Phoronix IRC Log: 2009-05-31
michaellarabel: uncle_fungus: did you get the new gromacs 4 running?
uncle_fungus: michaellarabel: I'm fixing up the gromacs 4 profile now
michaellarabel: uncle_fungus: Okay, thanks
uncle_fungus: the test now takes an age to compile, even longer than version 3 methinks
michaellarabel: is there any options to disable unused parts of gromacs from being built?
uncle_fungus: I'll have a look, but I'm pretty sure I can cut down the gromacs part by a factor of 2 due to some changes in the way parallelization works
uncle_fungus: michaellarabel: patch for gromacs 4 http://www.phoronix.com/forums/showthread.php?p=77047#post77047
uncle_fungus: there may be more I can do to reduce the build time further, but this at least gets the profile working with gromacs 4
michaellarabel: uncle_fungus: Great, thanks. Ultimately will there be updates for the other tests or are those still up to dat?
uncle_fungus: mrbayes needs to be updated against the latest mpi
uncle_fungus: I need to check if the other tests are still current
uncle_fungus: ok, a number need updating, also I found some potential coolness. A GPU version of HMMER has been released. I think that would be the first CUDA test in PTS right?
maligor: what would be the point of a cuda test?
maligor: it's limited to nvidia platform
maligor: opencl benchmark would be much more meaningful
maligor: once they get that out
maligor: you'd essentially compare the nvidia unified shader arch with cuda
maligor: the results wouldn't really be all that surprising
maligor: or even interesting
michaellarabel: uncle_fungus: Yes that would be cool
uncle_fungus: depends what you're using the test metrics for
maligor: except maybe for finding the optimal cuda bang/buck
uncle_fungus: if I was interested in seeing how much faster a cuda implementation was that a n other x86 impletementation
mattmatteh: anyone here experience a kernel crash/reboot when enabling cpu freq scaling with amd k8 and nvidia ?
uncle_fungus: especially since there is an MPI version of the same algorithm
michaellarabel: uncle_fungus: New profile now in git
maligor: has anyone done any benchmarks using the open64 compiler for x86_64 that amd released?
michaellarabel: uncle_fungus: Actually, can you pull it and try it again? It doesn't seem to be working the new gromacs profile. At least when using Villin, Single Node, Double Precision.
uncle_fungus: I think I see the problem already, without even running it
uncle_fungus: part of the patch hasn't been applied
uncle_fungus: there's a crucial line missing in install.sh
michaellarabel: uncle_fungus: what line is that? Part of the patch had failed but I thought I got all of the lines
uncle_fungus: line 130 of the patch
uncle_fungus: +\$HOME/gromacs40_/bin/grompp_\$PRECISION\_MPI -nov 1>/dev/null 2>&1
uncle_fungus: sometime my patches won't apply cleanly because no matter how many times I try and git pull, it refuses to stay up-to-date with the repo
uncle_fungus: mafft profile update: http://www.phoronix.com/forums/showthread.php?p=77052#post77052
michaellarabel: hmmm, still failing
michaellarabel: uncle_fungus: can you email the mafft patch to michael [at] phoronix.com ? I think vBulletin is messing up some of the patches
uncle_fungus: I've just pulled it again and I'm recompiling
uncle_fungus: sure, ok
uncle_fungus: should be winging its way to you now
michaellarabel: Got it, thanks. Maffty applied cleanly too, so it was the forums messing it up.
michaellarabel: the new mafft profile is working great
uncle_fungus: excellent. It appears to be running appallingly slowly on my machine at the moment :(
uncle_fungus: lol, ok ignore me, gromacs is still compiling in the background
uncle_fungus: that time looks fine to me
michaellarabel: uncle_fungus: still building?
mattmatteh: anyone have problems with asus bios crashing ?
mattmatteh: like reseting itself at boot ?
Deanjo: Sure your battery is good?
Milyardo: Yeah, see if any jumpers came out or something as well
mattmatteh: Deanjo, i would thinkso, this board has always been flakey since i got it
mattmatteh: Milyardo, ^
mattmatteh: linux can run stable for at least a month, could the batter or jumper be an issue ?
mattmatteh: and since the bios just crashed several hours ago
mattmatteh: it reset the next boot
mattmatteh: then i had to enable the fan speed again, else stuck on high
mattmatteh: tried to save that and it crashed when did
mattmatteh: but retained the fan setting
mattmatteh: once in linux i am fine.
mattmatteh: but... i was recently trying to get cpu freq scaling working
mattmatteh: any time i enable that it crashes. this last time it totally killed the bios
mattmatteh: the last thing i saw in dmesg was
mattmatteh: Clocksourse tsc unstable (delta -233429967 ns)
mattmatteh: then it froze and took about a half hour of messing with the bios to get it to boot at all
mattmatteh: and to fix, i have to use the power switch on the back, the reset and power buttons on the front are useless
mattmatteh: i am really stumped to why cpu freq scaling would take out the bios and cause it to reset
mattmatteh: and i am stumped to how a bios can be that crappy
mattmatteh: and... forgot to mention
mattmatteh: is that twice after it crashed, when i powered on, it made a weird buzzing sound. no, its not a wire hitting the fan, checked that
mattmatteh: and thats what it sounded like
mattmatteh: wonder if was the speaker
mattmatteh: wonders why he is soo doomed with a POS, first gigabyte, now asus
mattmatteh: is it really that hard to make a working mainboard and bios ?
Deanjo: mattmatteh, what version of kernel are you using?
Deanjo: if your seeing "Clocksource tsc unstable " it sounds like your trying to use apic_pm which is a known trouble maker
Deanjo: Enable APIC 2.0 and HPET in your bios
Deanjo: As far as powering up issues I would really look @ your power supply