Advanced search

Message boards : Graphics cards (GPUs) : Constant scheduler requests for more work

Author Message
chimmy
Send message
Joined: 24 Feb 09
Posts: 14
Credit: 1,261,660
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 7019 - Posted: 26 Feb 2009 | 21:42:50 UTC
Last modified: 26 Feb 2009 | 21:45:43 UTC

I signed up for GPUGrid a few days ago. I'm seeing constant scheduler requests for more work with the response 'per-CPU limit of 1 tasks, no jobs available'. Happens every 1-2 minutes.

Any idea why or how to stop this?

Running on dual core/dual proc (4 cores total) AMD with two 9600 GT's in NON-SLI.

Forgot to add, I have two WU's in progress and another two waiting to start.

Thanks,

Jim

Profile K1atOdessa
Send message
Joined: 25 Feb 08
Posts: 249
Credit: 407,628,877
RAC: 2,033,247
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 7025 - Posted: 27 Feb 2009 | 3:14:33 UTC - in response to Message 7019.

I think this has something to do with the change to the server side to get rid of the 24 hour backoff problem. Now it checks very frequently for new work.

I prefer this problem over the 24 hour backoff, and given you have 4 cores and 4 WU's you are right where you should be.

chimmy
Send message
Joined: 24 Feb 09
Posts: 14
Credit: 1,261,660
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 7028 - Posted: 27 Feb 2009 | 4:48:14 UTC - in response to Message 7025.

Okay cool. Just wanted to make sure I didn't have some screwed up config that was sending a bunch of traffic to the server. Not that it would be a lot but still :)

Is this fixed in a newer version? I'm on 6.5.0 now.

Thanks,

Jim

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 7029 - Posted: 27 Feb 2009 | 7:06:40 UTC - in response to Message 7028.
Last modified: 27 Feb 2009 | 7:08:17 UTC

Okay cool. Just wanted to make sure I didn't have some screwed up config that was sending a bunch of traffic to the server. Not that it would be a lot but still :)

Is this fixed in a newer version? I'm on 6.5.0 now.


You are getting appropriate messages, your client needs work, the server won't give you any because you have too many failures or you already have the most work you are allowed to queue ... as you return valid work the number will rise to the max of 4 ... in the mean time, as long as you have work in progress the only down side is that the request goes out and gets no satisfaction.

Post to thread

Message boards : Graphics cards (GPUs) : Constant scheduler requests for more work

//