Message boards :
Number crunching :
GPU task stuck - cannot process anymore GPU work
Message board moderation
Author | Message |
---|---|
David@home Send message Joined: 16 Jan 03 Posts: 755 Credit: 5,040,916 RAC: 28 |
I have an odd state: BOINC Manager reports GPU task as 100% progress but status of aborted. If I select this task the only option I have available is Suspend. Any ideas how to clear this task please as it is stopping any more GPU work. |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
I have an odd state: BOINC Manager reports GPU task as 100% progress but status of aborted. If I select this task the only option I have available is Suspend. Any ideas how to clear this task please as it is stopping any more GPU work. Exiting & restarting BOINC doesn't help? Manually clicking on Update? Grant Darwin NT |
David@home Send message Joined: 16 Jan 03 Posts: 755 Credit: 5,040,916 RAC: 28 |
Unfortunatley restarting BOINC does not help. Running update is a little odd, first the work unit disappears and then it reappears. Looks like I will have to do a project reset which would mean losing my stats in BOINC manager. Anybody have any other suggestions please? |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
And when suspended, there's still no option to abort? I'd suggest exiting BOINC (again), restart it, then post the messages from the Event log here. Generally you are only blocked from getting new work if current work is suspended, or if you have uploads & downloads accumulating. Are other WUs completing & being reported without issue? Grant Darwin NT |
David@home Send message Joined: 16 Jan 03 Posts: 755 Credit: 5,040,916 RAC: 28 |
Had a close look at the event log, I think BOINC Manager is down loading new GPU work one unit at a time and instantly aborting them which explains the effect I am seeing when doing an update where aborted disappears and then reappears. 07/09/2017 00:18:14 | SETI@home | Sending scheduler request: To fetch work. 07/09/2017 00:18:14 | SETI@home | Reporting 1 completed tasks 07/09/2017 00:18:14 | SETI@home | Requesting new tasks for NVIDIA GPU 07/09/2017 00:18:17 | SETI@home | Scheduler request completed: got 1 new tasks 07/09/2017 00:18:17 | SETI@home | [error] Missing coprocessor for task 16jn08ab.30762.83900.6.33.16_1; aborting What can be causing the missing coprocesser error message? |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
It seems you have 2 threads on this issue ... You moved account (data folder) to a different computer? https://setiathome.berkeley.edu/forum_thread.php?id=81898 |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
07/09/2017 00:18:17 | SETI@home | [error] Missing coprocessor for task 16jn08ab.30762.83900.6.33.16_1; aborting With your computer hidden, I can only offer wild guesses. If you're running Win10, it has a nasty habit of updating video drivers, with ones that don't work when it comes to GPU computing. Downloading and installing the current driver from the video card GPU manufacturer's web site (NVidia) should sort it out- as long as that is the type of card actually on that system. EDIT- posting the full log from when BOINC starts to when it first requests, and gets, and then errors out the work would be helpful. Grant Darwin NT |
David@home Send message Joined: 16 Jan 03 Posts: 755 Credit: 5,040,916 RAC: 28 |
It seems you have 2 threads on this issue ... You moved account (data folder) to a different computer? I tried to move to a different user account on the same computer, but that did not work. I wanted to run BOINC on a different user account (idea was it would be more secure) to my general user account but that did not work out as in Windows the GPU is only available to the user account on the console. I have since gone back to running BOINC on my main user account but now have a new problem. Don't think these are related as I dettached and reattached all projects in BOINC manager when I tried moving user accounts. |
David@home Send message Joined: 16 Jan 03 Posts: 755 Credit: 5,040,916 RAC: 28 |
07/09/2017 00:18:17 | SETI@home | [error] Missing coprocessor for task 16jn08ab.30762.83900.6.33.16_1; aborting Alas I am running Windows 10. I have suspended new work and once the current running tasks are finished I will try a system reboot to see if this clears things up. If that fails I will update the Nvidia driver. I think all those GPU aborting messages have now killed me for the day as now getting " SETI@home | This computer has finished a daily quota of 1 tasks" |
rob smith Send message Joined: 7 Mar 03 Posts: 22199 Credit: 416,307,556 RAC: 380 |
If Windows has decided to do an update then a re-boot will do nothing for your problem. Try, as has already been suggested, installing the drivers from the Nvidia website. Before doing this make sure to turn any windows driver update options OFF. For the installation do a "clean installation" which is hidden under the advanced option on one of the early pages. Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
David@home Send message Joined: 16 Jan 03 Posts: 755 Credit: 5,040,916 RAC: 28 |
If Windows has decided to do an update then a re-boot will do nothing for your problem. Try, as has already been suggested, installing the drivers from the Nvidia website. Before doing this make sure to turn any windows driver update options OFF. For the installation do a "clean installation" which is hidden under the advanced option on one of the early pages. Rebooting has cleared the GPU aborted work unit issue. Windows has not installed any updtaes since 29th AUgust and BOINC has been working fine up until last night. BOINC manager is now not downloading new GPU work units. 07/09/2017 10:17:58 | SETI@home | Requesting new tasks for CPU and NVIDIA GPU 07/09/2017 10:18:00 | SETI@home | Scheduler request completed: got 0 new tasks 07/09/2017 10:18:00 | SETI@home | No tasks sent 07/09/2017 10:18:00 | SETI@home | No tasks are available for AstroPulse v7 07/09/2017 10:18:00 | SETI@home | No tasks are available for SETI@home v8 07/09/2017 10:18:00 | SETI@home | Tasks for AMD/ATI GPU are available, but your preferences are set to not accept them 07/09/2017 10:18:00 | SETI@home | Tasks for Intel GPU are available, but your preferences are set to not accept them 07/09/2017 10:18:00 | SETI@home | This computer has finished a daily quota of 1 tasks 07/09/2017 10:18:00 | SETI@home | This computer has reached a limit on tasks in progress What do the line items in red mean please? I can see work units available on the server status web page. |
rob smith Send message Joined: 7 Mar 03 Posts: 22199 Credit: 416,307,556 RAC: 380 |
Basically the bits you have highlighted are the result of having had so many errors in the last few hours the servers have decided that your computer has a problem, so to prevent you trashing any more they have cut your ration to a very low number. If you have any tasks in progress (with your computer hidden we can't see the top level status of your work queues) as they complete you will find that you start to get new work. Additionally, in a few hours (I can't remember how many) the restriction will start to lift and more work will flow in your direction. Thanks for highlighting the test you are asking about, it makes life so much easier to see you problem - some folks just send a several hundred line dump and expect those trying to help to find the line in question! Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
Rebooting has cleared the GPU aborted work unit issue. Maybe. Won't know for sure till it downloads another WU & processes it. Grant Darwin NT |
David@home Send message Joined: 16 Jan 03 Posts: 755 Credit: 5,040,916 RAC: 28 |
Rebooting has cleared the GPU aborted work unit issue. Very true, I'll have to wait a while before SETI@home will send me any more GPU work units. Plenty of CPU work units to keep me going. I'll post an update when I get my next GPU work unit. |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Rebooting has cleared the GPU aborted work unit issue. If I am not mistaken, you will get one per day until 10 consecutive good WUs are returned. Unless you have a bunch in pending, in which case the first 10 that validate would do the trick as well. I think that's how it works. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
If I am not mistaken, you will get one per day until 10 consecutive good WUs are returned. Pretty sure the 10 valid WUs is for the runtime estimates when you first start crunching. For the limits due to errors, 1 valid WU gets your daily quota doubled to 2, another good WU gets that doubled to 4, and so on. Basically you have to clobber a lot of WUs to get down to that limit, but if you've got a few Pending, it doesn't take long to lift that limit if the work you do get is Validated. Grant Darwin NT |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
If I am not mistaken, you will get one per day until 10 consecutive good WUs are returned. Ahh....you might be correct. Been a long long time since I had to think about such things here. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
David@home Send message Joined: 16 Jan 03 Posts: 755 Credit: 5,040,916 RAC: 28 |
BOINC manager keeps downloading one new CPU workunit each time a CPU workunit finishes. Maybe it is just by chance they are CPU work units or is the scheduler clever enough to differenitate GPU and CPU work units. I.e. am I still being blocked for GPU work units? I have lots of CPU work units so I have put SETI@home into "do not allow new tasks" mode until a large number of CPU work units have been cleared. Maybe when I then allow work units again I will a GPU work unit? |
rob smith Send message Joined: 7 Mar 03 Posts: 22199 Credit: 416,307,556 RAC: 380 |
It doesn't always work like that :-( Yo may have to wait until tomorrow before you get any tasks for the GPU. If possible, watch the first few very carefully (I know that it is not always possible to do so due to other commitments. Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
I have lots of CPU work units so I have put SETI@home into "do not allow new tasks" mode until a large number of CPU work units have been cleared. Maybe when I then allow work units again I will a GPU work unit? Best just to let it return and receive work as it can get it. Generally the more you fiddle, the longer it takes things to settle down. Once again, having your computers hidden makes things difficult to help with. If you've got plenty of Pending or some Inconclusive work, as they Validate then the number of WUs you can receive will increase. Grant Darwin NT |
©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.