Message boards :
Number crunching :
My S@H Cuda 6.06 stops running at 00:00:21 seconds....
Message board moderation
Author | Message |
---|---|
Riker Send message Joined: 10 May 01 Posts: 2 Credit: 11,970,797 RAC: 12 |
I just discovered a quite serious problem - My Cuda Seti have stopped crunching - but it still seems active. I get the status running, then Cpu-time counts up to 00:19, freezes a second, the displays 00:21 and no work is done, a simple task keeps up running until i pause or abort it. I tried to leave it alone to see if it was crunching, but after 91:21:12 i aborted it. THe machine has been running fine before. Does anybody have a clue ??? I'm running XP Pro SP2, Q6600 and a Gainward GF8600GT on that specific machine. All the astropulses running on CPU runs fine.... /Goran |
Byron S Goodgame Send message Joined: 16 Jan 06 Posts: 1145 Credit: 3,936,993 RAC: 0 |
Without seeing the tasks it's hard to say exactly what it is, but if it's a CUDA task, the most common reason for this type of behaviour is Very Low Angle Range, or VLAR ( angle ranges 0.1 and below), which CUDA is not able to run. If you could post a link to the task you're talking about or un-hide your pc, it would be more possible to verify. |
RandMan Send message Joined: 15 Sep 07 Posts: 1 Credit: 143,477 RAC: 0 |
I have the same problem - CUDA job stuck at 00:21. |
Byron S Goodgame Send message Joined: 16 Jan 06 Posts: 1145 Credit: 3,936,993 RAC: 0 |
I have the same problem - CUDA job stuck at 00:21. Of the MB tasks of yours that were aborted they all appear to be Very Low Angle Range. At this time aborting those tasks is about all you can do when running CUDA. You also have a number of AstroPulse tasks aborted too, but I'm not sure why those were a probmlrm for you. If they were stuck too, snoozing them and making sure you completely close down Boinc Manager and then restarting Boinc can sometimes get those going again. Sometimes even a reboot of the system after the previous suggestion fails, can do the trick too. |
eaglescouter Send message Joined: 28 Dec 02 Posts: 162 Credit: 42,012,553 RAC: 0 |
I aborted one of these 'frozen' CUDA work units, the system had completed no work for several hours :( It's not too many computers, it's a lack of circuit breakers for this room. But we can fix it :) |
Fusky* Send message Joined: 9 Apr 07 Posts: 120 Credit: 415,845 RAC: 0 |
same happened to me had a task running next day same task was still running,so all i did was stop accepting cuda tasks and aborted all cuda tasks i had in stock,so back to crunching regular units now |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
someone had written and posted a script to end task when Cuda locked up. It was on this forum but I can't find it now. In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
kevin6912 Send message Joined: 18 Jul 99 Posts: 17 Credit: 10,539,602 RAC: 0 |
|
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
thats the one In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
Krzychu P. Send message Joined: 2 Feb 06 Posts: 1 Credit: 116,830 RAC: 0 |
Hi! I had the same problem. after 18÷23sek. CUDA WUs stopped then I have used VBscript, it helped for ~6 units after script's start but then even this script didn't help so after 3 days of trials I had to reset the project in the manager and reinstall all the graphic drivers For now it seems to be ok. But I wonder, what bug it was... |
Byron S Goodgame Send message Joined: 16 Jan 06 Posts: 1145 Credit: 3,936,993 RAC: 0 |
Hi! There are many threads that discuss the bugs that are causing problems in CUDA. This post probably describes what bug or bugs were in the tasks you had as well. As well as the VBscript you are running you should also use the batch file mentioned in several threads like this thread which will allow you to spot VLAR tasks and abort them before they run . |
Maik Send message Joined: 15 May 99 Posts: 163 Credit: 9,208,555 RAC: 0 |
Byron S Goodgame wrote: As well as the VBscript you are running you should also use the batch file mentioned in several threads like this thread which will allow you to spot VLAR tasks before they run and abort them. Thats right. The script cant abort VeryLowAngleRange - WU's by clicking the button in Boinc-Manager. You must make that as it is explained in this thread. The script is helping you to avoid non-VLAR WU's that they get stuck and blocking your host ... And its written down often: You have to babysit seti-cuda. That means: Download some WU's (in cache), suspend computation, suspend networkactivity (better: set to no new tasks), check them for VLAR, abort VLAR, let them crunch the complete cache, resume networkactivitity, upload results, report results, start at the beginning at every time your cache is crunched (if you set NNT -> allow new tasks). kevin6912 wrote: VBScript thread link VBscript - It fights with CUDA app freezing Please do NOT use the script posted there, use this link ;) VBscript Fights Cuda - Thread @ lunatics |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.