Questions and Answers :
GPU applications :
Computation Errors - never had them before - why now?
Message board moderation
Author | Message |
---|---|
BigWaveSurfer Send message Joined: 29 Nov 01 Posts: 186 Credit: 36,311,381 RAC: 141 |
Running 6.6.28 with 185.85 on an EVGA NVIDIA 9600GT KO. Ever since i updated to 6.6.28 & 185.85 I am getting WU that are coming up with a "Computation Error", which I have never seen before. Why is this happening? Nothing else in the system has changed and the card runs around 68 degrees with fan at 50%. I found a WU that only reported 0.03 for credit and here was the error: Work Unit Info: ............... WU true angle range is : 7.314308 Optimal function choices: ----------------------------------------------------- name ----------------------------------------------------- v_BaseLineSmooth (no other) v_GetPowerSpectrum 0.00021 0.00000 v_ChirpData 0.01760 0.00000 v_Transpose4 0.01716 0.00000 FPU opt folding 0.00442 0.00000 Cuda error 'cudaMalloc((void**) &dev_outputposition' in file 'c:/sw/gpgpu/seti/seti_boinc/client/cuda/cudaAcceleration.cu' in line 366 : out of memory. setiathome_CUDA: CUDA runtime ERROR in device memory allocation (Step 1 of 3). Falling back to HOST CPU processing... SETI@Home Informational message -9 result_overflow NOTE: The number of results detected exceeds the storage space allocated. Flopcounter: 6991832299.295586 Spike count: 29 Pulse count: 0 Triplet count: 1 Gaussian count: 0 called boinc_finish </stderr_txt> ]]> Just curious if I need to be worried about the new errors, thanks! |
perryjay Send message Joined: 20 Aug 02 Posts: 3377 Credit: 20,676,751 RAC: 0 |
The one you've posted here is a -9 result overflow. That just means the WU was noisy and had too many things causing noise. If you don't get too many of those it's nothing to worry about. If you do start to get too many, open your machine and check for dust bunnies. It can be a sign of overheating. I also notice you are getting some -5 errors. You might want to check this thread http://setiathome.berkeley.edu/forum_thread.php?id=53790 about those. PROUD MEMBER OF Team Starfire World BOINC |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
The one you've posted here is a -9 result overflow.... It's also a Cuda error 'cudaMalloc((void**) &dev_outputposition' in file 'c:/sw/gpgpu/seti/seti_boinc/client/cuda/cudaAcceleration.cu' in line 366 : out of memory. With a (512MB) card, you shouldn't be getting that unless your cache is so over-inflated that newly downloaded 'shorty' tasks go immediately into EDF/High Priority. And since I see tasks "sent 4 May, reported 22 May" in your tasks list, I suspect it may be. |
BigWaveSurfer Send message Joined: 29 Nov 01 Posts: 186 Credit: 36,311,381 RAC: 141 |
My Cuda taks are all "Running". My AP 5.03 tasks all run in "High Priority" because seti/boinc has not been able to fix the bug in reporting time to completion with both the CPU/GPU taksks. I have a 10 Day work buffer, should I change it to 5 days? I do not see a problem as my system is working on tasks that are due on the 2nd of June now and go up to the 13th of June. I see maybe 1 to 3 of those errors a day, maybe I will shorted my buffer and see what happens. |
©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.