Questions and Answers :
GPU applications :
Errors with optimised apps
Message board moderation
Author | Message |
---|---|
SmartWombat Send message Joined: 9 Jan 04 Posts: 64 Credit: 6,577,011 RAC: 0 |
Maybe this is my fault, for putting cudafft.dll in the Program Files\BOINC folder instead of in Documents and Settings\All Users\Application Data\BOINC\Projects\setiathome.berkeley.edu\ :( I corrected that today. http://setiathome.berkeley.edu/result.php?resultid=1312209686 Work Unit Info: ............... WU true angle range is : 0.433577 After app init: total GPU memory 536543232 free GPU memory 205757696 Cuda error 'cudaAcc_CalcChirpData_kernel2' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_CalcChirpData.cu' in line 106 : unspecified launch failure. CUFFT error in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_fft.cu' in line 139. http://setiathome.berkeley.edu/result.php?resultid=1312209475 Work Unit Info: ............... WU true angle range is : 0.415292 After app init: total GPU memory 536543232 free GPU memory 205757696 Cuda error 'cudaAcc_CalcChirpData_kernel2' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_CalcChirpData.cu' in line 106 : unspecified launch failure. CUFFT error in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_fft.cu' in line 139. http://setiathome.berkeley.edu/result.php?resultid=1312208701 Work Unit Info: ............... WU true angle range is : 0.439247 After app init: total GPU memory 536543232 free GPU memory 201956608 Cuda error 'cudaAcc_CalcChirpData_kernel2' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_CalcChirpData.cu' in line 106 : unspecified launch failure. CUFFT error in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_fft.cu' in line 139. http://setiathome.berkeley.edu/result.php?resultid=1312208283 Work Unit Info: ............... WU true angle range is : 0.445652 After app init: total GPU memory 536543232 free GPU memory 201956608 Cuda error 'cudaAcc_CalcChirpData_kernel2' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_CalcChirpData.cu' in line 106 : unspecified launch failure. CUFFT error in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_fft.cu' in line 139. PAul [IMG][/IMG] |
Leopoldo Send message Joined: 4 Aug 99 Posts: 102 Credit: 3,051,091 RAC: 0 |
After app init: total GPU memory 536543232 free GPU memory 205757696 It seems like your videocard has too little free videomemory for CUDA-processing... Did you try a reboot and saw the next completed result after the reboot? You should find what is eating your videomemory... For example, my Vista eats much less videomemory: Total GPU memory 805306368 free GPU memory 691068928 |
Fred J. Verster Send message Joined: 21 Apr 04 Posts: 3252 Credit: 31,903,643 RAC: 0 |
Hi, from my 8500GT :Total GPU memory 536870912 free GPU memory 453632000 After app init: total GPU memory 536870912 free GPU memory 360308736 and from my 9800GTX Device 1 : GeForce 9800 GTX/9800 GTX+ totalGlobalMem = 536543232 sharedMemPerBlock = 16384 regsPerBlock = 8192 warpSize = 32 memPitch = 262144 maxThreadsPerBlock = 512 clockRate = 1890000 totalConstMem = 65536 major = 1 minor = 1 textureAlignment = 256 deviceOverlap = 1 multiProcessorCount = 16 setiathome_CUDA: CUDA Device 1 specified, checking... Device 1: GeForce 9800 GTX/9800 GTX+ is okay SETI@home using CUDA accelerated device GeForce 9800 GTX/9800 GTX+ V12 modification by Raistmer Priority of worker thread rised successfully Priority of process adjusted successfully Total GPU memory 536543232 free GPU memory 491839488 After app init: total GPU memory 536543232 free GPU memory 489332736 RESULT WU |
Questor Send message Joined: 3 Sep 04 Posts: 471 Credit: 230,506,401 RAC: 157 |
You seem to be running 6.6.36. This version had a problem that under certain conditions when switching CUDA tasks the old one is left in CUDA memory which is probably why you sometimes see you have less free memory than you expect. Once you hit this condition then all the following tasks fail. The good news is this is fixed in 6.6.37 but this version is still only a test version so using it is at your own risk. GPU Users Group |
SmartWombat Send message Joined: 9 Jan 04 Posts: 64 Credit: 6,577,011 RAC: 0 |
Thanks, I'll live with this then until 6.6.37 (or later) is released. Optimised apps and the new CUDA libraries are making a terrific difference. PAul PAul [IMG][/IMG] |
Questor Send message Joined: 3 Sep 04 Posts: 471 Credit: 230,506,401 RAC: 157 |
I also noticed from your results that your are using the optimized app which includes the VLAR kill component - perhaps because you used the Lunatics unified installer? There is a version of the app now on Lunatics which does not include the kill aspect. CudaMB V12 non-VlarKill. Will require some manual tweaking to use but I'm happy to assist if you need any help. If you are getting a lot of errors and aborts you could run short of tasks to run. It might be worth looking at the Reschedule tool on the Lunatics site which allows you to move the VLAR tasks to your CPU for processing rather than aborting them. [VLAR tasks run very slowly on the Graphics card] John. GPU Users Group |
©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.