Phoronix IRC Log: 2009-11-03
bkero: http://www.phoronix.com/scan.php?page=news_item&px=NzY2NA Blobs are still bad, mm'kay
bkero: Same bad crack, twice the performance
Maxim1: hi people
Maxim1: is it just my feeling or does phoronix posts much more about AMD/ATI then Nvidia ?
Deanjo: lol, it is a pro ati site
bkero: Nouveau doesn't put out that many milestones/press releases.
bkero: and it's hard to know what nVidia's doing because their changelogs are often incomplete and it's completely closed source.
Kano: michaellarabel: did you test xvba-video yet
michaellarabel: I have in the past
michaellarabel: Deanjo: Looks like Jade is/was trying to make a comeback
Deanjo: Heh, that doesn't surprise me michaellarabel
michaellarabel: Posts already deleted as spam and blocked :)
Deanjo: So michaellarabel did you change to the 2.6.31-default kernel when installing openSUSE 11.2 for the tests?
Deanjo: or did you keep the preselection of the -desktop kernel?
michaellarabel: I used the kernel that ships with OpenSuSE by default
Deanjo: Well it ships with both. -desktop is used by default as most openSUSE users are using it in a desktop scenario
Deanjo: That explains alot actually
Deanjo: When running the server/workstation tests using the 2.6.31-default kernel is recommended
Deanjo: different kernel configs for different roles. The two separate kernels are new for 11.2
michaellarabel: maybe I'll run some benchmarks on them
Deanjo: Ya, particularly the apache tests and the likes
Deanjo: Timer on the -desktop kernel is 1000Hz and on the -default kernel it's 250
Deanjo: as well as some preempt differences
Deanjo: kernel-desktop flavor: disabled group scheduler and groups, disabled optimize for size, enabled full preemption, set HZ=1000
Deanjo: as far as the sqlite tests go, disabling barriers brings down the time drastically on openSUSE. SQLite doesn't like barriers on which is the openSUSE default