michaellarabel: Deanjo any new bugs?
Deanjo: Just the usual yafray, espeak, wine tests that won't run on openSUSE 64-bit
Deanjo: yafray requires heavy patching, espeak uses portaudio18.h instead of portaudio19.h and wine won't compile without some enviroment settings.
Deanjo: There also seems to be a bug where having the same test run description will end up merging unrelated test results
michaellarabel: test description or test name
Deanjo: test description
Deanjo: tets run description
Deanjo: http://global.phoronix-test-suite.com/?k=profile&u=deanjo-26663-10440-25671
Deanjo: It told me there was a conflict with the names, gave me the option to change it, changed it but it still pulled in the results with the name clash
peppo: why is it so god damn hard to switch between fglrx and radeonhd?
peppo: I was trying to test an rt kernel for ffado, and I couldn't get rid of fglrx properly
peppo: read the howto on purging it, no luck.
peppo: x kept booting with corrupt graphics on top, rest black.
peppo: sigh
maligor: it shouldn't be a issue
maligor: might have some mesa weirdness tho
peppo: it's insane. I just spent 30 minutes going between kernels and recovery mode trying to get something to boot
peppo: fglrx doesn't compile with an -rt kernel, so that's not an option
peppo: and radeonhd simply won't display anything worthwhile after fglrx has infected my system
peppo: sure wish the ubuntu people had shipped with an -rt kernel that doesn't hardlock immediately on boot
maligor: heh
maligor: brb, I'll try what happens here
maligor: heh, yeah, it does that here too
maligor: it didn't use to
peppo: -rt?
maligor: normal
maligor: and not ubuntu but debian
peppo: you mean switching between fglrx and radeonhd?
maligor: yep
peppo: hrm
peppo: a terminal launched from the gnome menu shows Direct rendering: yes, but a terminal launched by using the shortcut key shows NO: (LIBGL_ALWAYS_INDIRECT set)
peppo: shouldn't have tried vo=xv, machine hardlocked
peppo: god I hate this hardware
xcdfgkjhgcv1: Can anyone tell me why Dell laptops always have touchpads slightly left of the centre position?
Milyardo: Becuase most people tend to touch the pad with their right hands
Milyardo: however leftys like myself find it very off putting
xcdfgkjhgcv1: Milyardo: How does it being slightly left of the centre position make it easier to use for right handers?
Milyardo: RIght handed users can reach for the pad without having to bend their wrist
njdoyle: I am using the AMD Catalyst drivers and whenever I stop X and then start it again my kernel panics. Does anyone else get this or know how I might fix it?
mattmatteh: njdoyle_,
njdoyle_: mattmatteh,
mattmatteh: njdoyle_, not sure if this is related, but noticed that dri had a bug on the opensource with gentoo. i looked at the bugs and there are several
njdoyle_: mattmatteh, I am using gentoo, do you know where the bug is?
mattmatteh: nj
mattmatteh: njdoyle_, no
mattmatteh: just read it within 2 dyas though
mattmatteh: njdoyle_, it seems if you have dri enabled, you will crash
mattmatteh: njdoyle_, sync and it might be disabled in the profile
njdoyle_: I'll check gentoo bugzilla again. DRI is needed for 3D isnt it?
mattmatteh: i do believe it is
mattmatteh: but crashing is worse
mattmatteh: :(
njdoyle_: no
njdoyle_: it only crashes when I restart X
mattmatteh: oh
njdoyle_: I need 3D
mattmatteh: njdoyle_, then dont sync
mattmatteh: and dont quit X
njdoyle_: it means I can't log out though :/
njdoyle_: sorta a pain
mattmatteh: njdoyle_, not sure how the problem is related, i am using the opensource drivers
njdoyle_: oh
mattmatteh: njdoyle_, but obviously the bug is in the kernel
njdoyle_: hard to say if they would be related
njdoyle_: there are lots of ways to panic a kernel
mattmatteh: njdoyle_, ill give it some time and hopefuly a kernel dev can fix it
njdoyle_: I've had this problem for a few months :\
mattmatteh: njdoyle_, months ?
mattmatteh: hmmmm
mattmatteh: i think i have had my laptop on for 40 days
mattmatteh: and dont think i restarted X
mattmatteh: till the other day
mattmatteh: i usually dont reboot that often
njdoyle_: I don't reboot often but I would like to restart X once and a while :\
njdoyle_: http://bugs.gentoo.org/show_bug.cgi?id=268136
njdoyle_: just people saying they have the bug though
njdoyle_: no solution or devs or anything
mattmatteh: thats not the bug i read
mattmatteh: seems there are going to be alot of duplicates
GNU\colossus: Deanjo: sorry to distrub/highlight you, but I just need a quick confirmation and you're the perfect source, I guess ;) AMD-V is available on geforce 8xxx chipsets given the CPU supports it, right?
michaellarabel: http://www.phoronix.com/scan.php?page=article&item=moblin_v2_ui&num=1
xcdfgkjhgcv1: Moblin can't do everything Ubuntu can.
