Quadro FX 5600 slow down


log in

Advanced search

Questions and Answers : GPU applications : Quadro FX 5600 slow down

Author Message
Nelson
Send message
Joined: 29 May 07
Posts: 3
Credit: 1,958,088
RAC: 0
United States
Message 930689 - Posted: 3 Sep 2009, 21:34:21 UTC

I just put this card on-line last night and it seemed to be working perfectly. It would finish a SETI CUDA project in minutes, but now it seems to be doing almost nothing and is at less than 1% after 25 minutes. Also, my whole machine seems to be lagging whenever I have BOINC manager open. I have rebooted and checked the temps using GPUz (max @ about 80c), but the problem still exists? any suggestions?

http://setiathome.berkeley.edu/workunit.php?wuid=499626048

latest drivers as of yesterday (190_something)

XP 64-bit

Profile Gundolf Jahn
Send message
Joined: 19 Sep 00
Posts: 3184
Credit: 361,386
RAC: 39
Germany
Message 930706 - Posted: 3 Sep 2009, 22:58:47 UTC - in response to Message 930689.

You probably now have a VLAR task (Very Low Angle Range).

You could enter "vlar" in the search box at the top of this page, or look at my answers in this thread. Look also for "rescheduling tool" in the same thread.

Gruß,
Gundolf
____________
Computer sind nicht alles im Leben. (Kleiner Scherz)

SETI@home classic workunits 3,758
SETI@home classic CPU time 66,520 hours

Nelson
Send message
Joined: 29 May 07
Posts: 3
Credit: 1,958,088
RAC: 0
United States
Message 930725 - Posted: 3 Sep 2009, 23:39:20 UTC - in response to Message 930706.

Yeah, after reading that, it sounds about right. So, is CUDA inefficient at running them, or is it just harder for any type of processor to run them? i.e. is this an operation that a CPU would better handle? Or is a GPU still the best option, it just takes a long time?
If they need to be done by GPU, do people bypass them because they give you less credit? Or is there some other reason that I am missing?
Also, is the stderr file readable while the project is still running, or is a printout made on completion? If it is readable, where can i find it?

Thanks for your help and quick response.

Profile Gundolf Jahn
Send message
Joined: 19 Sep 00
Posts: 3184
Credit: 361,386
RAC: 39
Germany
Message 930800 - Posted: 4 Sep 2009, 8:25:52 UTC - in response to Message 930725.

Yeah, after reading that, it sounds about right. So, is CUDA inefficient at running them, or is it just harder for any type of processor to run them? i.e. is this an operation that a CPU would better handle? Or is a GPU still the best option, it just takes a long time?

It seems that the CPU application deals better with VLAR.

If they need to be done by GPU, do people bypass them because they give you less credit? Or is there some other reason that I am missing?

Many let them being aborted automatically by the VLAR-kill application because they think it easier than running the rescheduler program.

Also, is the stderr file readable while the project is still running, or is a printout made on completion? If it is readable, where can i find it?

There's a "slots" directory beneath the BOINC data directory. For each running task, you find a slot directory in there, containing the stderr.txt file.

Gruß,
Gundolf

Nelson
Send message
Joined: 29 May 07
Posts: 3
Credit: 1,958,088
RAC: 0
United States
Message 930865 - Posted: 4 Sep 2009, 15:20:45 UTC - in response to Message 930800.

Thanks a lot, Gundolf! You've been a lot of help!

Questions and Answers : GPU applications : Quadro FX 5600 slow down

Copyright © 2014 University of California