Message boards :
Number crunching :
What is Computation error (-5,) running MilkyWay@Home GPU tasks?
Message board moderation
Author | Message |
---|---|
Send message Joined: 3 May 10 Posts: 1 Credit: 22,583,933 RAC: 0 |
BoincTasks (efmer.com) History display Status column sometimes shows "Reported: Computation error (-5,)" running MilkyWay@Home GPU tasks on Zotac GTX980Ti GPUs (not overclocked) under Windows 10 and Nvidia driver 461.40. Fred at Boinctasks says it is not a BoincTasks error code and to check with the MilkyWay@home project. 1) What is this error code? 2) Is there a list of MW@h error codes? Thanks for your help. Art Brown |
Send message Joined: 9 Dec 11 Posts: 38 Credit: 1,497,896,956 RAC: 0 |
Having a peek at your tasks you seem to have several computation errors. Did you recently update your drivers (or did windoze decide you needed updated drivers)? Driver issues have been known to cause such problems. Especially if they are installed while you are crunching, as they basically remove the old drivers first which causes the card(s) to error out as they have no current driver installed. It could also be an aging card going to bed. I've heard tales of the 900 series cards going to bed in the last year or so. Either way, I'd recommend rebooting the system. If it was a driver update the new drivers will load on boot and you'll be back in business. I mean in actuality it could be many things, but let's start simple. |
Send message Joined: 8 May 09 Posts: 3319 Credit: 520,327,381 RAC: 21,204 |
BoincTasks (efmer.com) History display Status column sometimes shows "Reported: Computation error (-5,)" running MilkyWay@Home GPU tasks on Zotac GTX980Ti GPUs (not overclocked) under Windows 10 and Nvidia driver 461.40. I found this on the Net: "The computation error happens on the client, in the science application. This will also send back information about the error, just click on any of the task's resultIDs, which will show the stderr.txt that will show what is giving the error. Client (BOINC) errors are always three digits and negative, e.g. -227 Application errors are normally positive, can be negative, but then are +1, -1 or something like -1073528045." https://boinc.berkeley.edu/dev/forum_thread.php?id=7156 So it sounds like it's an MW error code and may mean something or nothing depending on which version of the app you are crunching, they just went thru a whole bunch of tasks having problems and MW even cancelling alot of tasks to try and fix it and make new ones. You may have exposed a new one in this version. In the end it means keep on crunching and ignoring them or go to another project until they figure things out and fix it. |
Send message Joined: 10 Apr 19 Posts: 408 Credit: 120,203,200 RAC: 0 |
Hello, Like Mikey said, could you please share the stderr.txt file associated with the workunit that produced the error? I took a quick peek through the source code for something that would throw that error but didn't see anything that jumped out at me. Best, Tom |
Send message Joined: 16 Mar 10 Posts: 210 Credit: 105,942,716 RAC: 24,833 |
Tom, I had a very brief look look at one or two of Art's failed jobs yesterday and noticed a common theme... Error creating context (-5): CL_OUT_OF_RESOURCES Error getting device and context (-5): CL_OUT_OF_RESOURCES Failed to calculate likelihood 09:16:48 (14100): called boinc_finish(-5) So it's not an error in the code per se! Your guess is as good as mine (or probably better!) as to why that happens. There are some recent tasks awaiting validation, and the error tasks had already managed to produce one or more of the results, then failed out on the third or fourth work unit so it's not as if the card can't do the work!... Happy hunting! Cheers - Al. |
©2024 Astroinformatics Group