Advanced search

Message boards : Server and website : reached daily quota of 2 results

Author Message
HillFamily
Send message
Joined: 2 Mar 09
Posts: 6
Credit: 584,337,082
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 7975 - Posted: 29 Mar 2009 | 22:14:03 UTC

Hi,
Since about Mar. 24 I've flatlined on GPU credit, and the log says:

Message from server: No work sent
Message from server: (reached daily quota of 2 results)

my computer ID is 28054. Can someone advise how I can go about figuring out what happened?

Thanks.

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 7,369,744
RAC: 0
Level
Ser
Scientific publications
watwatwatwatwatwatwat
Message 7978 - Posted: 30 Mar 2009 | 1:03:58 UTC - in response to Message 7975.
Last modified: 30 Mar 2009 | 1:13:31 UTC

Hi,
Since about Mar. 24 I've flatlined on GPU credit, and the log says:

Message from server: No work sent
Message from server: (reached daily quota of 2 results)

my computer ID is 28054. Can someone advise how I can go about figuring out what happened?

Thanks.



This is the result page for that computer: http://www.gpugrid.net/results.php?hostid=28054.

You'll notice that all the WUs errored out. BOINC will reduce the max number of daily WUs per day when you return an error, and increase it when you return a success. It does this to prevent a race condition where a client computer is instantly returning an error -- if the server kept supplying WUs, the client would chew up thousands of WUs every day without doing anything useful. To prevent this, the server limits the number of WUs it will send to computers that are always returning errors.

If you fix the problem on your computer and start returning successful results, the server will start supplying more WUs to you each day. I think the standard algorithm is that the number of WUs per day per core starts at 100, and id decremented by 1 for every error returned, and incremented by 4 for every success returned. Once the problem is fixed and you start returning successful results, almost immediately you'll be back up to having a full daily workload.

Edit: I'm not an expert on GPU errors, but perhaps rebooting your computer might clear up the error.
____________
Want to find one of the largest known primes? Try PrimeGrid. Or help cure disease at WCG.

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 7982 - Posted: 30 Mar 2009 | 2:01:08 UTC

You may need to do a cold start with powering off the computer.

Errors like this can occur if you crash some tasks on SETI@Home. Are you running them?

HillFamily
Send message
Joined: 2 Mar 09
Posts: 6
Credit: 584,337,082
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 7985 - Posted: 30 Mar 2009 | 2:12:44 UTC - in response to Message 7978.

Thank you for the information...I got into an error condition that looked like some kind of GPU SNAFU (graphics all stretched and computer was unresponsive), but I bet I didn't go all the way to a cold start.

I am also running SETI. I'll stop that for a while and see if GPUGrid recovers.

Thank you very much!


Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 7987 - Posted: 30 Mar 2009 | 2:50:12 UTC

Sometimes a cold start is the only way to force a complete reset of hardware.

And, if the image is distorted, that is a real sign that a restart is required.

HillFamily
Send message
Joined: 2 Mar 09
Posts: 6
Credit: 584,337,082
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 8048 - Posted: 2 Apr 2009 | 3:44:03 UTC - in response to Message 7987.

Hmm, I still seem to be getting client errors even after numerous reboots.

It looks like I had at least one successful work unit though, but failures since:
http://www.gpugrid.net/results.php?hostid=28054.

Is there a special debug mode I could run the client in to get extra logging? Or is there some other way I could figure out what's going on?

Thanks for any advice.

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 8052 - Posted: 2 Apr 2009 | 5:05:36 UTC

Ok, first thing to check is temperature.

You can use Nvidia's tools or GPU-Z or others to find this out.

Second, the "best" driver that *I* know of is 181.22 (what I am using) ...

THird, it over-clocking ... slow down the OC

Fourth, when was the last time you cold started the system?

Fifth, are you running SaH or TLP on this system?

Post to thread

Message boards : Server and website : reached daily quota of 2 results