Message boards :
Number crunching :
(unknown error) - exit code -5 (0xfffffffb)
Message board moderation
Author | Message |
---|---|
Send message Joined: 12 Jun 16 Posts: 3 Credit: 6,858,927 RAC: 0 |
Hello, more than 150 WU failed immediately (2-3 CPU seconds) during the last few days on one of my PCs with unknown error. Example: http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1951525908 May I kindly ask the project team / experts for support ? br Michael |
Send message Joined: 8 May 09 Posts: 3321 Credit: 520,500,528 RAC: 26,616 |
Hello, Error Code -5 means Boinc can't open the file, it could be on your end or that you never got the file it is looking for so couldn't open it. If everything is working now it probably wasn't on your end. The Error Code 0xfffffffb is explained here: http://serverfault.com/questions/197206/scheduled-task-last-run-result-0xfffffffb-works-when-running-from-command-p |
Send message Joined: 12 Jun 16 Posts: 3 Credit: 6,858,927 RAC: 0 |
Hello, which file are you referring to ? From the WU details I see the following: Found 1 CL device Device 'GeForce GTX 980 Ti' (NVIDIA Corporation:0x10de) (CL_DEVICE_TYPE_GPU) Board: Driver version: 378.49 Version: OpenCL 1.2 CUDA Compute capability: 5.2 Max compute units: 22 Clock frequency: 1240 Mhz Global mem size: 6442450944 Local mem size: 49152 Max const buf size: 65536 Double extension: cl_khr_fp64 Error creating context (-5): CL_OUT_OF_RESOURCES Error getting device and context (-5): CL_OUT_OF_RESOURCES Failed to calculate likelihood Not sure what this means, but potentially not a missing file ?! Thanks for any help in advance br Michael |
Send message Joined: 16 Mar 10 Posts: 210 Credit: 106,052,955 RAC: 24,017 |
Hello, It appears you had a driver update from version 376.33 (which shows up in your valid results from earlier on 26th January [UTC time]) to version 378.49 (which shows up in your invalid results from later on 26th January and thereafter) Your other machine listed here appears to be running an even earlier driver (version 372.90) and is not showing errors. I notice that at SETI@Home the "problem" machine is using the Intel GPU rather than your NVIDIA GPU, so that's not showing the same problem. However, I wonder if something in BOINC getting confused by having a pair of active GPUs that don't use the same driver for computing?... I'm not a Windows user myself, so I've no personal knowledge of the current state of NVIDIA drivers for Windows 10, or potential problems with multiple OpenCL providers and their [different?] drivers - the above is, therefore, a combination of observation and speculation. My only NVIDIA GPU is in a Linux laptop where it is the only OpenCL device enabled, and that has no problems (using a much older driver, though!) Hope you manage to get it sorted soon. Good luck - Al. |
Send message Joined: 8 May 09 Posts: 3321 Credit: 520,500,528 RAC: 26,616 |
Hello, That's the rub it doesn't say which file is missing! I would try rolling back the gpu driver to an earlier one that worked for you. |
Send message Joined: 12 Jun 16 Posts: 3 Credit: 6,858,927 RAC: 0 |
Hello, I am back to driver: 376.33 and WUs are calculated correctly since then. Thanks for the hint. The project team might want to look into this for general assessment as I assume most participants might update sooner than later. br Michael |
©2024 Astroinformatics Group