Since a few days (at least I did not observe earlier) BOINC aborts these Work Units after (usually) 3 to 5 minutes computation time with a computation error. Other NVIDIA work units are processed regularly, as far as they get the chance to get a hit.
Someone any idea what to do or check? Or is it possible to block just these Work Units?
TIA,
Roelof
Copyright © 2024 Einstein@Home. All rights reserved.
I see this error in stderr
)
I see this error in stderr for a recent one:
you are using a 4GB card to run OAS3. Look around here on the forum, but I think that is right at the edge, possibly WU dependent. You could just deselect O3AS in the Applications section of Project Preferences for the Location (aka venue) on which you have that machine running.
If I've guessed wrong, I hope someone more in the know will drop by and give better advice.
archae86 wrote: I see this
)
you’re right. 4GB is insufficient for these tasks when you factor in VRAM overhead by running the windows desktop. You might be able to get by with a 4GB GPU on a very lightweight Linux distribution or a headless CLI-only setup, but not windows.
_________________________________________________________________________
Many thanks, both of you! I
)
Many thanks, both of you! I managed to deselect the O3AS after a bit of searching. Now let's see if other tasks will pop-up again and run to completion.