Advanced search

Message boards : Number crunching : "Error while computing" Task Status

Author Message
Reysic
Send message
Joined: 3 Apr 11
Posts: 11
Credit: 2,052,552
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 24243 - Posted: 4 Apr 2012 | 18:24:43 UTC
Last modified: 4 Apr 2012 | 18:26:58 UTC

The status of my tasks keeps showing up as "Error while computing". As far as I can tell, I've been unable to get any successful results with GPUGRID since I've started running BOINC again. I've tried BOINC versions 6.12.34 and 7.0.23. I receive this error on two different machines, one with two NVIDIA GTX 260s and the other with two NVIDIA GTX 460s. The NVIDIA driver version on both is 296.10. I'm not currently at home so I can't provide log information or set flags on the machines, but I will be soon if that would be useful. I'm running Windows 7 64-bit on both computers.

Any idea what could be going on? Just let me know if there's any other information you need. Thanks in advance.

Reysic
Send message
Joined: 3 Apr 11
Posts: 11
Credit: 2,052,552
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 24244 - Posted: 4 Apr 2012 | 18:32:32 UTC - in response to Message 24243.

Just to provide some additional info, I'm noticing that SETI tasks are reporting the same status. The online SETI logs state that no CUDA devices are being found and that crunching is falling back to the CPU. A computation error subsequently occurs.

Reysic
Send message
Joined: 3 Apr 11
Posts: 11
Credit: 2,052,552
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 24245 - Posted: 4 Apr 2012 | 18:53:21 UTC - in response to Message 24244.

Based upon what I'm seeing in other threads, it looks like the NVIDIA driver and monitor sleeping might be to blame. I'll try an older driver and post back with my results.

5pot
Send message
Joined: 8 Mar 12
Posts: 411
Credit: 2,083,882,218
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24246 - Posted: 4 Apr 2012 | 18:54:46 UTC

I was just going to post that. Try using the 285.63 I think, its a WHQL driver. Sounds like the issue.

Reysic
Send message
Joined: 3 Apr 11
Posts: 11
Credit: 2,052,552
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 24249 - Posted: 5 Apr 2012 | 1:46:51 UTC - in response to Message 24246.
Last modified: 5 Apr 2012 | 1:47:41 UTC

Thanks for the suggestion 5pot. The 285.63 driver has made a definite improvement as far as SETI tasks go (no more claims that CUDA devices cannot be found), but GPUGRID seems to be a different story. I'm still seeing computation errors. I'll post the stderr output available online for one of the failed tasks below.

<core_client_version>7.0.23</core_client_version>
<![CDATA[
<message>
The system cannot find the path specified. (0x3) - exit code 3 (0x3)
</message>
<stderr_txt>
# Using device 0
SWAN: FATAL : Unable to enumerate devices
Assertion failed: 0, file swanlib_nv.c, line 390

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.

</stderr_txt>
]]>

Any ideas? Thanks again.

Reysic
Send message
Joined: 3 Apr 11
Posts: 11
Credit: 2,052,552
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 24250 - Posted: 5 Apr 2012 | 1:52:37 UTC - in response to Message 24249.

Below is a slightly different version of the stderr output that is also popping up.

<core_client_version>7.0.23</core_client_version>
<![CDATA[
<message>
The system cannot find the path specified. (0x3) - exit code 3 (0x3)
</message>
<stderr_txt>
# Using device 1
# There are 2 devices supporting CUDA
# Device 0: "GeForce GTX 460"
# Clock rate: 1.44 GHz
# Total amount of global memory: 805306368 bytes
# Number of multiprocessors: 7
# Number of cores: 56
# Device 1: "GeForce GTX 460"
# Clock rate: 1.44 GHz
# Total amount of global memory: 805306368 bytes
# Number of multiprocessors: 7
# Number of cores: 56

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.

</stderr_txt>
]]>

5pot
Send message
Joined: 8 Mar 12
Posts: 411
Credit: 2,083,882,218
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24251 - Posted: 5 Apr 2012 | 2:13:06 UTC - in response to Message 24250.

You don't have SLI enabled do you?

Reysic
Send message
Joined: 3 Apr 11
Posts: 11
Credit: 2,052,552
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 24252 - Posted: 5 Apr 2012 | 2:37:01 UTC

Yes, I do have SLI enabled. Does SLI cause problems when crunching?

5pot
Send message
Joined: 8 Mar 12
Posts: 411
Credit: 2,083,882,218
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24253 - Posted: 5 Apr 2012 | 2:44:27 UTC

I would highly recommend disabling SLI, i think this may be your problem. Pretty.sure you can leave bridge connected if u game and would like to switch quickly. Computer basically recognize one gpu (in terms of the computer uses it. with SLI enabled. Depending on the coding of project, this may be why it works on seti and not here. [/b]

Reysic
Send message
Joined: 3 Apr 11
Posts: 11
Credit: 2,052,552
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 24254 - Posted: 5 Apr 2012 | 3:39:57 UTC - in response to Message 24253.

Thanks for your reply 5pot. Do you know if BOINC will still be able to see and use both GPUs with SLI disabled?

5pot
Send message
Joined: 8 Mar 12
Posts: 411
Credit: 2,083,882,218
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24255 - Posted: 5 Apr 2012 | 3:54:04 UTC

Someone a lot more knowledgeable than me will probably post tomorrow. I would just wait and see . I'm still waiting for post from message regarding Linux I made earlier today.

5pot
Send message
Joined: 8 Mar 12
Posts: 411
Credit: 2,083,882,218
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24256 - Posted: 5 Apr 2012 | 4:07:03 UTC
Last modified: 5 Apr 2012 | 4:09:23 UTC

Sorry its getting late. Thought that said enabled. Yes it should but as I said someone much more knowledge than me.should respond tomorrow[/b]. Things like different boinc version can also cause issues with multiple gpu but.since they're same.card.there should not be an issue i think. Disable SLI and give it a try

Profile dskagcommunity
Avatar
Send message
Joined: 28 Apr 11
Posts: 456
Credit: 817,865,789
RAC: 0
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24257 - Posted: 5 Apr 2012 | 6:09:16 UTC
Last modified: 5 Apr 2012 | 6:12:11 UTC

Yes it should work with both gpus, done that a time ago. If not (possible depends on boinc, drivers and so on) there are some threads about multiple gpu config changes.

Example http://www.gpugrid.net/forum_thread.php?id=2732#22419
____________
DSKAG Austria Research Team: http://www.research.dskag.at



Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24262 - Posted: 5 Apr 2012 | 15:32:31 UTC - in response to Message 24257.

I think all the answers to all those questions are in the Best configurations for GPUGRID FAQ thread.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Reysic
Send message
Joined: 3 Apr 11
Posts: 11
Credit: 2,052,552
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 24292 - Posted: 7 Apr 2012 | 2:18:56 UTC - in response to Message 24262.

Good news! Swapping to the 285.63 driver and disabling SLI seems to have stopped the "Error while computing" task statuses. Thanks everyone for your responses! :)

Post to thread

Message boards : Number crunching : "Error while computing" Task Status

//