarily CPUfreq is us= ed, as it > > + * offers (in most cases) voltage scaling in addition to frequency sca= ling, and > > + * thus a cubic (instead of linear) reduction of energy. Also, we allo= w for > > + * _any_ cpufreq driver and not only the acpi-cpufreq driver. > > + */ >=20 > Just a stupid question: >=20 > What is best if processor heat issues (apart turning on the fan)? >=20 > Reducing voltage of the processor, but still allowing it to run execution > at 100% (which is the case if the processor is heating), or reduce > amount of time allowed for the processor to execute? voltage scaling. It offers a much better (quadratic) saving than clock modulation (linear saving). Doing both [and you need to do it, as the CPU won't run with fewer volts at the same frequency] gives you cubic savings. Dominik --J/dobhs11T7y2rNN Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iD8DBQFAAtvKZ8MDCHJbN8YRAsQhAJ45hhFDMKiU02RqV/vuBIrbL/51oACfUNoK BLgQ9ImjnuvyQT8gS/pNHEI= =VjvN -----END PGP SIGNATURE----- --J/dobhs11T7y2rNN-- --===============57847486923559766== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ Cpufreq mailing list Cpufreq@www.linux.org.uk http://www.linux.org.uk/mailman/listinfo/cpufreq --===============57847486923559766==--