Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
On Tue, 31 Oct 2006 23:25:46 -0500 Anthony DeRobertis [EMAIL PROTECTED] wrote: Peter Palfrader wrote: How come it never outputs p4_clockmod? I use that happily on several different machines. I can confirm I've tried it on a machine. The performance impact was substantial (at least if I allowed it to go to its lowest speed); the power consumption difference was immeasurable[0]. Have you actually measured the power savings? [0] Immeasurable as in the measurement equipment (an UPS with load indication, in this case) could not notice any difference. The same UPS does notice a substantial difference with K8 chips. Anecdotal confirmation - on my Celeron M 420 (1.6 GH), using p4_clockmod causes an enormous drop in perceived responsiveness, but no apparent gain in battery life; I get the same time to discharge either way. Celejar -- mailmin.sourceforge.net - remote access via secure (OpenPGP) email ssuds.sourceforge.net - A Simple Sudoku Solver and Generator -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
Anthony DeRobertis wrote: Peter Palfrader wrote: How come it never outputs p4_clockmod? I use that happily on several different machines. I can confirm I've tried it on a machine. The performance impact was substantial (at least if I allowed it to go to its lowest speed); the power consumption difference was immeasurable[0]. Have you actually measured the power savings? [0] Immeasurable as in the measurement equipment (an UPS with load indication, in this case) could not notice any difference. The same UPS does notice a substantial difference with K8 chips. From what I remember reading on some LKML threads a while ago, p4-clockmod is not that interesting because it only does throttling (divides the frequency only) while the important thing to do to reduce power consumption is to use C-states as other drivers/processors do (reduce both frequency and voltage). http://marc.theaimsgroup.com/?l=linux-kernelm=114072746504710w=2 http://marc.theaimsgroup.com/?l=linux-kernelm=114123706515874w=2 Brice -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
On Tue, 31 Oct 2006, Anthony DeRobertis wrote: Peter Palfrader wrote: How come it never outputs p4_clockmod? I use that happily on several different machines. I can confirm I've tried it on a machine. The performance impact was substantial (at least if I allowed it to go to its lowest speed); the power consumption difference was immeasurable[0]. Have you actually measured the power savings? With 2.8GHz IPMI claims the machine is using 88+96 Watts (two powersupplies), when I let ondemand take down the CPUs to 700MHz it's usually 76+88 Watts. Not a huge difference but it adds up. -- | .''`. ** Debian GNU/Linux ** Peter Palfrader | : :' : The universal http://www.palfrader.org/ | `. `' Operating System | `-http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
Peter Palfrader: How come it never outputs p4_clockmod? I use that happily on several different machines. It's disabled in Ubuntu with the following comment: # Disabled for now - the latency tends to be bad enough to make it # fairly pointless. # echo FREQDRIVER=p4-clockmod /etc/default/powernowd # to override this Maybe they're wrong? -- Pelle -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
Peter Palfrader wrote: How come it never outputs p4_clockmod? I use that happily on several different machines. I can confirm I've tried it on a machine. The performance impact was substantial (at least if I allowed it to go to its lowest speed); the power consumption difference was immeasurable[0]. Have you actually measured the power savings? [0] Immeasurable as in the measurement equipment (an UPS with load indication, in this case) could not notice any difference. The same UPS does notice a substantial difference with K8 chips. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
On Oct 30, Michael Biebl [EMAIL PROTECTED] wrote: Why do you want to make a separate package for such a tiny shell script? The overhead is just two much. I'd propose to include it into the powermgmt-base package. I fully agree, it's silly to create a package for a 2 KB shell script. -- ciao, Marco signature.asc Description: Digital signature
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
Hi, Marco d'Itri: On Oct 30, Michael Biebl [EMAIL PROTECTED] wrote: Why do you want to make a separate package for such a tiny shell script? The overhead is just two much. I'd propose to include it into the powermgmt-base package. I fully agree, it's silly to create a package for a 2 KB shell script. Chris, would you accept the cpufreq-detect script into powermgmt-base? -- Pelle -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
Per Olofsson wrote: Package: wnpp Severity: wishlist Owner: Per Olofsson [EMAIL PROTECTED] * Package name: cpufreq-detect Version : 0.1 Upstream Author : Per Olofsson [EMAIL PROTECTED] * License : GPL Programming Lang: shell Description : detect CPU frequency control driver This package will attempt to detect and load the appropriate CPU frequency control driver for your hardware. This is useful together with packages such as powernowd, which can dynamically adjust the CPU frequency in order to save power. You can download the package here: http://dsv.su.se/~pelle/tmp/cpufreq-detect/ It is based on the detection code from Ubuntu's powernowd package. Hi Per, one more comment: The shell script contains a bashism (==) at line 95, which is not posix compliant. Use a simple = instead. Cheers, Michael -- Why is it that all of the instruments seeking intelligent life in the universe are pointed away from Earth? signature.asc Description: OpenPGP digital signature
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
Michael Biebl: one more comment: The shell script contains a bashism (==) at line 95, which is not posix compliant. Use a simple = instead. Thanks. There were many bashisms in the original script from Ubuntu, and I apparently missed this one. -- Pelle -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
Per Olofsson wrote: Hi, Marco d'Itri: On Oct 30, Michael Biebl [EMAIL PROTECTED] wrote: Why do you want to make a separate package for such a tiny shell script? The overhead is just two much. I'd propose to include it into the powermgmt-base package. I fully agree, it's silly to create a package for a 2 KB shell script. Chris, would you accept the cpufreq-detect script into powermgmt-base? Probably yes, but I'd like to see it before I decide. Does this have to make it into etch? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
Chris Hanson: Probably yes, but I'd like to see it before I decide. Attached. I will probably change a few things though, it should probably by default try to load the module instead of merely outputting it. Does this have to make it into etch? Nope. Although it would be nice, of course. -- Pelle cpufreq-detect.sh Description: Bourne shell script
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
On Mon, 30 Oct 2006, Per Olofsson wrote: Chris Hanson: Probably yes, but I'd like to see it before I decide. Attached. I will probably change a few things though, it should probably by default try to load the module instead of merely outputting it. How come it never outputs p4_clockmod? I use that happily on several different machines. Peter -- | .''`. ** Debian GNU/Linux ** Peter Palfrader | : :' : The universal http://www.palfrader.org/ | `. `' Operating System | `-http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#396117: ITP: cpufreq-detect -- detect CPU frequency control driver
Package: wnpp Severity: wishlist Owner: Per Olofsson [EMAIL PROTECTED] * Package name: cpufreq-detect Version : 0.1 Upstream Author : Per Olofsson [EMAIL PROTECTED] * License : GPL Programming Lang: shell Description : detect CPU frequency control driver This package will attempt to detect and load the appropriate CPU frequency control driver for your hardware. This is useful together with packages such as powernowd, which can dynamically adjust the CPU frequency in order to save power. You can download the package here: http://dsv.su.se/~pelle/tmp/cpufreq-detect/ It is based on the detection code from Ubuntu's powernowd package. -- Pelle -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]