Advanced search

Message boards : Graphics cards (GPUs) : gpu tasks too large for my gpu, they never finish

Author Message
tumaru
Send message
Joined: 3 Apr 09
Posts: 2
Credit: 0
RAC: 0
Level

Scientific publications
wat
Message 17876 - Posted: 4 Jul 2010 | 21:37:30 UTC

I have finally gotten my gpu to work as it is supposed to work on boinc to find that the tasks sent to me are larger then it can handle by far. I was wondering if there was an error in the way you cut up the tasks or the way you tell which computers can handle what tasks based on performance.

PS am I just wasting power trying to complete these tasks or am I just getting no credits due to the fact that even if I run it 24/7 from start to finish with only it running it would still finish a long time afterwards.

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 17878 - Posted: 5 Jul 2010 | 8:59:34 UTC - in response to Message 17876.
Last modified: 5 Jul 2010 | 9:01:20 UTC

tumaru, I can confirm that your G102M is too slow to finish Work Units.
It only has 16 CUDA Cores. The entry level is about 48 CUDA Cores on that CC1.2 range of card. To put it into comparison, the best cards can finish a work unit in about 2hours, and a G102M would take about 4days if it ran 24/7 and the system was unused; which defeats the purpose of having a system. The cut-off period is 5days, as the next tasks depend of the quick completion of present tasks.
Also, I'm guessing it is a laptop. Generally it is not a good idea to run GPU WU on laptop systems, due to their structure and the heat generation by the GPU.
Sorry,

Post to thread

Message boards : Graphics cards (GPUs) : gpu tasks too large for my gpu, they never finish

//