It is too bad that when these uploads accumulate to the number of 2 x number of CPU cores, you won't be able to download any new work from Einstein anymore.
I still have a big amount of BRP7-WUs - I guess about 40% to 50% - that seem to stuck in some kind of way. GPU-utilisation is very low from the beginning and they have very long runtimes. The more runtime increases progress decreases. After 8 hours of crunching the WU seems to stop and show no more progress.
Hope this will be fixed in later versions of the BRP7-WUs. At the moment, they're a real pain in the ass. Because of the high quota of WUs affected, I need to abort the WUs manually to prevent my computers from idling. The high maintenance of these WUs is kind of stressfully kept in mind I knew Einstein as a "start-and-forget"-project.
While I agree that it’s an issue that should be looked into, there’s no reason to let it stress you out.
these tasks are in Beta. With no expectation of 100% success. The whole point of beta testing is to discover issues and bugs that can be reported back to the devs. if you aren’t ok with that, just disable beta task processing and you’ll only get the stable non-beta work.
I seem unable to get any new Meerkat tasks. I got one earlier today (just the one is waiting to be uploaded). I have tried to get additional tasks yet the server won't send me any. The messages send me to Host Log but I don't see a cause within that that would prevent me from downloading Meerkat tasks. Are we out of Win-Nvidia tasks and only have Linux and/or ati tasks and the one task I got was just dumb luck?
I still have a big amount of BRP7-WUs - I guess about 40% to 50% - that seem to stuck in some kind of way. GPU-utilisation is very low from the beginning and they have very long runtimes. The more runtime increases progress decreases. After 8 hours of crunching the WU seems to stop and show no more progress.
I've seen this with some of the MeerKAT tasks on my RX 570 host. I have FGRPBG1 and BRP7 running together at 3X. Usually, the M22 tasks take about an hour to complete. When one has been running over 1.5 hr, I have found that when I suspend and resume it, it will then go on to completion.
Ideas are not fixed, nor should they be; we live in model-dependent reality.
Yeah I’ve had maybe 10% of my
)
Yeah I’ve had maybe 10% of my tasks be ones that get “stuck” right at the beginning. They show running but never make any progress.
Im going to guess it’s a problem with individual work units and not the application. Since not all tasks seem to be affected.
_________________________________________________________________________
Got a few and also have the
)
Got a few and also have the upload problem. At least the tasks don't end in error on my Nvidia.
Anyway, with the 1 day deadline, will server keep sending out task to a new wingman once a day if the upload isn't fixed?
Bernd Machenschalk
)
By "old" wu's do you mean BRP7 tasks or other types of tasks that we could help get off the books.
It is too bad that when these
)
It is too bad that when these uploads accumulate to the number of 2 x number of CPU cores, you won't be able to download any new work from Einstein anymore.
As reported, no upload is
)
As reported, no upload is possible for brp7.
Also the first of my finished computed WUs expire in an hour today.
I still have a big amount of
)
I still have a big amount of BRP7-WUs - I guess about 40% to 50% - that seem to stuck in some kind of way. GPU-utilisation is very low from the beginning and they have very long runtimes. The more runtime increases progress decreases. After 8 hours of crunching the WU seems to stop and show no more progress.
Hope this will be fixed in later versions of the BRP7-WUs. At the moment, they're a real pain in the ass. Because of the high quota of WUs affected, I need to abort the WUs manually to prevent my computers from idling. The high maintenance of these WUs is kind of stressfully kept in mind I knew Einstein as a "start-and-forget"-project.
While I agree that it’s an
)
While I agree that it’s an issue that should be looked into, there’s no reason to let it stress you out.
these tasks are in Beta. With no expectation of 100% success. The whole point of beta testing is to discover issues and bugs that can be reported back to the devs. if you aren’t ok with that, just disable beta task processing and you’ll only get the stable non-beta work.
_________________________________________________________________________
I seem unable to get any new
)
I seem unable to get any new Meerkat tasks. I got one earlier today (just the one is waiting to be uploaded). I have tried to get additional tasks yet the server won't send me any. The messages send me to Host Log but I don't see a cause within that that would prevent me from downloading Meerkat tasks. Are we out of Win-Nvidia tasks and only have Linux and/or ati tasks and the one task I got was just dumb luck?
Ereignishorizont wrote: I
)
I've seen this with some of the MeerKAT tasks on my RX 570 host. I have FGRPBG1 and BRP7 running together at 3X. Usually, the M22 tasks take about an hour to complete. When one has been running over 1.5 hr, I have found that when I suspend and resume it, it will then go on to completion.
Ideas are not fixed, nor should they be; we live in model-dependent reality.
cecht wrote: When one has
)
Just what we needed to know. Give the tasks a rest break and they bounce back!
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)