Owen, Thanks for your pointer to BLTK earlier. I completed my class project and BLTK was a big part of my success. I compared the relative power usage and performance of the "office" and "developer" task in BLTK under the performance, conservative, powersave, and ondemand governors. I also tested the "powernowd" userspace governor. I am more interested in desktop environments with frequency scaling CPUs than conserving laptop battery life. I found that ondemand and "powernowd" (with default options) performed equally well in both cases.
If you are interested in my results I'd be glad to send the PDF along. If you know the BLTK developers I'd be glad to send some bugs too :) Justin On Mon, May 5, 2008 at 5:14 PM, Kwon, Owen <[EMAIL PROTECTED]> wrote: > > Hello, > > We have conducted power measurement on Q1U with ondemand and performance > governors. I don't know about the Power and Performance project, but the > test result seems to use BLTK. > http://www.lesswatts.org/downloads/#bltk > BLTK is a kind of logger and workload. It collects battery capacity and > various speedstep related values from sysfs periodically. Also it > generates very intuitive graph of the values. I don't know this can be > of any help, but I think BLTK is a good starting point for the > measurement. > > Thanks, > Owen > > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Alan Jenkins > Sent: Friday, May 02, 2008 6:14 PM > To: Justin Bailey > Cc: [email protected] > Subject: Re: [Discuss] Power and Performance Measurement tests? > > Justin Bailey wrote: >> Greetings! >> >> As part of a class project, I am interested in measuring the power vs. >> performance of various cpufreq governors. On LessWatts.Org, there is a >> page about a Power and Performance project[1]. Are the tests used >> available for download somewhere? They make mention of a Kernel >> Performance Project[2], but I don't see where that project is >> measuring any power either. >> >> Any info is much appreciated! I will be sure to post my results back > here. >> >> Justin >> >> [1] http://www.lesswatts.org/projects/power-performance/ >> [2] http://kernel-perf.sourceforge.net/ >> >> _______________________________________________ >> Discuss mailing list >> [email protected] >> http://mail.lesswatts.org/mailman/listinfo/discuss >> > Powertop will report battery usage in Watts for laptops which implement > it (remember to unplug the power cable first!) > > It may be that other software deliberately doesn't include this, because > > it's designed to be used with a special power measuring device. That > presumably avoids issues with variable battery management > implementations, makes it easier to compare across different computers, > and allows desktops to be tested as well. Though it does mean you have > to test with the power cable _plugged in_, which can make a difference > on some laptops (when unplugged, some BIOS's will automatically switch > some hardware to low power mode; I'm thinking of something like enabling > > CPU C4 deep sleep, though that shouldn't affect raw power/performance > figuires). > > I don't know about the performance side of the tests though. > > Alan > > _______________________________________________ > Discuss mailing list > [email protected] > http://mail.lesswatts.org/mailman/listinfo/discuss > _______________________________________________ Discuss mailing list [email protected] http://mail.lesswatts.org/mailman/listinfo/discuss
