These all seem to be old ("Gn" as opposed to "G2") workunits that were created before ~Oct 22. Sorry, it might still happen with these, I didn't expect GPU results to come back that late. I'll see if I can find the time to update that validator, too.
The "sensitivity fix" apparently changes the run-time characteristics of the application so much that the workunit generator has trouble slicing eually-sized workunits, thus we get a pretty large variation in run-time per task. Sorry for that. We will take this into account in subsequent "sub-runs", but can't do much about it in the current "G2".
We'll shift the GPUs to a new applcation ("O2MDF") where they'll basically run on their own, i.e. not slowed down by validation by CPUs.
Validation for NVidia cards is almost perfect (invalids < 0.2%), but as some AMD cards are worse (invalids >5%), I'll leave the AMDs in Beta test, at least over the weekend, so their results will be validated by NVidias.
Here's
)
Another one snuck
)
Another one snuck thru.
https://einsteinathome.org/workunit/423138277
My host seems to get about 1
)
My host seems to get about 1 validate error a day.
https://einsteinathome.org/workunit/423296811
These all seem to be old
)
These all seem to be old ("Gn" as opposed to "G2") workunits that were created before ~Oct 22. Sorry, it might still happen with these, I didn't expect GPU results to come back that late. I'll see if I can find the time to update that validator, too.
BM
The "sensitivity fix" in the
)
The "sensitivity fix" apparently changes the run-time characteristics of the application so much that the workunit generator has trouble slicing eually-sized workunits, thus we get a pretty large variation in run-time per task. Sorry for that. We will take this into account in subsequent "sub-runs", but can't do much about it in the current "G2".
BM
https://einsteinathome.org/wo
)
https://einsteinathome.org/wo
)
Betreger
)
That's still an old workunit ("Gn").
BM
Betreger
)
That's a Gamma-Ray workunit (FGRP).
BM
We'll shift the GPUs to a new
)
We'll shift the GPUs to a new applcation ("O2MDF") where they'll basically run on their own, i.e. not slowed down by validation by CPUs.
Validation for NVidia cards is almost perfect (invalids < 0.2%), but as some AMD cards are worse (invalids >5%), I'll leave the AMDs in Beta test, at least over the weekend, so their results will be validated by NVidias.
BM