returning to the horde today

[shabazkilla]

It's been about a year since I last folded. I've got some new servers at work that I need to "stress test" and that's brought me back. Right now I've just got one dual processor/quad core 2.33ghz Proliant DL380, but the real fun should be in about 3 weeks. We should be getting in two quad processor/quad core 2.93ghz DL580s and a couple of single processor/quad core DL360s.

I installed the SMP client on the DL380 but it looks like it is only utilizing 4 cores. Do I just need a second installation and set machine numbers like back in the old days?

 

[Majic]

well wb man man.

looks like you are going to have some fun times with those servers. lol As for running 2 different smp clients for a quad no you dont they (the smp client) is actually setup for quads. i am sure if i got something wrong here one of these other guys will edit me faster then a Clinton wiki.

 

[shabazkilla]

Well with one install the SMP client is only using 1 processor (4 cores). The other 4 cores are idle. I did a 2nd install and all 8 core are being used, but now the second install keeps failing...

[22:29:35] Folding@Home Gromacs SMP Core
[22:29:35] Version 1.74 (March 10, 2007)
[22:29:35]
[22:29:35] Preparing to commence simulation
[22:29:35] - Ensuring status. Please wait.
[22:29:52] - Looking at optimizations...
[22:29:52] - Working with standard loops on this execution.
[22:29:52] - Previous termination of core was improper.
[22:29:52] - Going to use standard loops.
[22:29:52] - Files status OK
[22:29:53] tarting from initial work pa- Starting from initial work packet
[22:29:53]
[22:29:53] Project: 26Entering M.D.
[22:29:53] ne 493, Gen 19)
[22:29:53]
[22:29:53] Entering M.D.
[22:29:53] ne 493, Gen 19)
[22:29:53]
[22:29:53] Entering M.D.
[22:29:59] rotein
[22:29:59] : Protein
[22:29:59] Writing local files
[22:30:00] boost OK.
[22:30:00] Writing local files
[22:30:00] Completed 0 out of 1000000 steps (0 percent)
[22:39:57] Writing local files
[22:39:57] Completed 10000 out of 1000000 steps (1 percent)
[22:49:52] Writing local files
[22:49:52] Completed 20000 out of 1000000 steps (2 percent)
[22:50:25] delete work/wudata_02.xtc
[22:50:25] - Failed to delete work/wudata_02.bed
[22:50:25] - Failed to delete work/wudata_02.sas
[22:50:25] - Failed to delete work/wudata_02.goe
[22:50:25] Warning: check for stray files
[22:50:25]
[22:50:25] Folding@home Core Shutdown: EARLY_UNIT_END
[22:50:25] Finalizing output
[23:15:30] CoreStatus = 1 (1)
[23:15:30] Client-core communications error: ERROR 0x1
[23:15:30] Deleting current work unit & continuing...
[23:17:50] - Preparing to get new work unit...


The first install is running fine. Any ideas?

 

[Tigerbiten]

Check you temps.
SMP folding on a dual-quad will stress all 8 cores to the max so temps may start to be a factor.

If its a p2652 then that protien can be a little buggy as well.
You will find that after an early end you will download the same work unit two more times to see if you can finish it before you download a compleatly new work unit.

If you have any results from the early end left orphaned in the work folder, you can try running qfix to see if it will repair the queue.dat file so you can send them in.

Edit:- Be warned.
Shutting one client down on a dual-quad may well cause the other client to crash so you loose its work.
So make a copy of both clients just before you shut down so that if one does crash and you loose the work, you'll have a back up to swop in.

Luck ...........

 

[Pocatello]

hey tiger bit... how many dual quads do you have? just curious.

 

[Tigerbiten]

I've only 1x Dual-Quad.
Also 1x Dual-Dual, 1x Quad & 6x Dual.
Not a lot.

Luck ...........