Friends!
There are a lot of topics and uncertainties surrounding the performance of various GPUs in all E @ H applications.
An excellent solution would be the existence of a simple test, where for each of the applications there is test-wu. Now the big problem is to compare performance, since the same GPU can spend different amounts of time processing one wu.
I don't know if it is possible to package such a test in BOINC, but I think it won't be a big problem to make a separate application for those who understand this.
Thus, having a certain reference WU (it can be any valid one), it will be possible to have objective statistics for all GPUs.
This is always relevant, because over and over again each of us comes back to this issue. Especially those users who specifically equip computing power for E @ H or want to understand if a specific system really achieves optimal results.
Thanks!
Copyright © 2024 Einstein@Home. All rights reserved.
Andrew Petkin
)
They already do that on the test site so they don't need to test the tasks, if our pc's have problems with the tasks it's often more a problem on our side than the Projects side. Lately there were ALOT of problems with several different kinds of tasks and that was on the Project side that we users proved, they seem to have sent out new versions of the tasks having problems and they seem to be doing very well.