14:04:57 <jskarvad> #startmeeting PowerManagement -- http://fedoraproject.org/wiki/SIGs/PowerManagement 14:04:57 <zodbot> Meeting started Wed May 4 14:04:57 2011 UTC. The chair is jskarvad. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:04:57 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:06:03 <jskarvad> hi 14:06:07 <jmarko> hi 14:06:12 <fenrus02> is there a new power mgmt guide coming soon for f15 ? :) 14:07:49 <jskarvad> fenrus02: currently I do not work on it, but I will look on it and add the request to our wiki (at least ;) 14:08:04 <jskarvad> good point 14:08:22 <fenrus02> thanks :) 14:08:28 <jskarvad> np 14:09:09 <jskarvad> now the news about the measurementes/benchmarking I have performed during the last week 14:10:25 <jskarvad> I tried to use our home-brewed automated testsuite to measure power consumption of various Fedoras on server 14:11:25 <jskarvad> The F13-F14 test is currently finished, but the F15 test is still runing - the delay was caused by some troubles during the provisioning process (I had to update my scripts ;) 14:11:30 <jskarvad> Current results are here: 14:11:35 <jskarvad> http://jskarvad.fedorapeople.org/pm-tests/ 14:12:07 <jskarvad> Currently there is only one server machine dedicated to PM regression testign. 14:12:25 <jskarvad> But I am planning to user more machines to cover more HW profiles. 14:13:26 <jskarvad> The testsuite is still under heavy development (more tests/benchmarks are added). 14:13:52 <jskarvad> btw: does' anybody know good benchmark for openldap? 14:14:07 <jskarvad> it would be good candidate to add 14:16:10 <jskarvad> There are also some inconsistencies it the above mentioned results and I am working on improvements - namely different kernels were used in the kernel rebuild testcase (because the kernel used for the F13 test didn't build on F14) 14:16:42 <jskarvad> and For the seqread test the F14 used HP Smart Array Controller (/dev/cciss/c0d0) but the F13 ATA (/dev/sda) so that's probably why the performance of raw disk access doesn't match 14:17:38 <jskarvad> Some more tests are scheduled: test of the rawhide, tests of F15 with the kernel rebuilt with the default upstream settings and F15 kernel rebuilt with my custom (may be optimized :) settings, so stay tuned, more results will come :) 14:20:13 <jskarvad> Next info: vpodzimek has been trying to corelate the scomes score (systemtap gathered data) and the real power consumption of the system (or relative delta). But so far 14:20:14 <jskarvad> this approach doesn't end in usable formula. 14:20:34 <jmarko> i found some paper about performance testing on open ldap 14:20:39 <jmarko> www.lizenzfrei.at/downloads/ldap-paper.pdf 14:21:07 <jskarvad> jmarko: thanks I will look on it 14:22:40 <jskarvad> my idea is to cover the most "common" services and use cases in the PM regression testing, so the PM (polling) regression in individual services could be easily catched 14:24:08 <jskarvad> Next info: there is planned F15 release party in the Brno office. I am planning to give there short talk with demo covering the powermanagement features/improvements in the F15. More info will come later. 14:27:04 <jskarvad> Next: there are reported some i915 PM regressions in the 2.6.38 kernel, which can result in dark screen after laptop resume or unability to resume at all (but not all i915 users seems to be affected): 14:27:38 <jskarvad> https://bugzilla.kernel.org/show_bug.cgi?id=32522 14:27:40 <jskarvad> https://bugzilla.kernel.org/show_bug.cgi?id=31522 14:28:58 <jskarvad> Currently I don't know more, but there are some similar bug reports in our bugzilla. 14:29:51 <jskarvad> That's all news from me for today. 14:32:59 <jskarvad> So any tips, comments, volunteers (e.g. for writing the pm mgmt guide ;) or anything else? 14:37:20 <jskarvad> OK, thanks, that's all for today 14:37:40 <jskarvad> More news next week 14:39:10 <jskarvad> #endmeeting