![]() If this is something that needs to be debugged, feel free to point me to instrumented EXEs/DLLs and/or instructions to give you what you need. And that has slowed down my cracking into the top-400 users. Therefore, one of my best machines has been not been fulfilling its PPD capacity. If I OK that dialog, the WU either continues and finishes later, or continues and crashes again after a random amount of further progress, and if it crashes enough times, some restart threshold is hit and FAH terminates the WU. If so, I have that many crash dialogs up on my screen. Sometimes it is just 1 GPU, sometimes 2, sometimes 3. When the log shows this error, I can see a Windows dialog that indicates that FahClient_22.exe associated with that GPU has crashed. Lately, I have been randomly getting CUDA_ERROR_LAUNCH_FAILED on different GPUs at random WU progress points, for variety of WUs. (In comparison, many BOINC projects I ran in the past used to keep them at 99-100% 24/7) with FAH WUs usually keeping the GPUs busy anywhere between 60-90%, but never really hitting 90%+. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |