After using BOINC for E@H for a few months now, since the last few days I discover a task that does not make any progress any more (stuck at 98,609%), although its estimated completion time counts down just like the other tasks and eventually show the 'finished' state (3 dashes). I tried to stop and restart the taks, the system acknowledges my actions, but the task still seems is stuck. Anyting else I can try please? Thanks.
Copyright © 2024 Einstein@Home. All rights reserved.
E@H task is not completing
)
Exit Boinc, restart the pc and see what happens to the unit, I'm guessing it will jump back to a checkpoint and start crunching again from that point forwards. If not just abort it and move on, not all units can be crunched, some are just flawed. Aborting a unit just puts it back into the cache for some other pc to try and crunch.
RE: RE: After using BOINC
)
Thanks for the reply.
The stuck task was already there the day before; my attempts to trigger the task were taken the next day, hence after a system restart.
But...
In teh meantime, the task miraculously appeared to have finished after quite some time crunching the next day. And I discovered another stuck task later, which appeared to have disappeared also.
Now I think this type of tasks has a progress flow far different from the others (they had the same name prefix).
So next time I'll have to be more patient... ;-)
RE: RE: RE: After using
)
Yes you have some LONG crunching times sometimes, here is a short list of your cpu tasks times: 29,086.37, 29,019.71, 29,172.09, 28,597.88, 29,457.23, 50,837.60, 46,436.72, 56,109.31 etc. The key is they all seem to be validating and your getting credits for them, so yes I too would just persevere thru them.
I see you are also running gpu tasks too, do you leave a cpu core free just to keep the gpu fed?
RE: Yes you have some LONG
)
These times are probably measured between start and finish. This can be quite long, because my computer is powered on during a few hours a day only.
I have no idea. Things run as default configured by BOINC. Perhaps I may have enabled the usage of the GPU when I first installed it a few months ago. The only thing I observed is that GPU cruching starts a few minutes after I left the computer idle. Meanwhile all 8 cores are running other tasks. Should I dedicate one core to feed the GPU in my situation where the system is powered on and off quite often? (And sometimes I initiate the project update manually.)
RE: RE: Yes you have some
)
No problem.
Yes I would write down how long the current gpu units are taking, just use the valid units list from the website, then change it so one cpu core is free, ie only running 7 cpu units at a time, and see what the new time to completion is on your gpu units after 24 hours. Since you have an Nvidia gpu it may not make alot of difference, but it could and that's what you are looking for, doing more work faster.