Is someone more knowledgeable that me able to have a look at the results I have returned to see if they can work out why most are failing to validate?
https://einsteinathome.org/host/11700141/tasks
Thanks :)
Copyright © 2024 Einstein@Home. All rights reserved.
If you search or look around
)
If you search or look around on these forums you'll find lots of advice on this sort of case. I suggest you review some of the many times Gary Roberts has addressed this sort of question from other users.
There is more than one possible cause, but a unifying theme is that quite likely something in your system is being clocked faster than it can reliably get the right answer to these WUs as it is currently operating.
If you have cooling deficiency, such as a failed fan, a buildup of air flow blocking material such as dust, a deteriorated interface material between your GPU and the heat sink, or a configuration error which just means the fans are not trying very hard, that can be one reason--which is common and often top of the list to check.
If you are knowingly overclocking, in your situation it would be wise to fully remove the overclocking and run a while to see if it changes the error rate.
If you have removed suspicion of fixable cooling trouble, I, personally, in your case recommend further reducing clock rate--even if that means going below stock.
I base this recommendation on your comments and on the extremely high invalid rate shown in your task list. I did not review your stderr files, which just possibly might reveal something useful, but in this case are not likely to.
GPU is not overclocked and is
)
GPU is not overclocked and is watercooled so I dont think its that?
AS others have found out How
)
AS others have found out
How many work units are you running in parallel ?
Multiple GPU tasks on a FIJI can cause Invalids, try running just one at a time.
Only running one work unit at
)
Only running one work unit at a time.
I checked the recent results
)
I checked the recent results your computer send back to us. They contain a non number in a column where there should be a number. Instead I see #INF which usually means the GPU did some wrong calculations. In the BRP4 case we see a power that is too high to be plausible which also stems from the GPU having problems doing math. This started on August 10th so if you did a driver update before that I would say you should maybe revert to the previous one and see if that fixes it.
Ok thanks, I think I did do a
)
Ok thanks, I think I did do a driver update then, ill try a full uninstall and roll back