Milyardo: Nice michaellarabel, I think I'm going to try out moblin now as well
xcdfgkjhgcv1: I use Ubuntu on my Eee PC 901.
xcdfgkjhgcv1: I use an encrypted home.
xcdfgkjhgcv1: Because I want to keep my data secure.
xcdfgkjhgcv1: Am I the only person unimpressed with Moblin V2?
xcdfgkjhgcv1: It doesn't look all the great to me.
xcdfgkjhgcv: When is UT3 coming to Linux?
mattmatteh: xcdfgkjhgcv, i am not familiar with moblin, is that an intel thing ?
bexamous: jxrendertest, part of PTS, has some issues
bexamous: Doing the same rectsFill test I can get variances between 150k to 200k/sec... so I look a program... it does this crazy ass thing to compute time
bexamous: tv.tv_sec*750 + tv.tv_usec/750
bexamous: either i don't understand how that is supposed to work or it makes no sense
bexamous: in any case I got rid of it and did tv.tv_sec*1000000+tv.tv_usec
bexamous: and now I get consistent results
bexamous: all withing a 1000ops/sec range
bexamous: its like 177760-178325 rather than 150000-200000
Kano: hi michaellarabel , do you have got an option to blacklist tests which are in a suite
Kano: like to blacklist vdrift that it never will run
Kano: or do you want to disalbe shaders, that would work too
michaellarabel: NO_VDRIFT_TEST=1 might be the environmental option to set. I don't recall what it is offhand. Or just: rm ~/.phoronix-test-suite/installed-tests/vdrift/vdrift
Kano: well thats a bit restricted to vdrift
Kano: how about a general solution
Kano: to blacklist checks in suites
michaellarabel: It is something like NO__TEST=1
Kano: like a new option in user-config
Kano: backlisted tests
Kano: or so
michaellarabel: why not: rm ~/.phoronix-test-suite/installed-tests//
Kano: does not help when you use benchmark
Kano: as it installs it
michaellarabel: only if you remove the pts-install.xml will it reinstall it
michaellarabel: if you remove just the executable, it should work
Kano: hmm, why not add a nicer blacklist?
michaellarabel: What other use cases are there for a nicer blacklist besides vdrift bug?
Kano: well blacklists for commerial tests
michaellarabel: In none of the mainline suites do I use a commercial test profile
Kano: is the env var check already in?
michaellarabel: should be for a while, IIRC
Kano: all uppercase even if your tests have got lowercase names?
michaellarabel: yep
michaellarabel: I will doublecheck in a minute that it's still in there
Kano: NO_VDRIFT_TEST=1 phoronix-test-suite benchmark vdrift
Kano: why does it ask then
Kano: i would like -> setting in user config -> if in blacklisted tests -> do nothing
Kano: maybe , seperated list like in other options
Kano: with options like
Kano: add/remove-backlist for pts
Kano: and list-backlist
Kano: blacklist of course
Deanjo: GNU\colossus, yes AMD-V is fully supported on nvidia chipsets
Kano: thats a cpu thing
Kano: s939 cpus do not have got amd-v
Deanjo: Yes it must be Pacifica or >
Kano: thats just another name
Kano: every am2/3 cpu from amd has it
Deanjo: Nope
Kano: but of course not the cpu i have got... s939
Kano: Deanjo: which am2 cpu has got no amd-v
Deanjo: The semprons for one
Kano: ok, well the usual athon/phenoms have it
Deanjo: Early AM2 X2's not of the F or G families did not support it either
Kano: well i still need somebody that buys my old amd x2 ;)
Deanjo: ie the windsor cores
Deanjo: like the Athlon 64 FX-62
Deanjo: Wait a minute, ya windsor did support it
Deanjo: Just the semprons don't
Kano: at least it is not that much restricted as intel does
Deanjo: Ya well when it comes to virtualization AMD pretty much has always had the upperhand
Kano: some oems will get even vt enabled E5200 cpus, but usually they want much more for vt...
Kano: i dont need it that often,but when i run 32 bit i still can use 64 bit in vbox is nice
Deanjo: Ya, vbox still can't run a 64-bit client though if no AMD-V or intel equiv is present
Deanjo: vmware can
Deanjo: But all be said. The BIOS still has to support AMD-V too. Believe it or not there are still MB's out there that do not have the option
Deanjo: lol, ouch http://global.phoronix-test-suite.com/?k=profile&u=grazz-23129-3969-328
Kano: well reiserfs was definitly faster than ext3
Kano: i know that, kompiling was faster on reiserfs
Kano: michaellarabel: you still do not check for results
Deanjo: heh, I could imaging how much slower it would be if those distro's enabled barriers by default
Kano: ./.libs/libsqlite3.so: undefined reference to `sqlite3BtreeCopyFile'
Kano: collect2: ld returned 1 exit status
Kano: make: *** [sqlite3] Fehler 1
Kano: but still tried to run it
Kano: http://paste.debian.net/36598/
Deanjo: hey is there a way to just generate a test profile that has all tests without having to enter them all one by one?
Kano: michaellarabel: missing depends check: /usr/lib/tclX8.3/help/tcl/intro/tclsh -> install tclx8.3
michaellarabel: Deanjo: phoronix-test-suite benchmark all
michaellarabel: all is a virtual test suite.
Kano: michaellarabel: for that sqlite test
Deanjo: OK but I don't want to actually run it yet though michaellarabel
Deanjo: I just want it to dump a xml
michaellarabel: Deanjo: For what purpose?
Kano: at least it rans now...
Deanjo: So I can weed out the tests that don't run
Kano: michaellarabel: why dont you check for error levels of install script and delete executeable when not 0
michaellarabel: Kano: I have been meaning to do that
Kano: as you see 1.8.1 does not have such a check
Kano: also it is hard to test for correct depends
Kano: you can only use a vm as snapshot
Kano: or a live image
Kano: and then you test if one SINGLE test runs
Kano: when you install lots of tests, then it is more likely that you isntall the needed depends
Kano: you would need to try on a clean system every benchmark
Kano: similar to what i do
Kano: i use pbuilder
Kano: instead of removeing depends
Kano: you can create a chroot via debootstrap
Kano: and run your tests there
Kano: just mount --bind proc,sys + dev
Kano: and resolv.conf
Kano: pbuilder works the same way
Kano: it extracts the chroot
Kano: then installs depends
Kano: by
Kano: e