Advanced search

Message boards : Number crunching : Failed WUs: Out of GPU memory

Author Message
capeITLabs
Send message
Joined: 17 Nov 12
Posts: 30
Credit: 111,887,025
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwat
Message 34590 - Posted: 7 Jan 2014 | 8:21:19 UTC

Hi there,

at the moment, one of my machines has some problems with NATHAN WUs. They are running for days without end and I have to abort them manually. The stderr file shows "Out of GPU memory". That's interesting, since the card is a GTX480 with 1.5GB RAM. In the past there was no such problem with the NATHAN WUs. My other GPUGrid machine, equipped with two GTX560Ti, seems not to receive any NATHAN WUs.

How can I suppress those WU type ?

best regards,
Rene

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 7,520
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 34598 - Posted: 7 Jan 2014 | 22:55:18 UTC - in response to Message 34590.

at the moment, one of my machines has some problems with NATHAN WUs. They are running for days without end and I have to abort them manually. The stderr file shows "Out of GPU memory". That's interesting, since the card is a GTX480 with 1.5GB RAM.

It's interesting because these workunits consume only 670MB (GPU) RAM.
Is there any other GPU application running on this host?

In the past there was no such problem with the NATHAN WUs.

I don't have such problems with these workunits now, however I don't have any GTX 480s in my machines at the moment.
Maybe you should try to uninstall your old GPU driver, and install the latest one.

My other GPUGrid machine, equipped with two GTX560Ti, seems not to receive any NATHAN WUs.

That is only a matter of chance.

How can I suppress those WU type ?

You can't. You can only choose between the long and the short queue.

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,601,861,851
RAC: 8,767,822
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 34599 - Posted: 7 Jan 2014 | 23:10:45 UTC

Did you try rebooting the computer to free any 'stuck' memory?

capeITLabs
Send message
Joined: 17 Nov 12
Posts: 30
Credit: 111,887,025
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwat
Message 34600 - Posted: 8 Jan 2014 | 8:51:28 UTC - in response to Message 34598.

It's interesting because these workunits consume only 670MB (GPU) RAM.
Is there any other GPU application running on this host?

yes, there's also a GTX460 in this machine running PrimeGrid. Do you think this might cause these problems ?

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 7,520
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 34602 - Posted: 8 Jan 2014 | 13:24:35 UTC - in response to Message 34600.

It's interesting because these workunits consume only 670MB (GPU) RAM.
Is there any other GPU application running on this host?

yes, there's also a GTX460 in this machine running PrimeGrid. Do you think this might cause these problems ?

It's worth a try to stop it, and simplify your BOINC configuration.

capeITLabs
Send message
Joined: 17 Nov 12
Posts: 30
Credit: 111,887,025
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwat
Message 34604 - Posted: 8 Jan 2014 | 15:22:49 UTC - in response to Message 34602.

Hmmm...at the moment a NOELIA WU is running fine and this morning a SANTI was completed without error. Lets see what happens next... ;)

Post to thread

Message boards : Number crunching : Failed WUs: Out of GPU memory

//