Phoronix IRC Log: 2009-01-09
michaellarabel: ndim: Is PTS running alright for you now?
ndim: michaellarabel: Appeared to. I have aborted it, though, when it just flipped through some gtk drop-down list for 10 minutes. I got kind of bored. What is the run time for these tests supposed to be? Minutes? Hours? Days? Weeks?
ndim: Some hint on the approximate storage and runtime requirements when starting a benchmark or something similar would be really useful.
ndim: I'm well aware that some machine might take 10 hours, and another might take 0.3 hours for the same task. But some idea as to whether we are talking about minutes or weeks would be useful.
michaellarabel: ndim: Well, it all depends upon what you're choosing to run. Running universe-x is quite huge and takes a LONG time since there's so many tests. You'd probably be better off for your testing of just running say opengl-demos or gaming-free.
ndim: I guess. I'll try that later. Right now, the box is test-building stuff.
michaellarabel: ndim: Yep, that capability for run-time is being worked on. After you've run a test before, if you run: phoronix-test-suite list-test-usage it should start showing you average run-times. Then soon I'll integrate it so that it will show it on future runs about how long it will take.
ndim: Does not need to be accurate. A rough idea is sufficient.
Ivanovic: michaellarabel: http://bugs.gentoo.org/show_bug.cgi?id=216656
Ivanovic: michaellarabel: someone helpfull added a *huge* patch with changes to download urls
michaellarabel: Ivanovic: Thanks, will work on it in a minute.
Ivanovic: yes, several *nice* things
michaellarabel: ndim: Just committed support for reporting total sizes and estimated lengths, when running multiple things
Ivanovic: michaellarabel: please run a force update on the git...
michaellarabel: Ivanovic: Just did :)
Ivanovic: and now the update worked
michaellarabel: Ivanovic: He was using PTS 1.4 and I think in 1.6 I corrected some of those broken URLs already, but will get that patch cleaned up and merged soon.
Ivanovic: just make sure that they are fixed and then it should be good
Ivanovic: though the lib patch you should apply
Ivanovic: thugh change "/usr/include/AL/al.h, /usr/lib/libopenal.so.0 OR /usr/lib64/libopenal.so.1" (from the patch) to "/usr/include/AL/al.h, /usr/lib/libopenal.so.0 OR /usr/lib/libopenal.so.1 OR /usr/lib64/libopenal.so.1"
kaktuspalme: hi there
kaktuspalme: am i right with x264 decoding works only with hd4xxx graphic cards and not with hd3xxx cards?
Deanjo: Where do you get the idea that ATI cards do any x264 decoding in linux?
kaktuspalme: Deanjo: http://www.phoronix.com/scan.php?page=article&item=amd_xvmc_xvba&num=1
Deanjo: Ya, that's planned, but as of yet nothing has showed up
Deanjo: The only ones capable right now are the nvidia's with vdpau
kaktuspalme: hmm, damn, im planning a mediacenter pc
Deanjo: "The API for XvBA isn't published yet "
kaktuspalme: ok, then i have to buy a nvidia card :(
Deanjo: Ya it's the only option right now
Deanjo: MythTV for example takes full advantage of it right now. xine, ffmpeg, vlc etc are all working on vdpau support
kaktuspalme: ok thx
ndim: michaellarabel: Great!
kaktuspalme: i hope nvidia won't end linux support anytime
Deanjo: Very doubtful it will
Deanjo: They have been supporting linux for a long long time now
GNU\colossus: hmm, I can't seem to get arecord to record sound from my mic
GNU\colossus: although I'm able to set everything up in a way I can hear mic input looped back to my headphone output
GNU\colossus: anyone willing to walk me thru this? I never understood all that sound-related stuff very well :/
maligor: Deanjo, if you can call it support ;)
maligor: windows shows how bad binary drivers are inheritly
Deanjo: Hey it is support. Just because it get Foss fanatics panties all in a knot doesn't mean it's not supported
maligor: yeah... the kernel module that's larger than the kernel
Deanjo: maligor, the windows drivers work fine here as well, same with the os x drivers
maligor: sure, I'm not saying nvidia makes bad drivers, or atleast they don't appear so outside
maligor: it's just a bad trend
bkero: We wouldn't know though, that's the problem.
maligor: like the mess with creative
maligor: their windows drivers suck also tho
maligor: good thing about creative is that they give up eventually and release the sources
Deanjo: maligor, there are just as many issues with the opensource drivers though so it's not really any worse
maligor: 'the opensource drivers'?
Deanjo: for other cards
maligor: They rather got hit by the xfree86 problem too
Deanjo: again opensource drivers in no way guarantee that they will be good drivers
maligor: no, but they guarantee you can do something about it
Deanjo: it doesn't guarantee that anything will be done about it though
maligor: unless the hw is broken, they quite often don't infact cause a crash
bkero: That's not the point :/
maligor: points at Creative
GNU\colossus: proprietary drivers, otoh, guarantee you CANNIOT do anything about it
Deanjo: The difference is though maligor Creative always did have a shitty driver team
maligor: Deanjo, yes, but high profile company
maligor: the linux opensource drivers are fantastic
Deanjo: high profile in a small market
maligor: infact, they're almost the only player in sound cards
Deanjo: Fantastic? Basic 2 channel support is fantastic?
maligor: heh, now you refer to X-Fi
maligor: I meant Audigy/Live
maligor: the drivers are fully featured
Deanjo: There are plenty of other chipsets that work just as well
maligor: I never said there wasn't
maligor: The most common soundcard in a computer part shop is Creative
Deanjo: Besides audio is far less complex then video
maligor: Yes, nvidia has used that
maligor: I remember
Deanjo: It's true
maligor: but it's a bad reason not to release register specs
maligor: Cars are complex
maligor: ofcourse the problem here is that with cars you can just use eyes
maligor: with software you have the EULA that says you aren't allowed to use your eyes
maligor: mm, the ati 9.1 drivers have references to KONGM and CYPRESS codename chips, I wonder what they are
maligor: kinda interesting that they also refer to RAGE6 which is Radeon R100 I think
maligor: but cypress and kongm are pcie 2.0 according to this
michaellarabel: Ivanovic: Both patches have landed in master
mattmatteh: i updated a few packages using gentoo and noticed that rendering is kinda slow, i can see it draw from top to bottom. i was thinking dri might be messed up but 3d apps still work fine. could anyone suggest what to look for ?