I just looked and it appears there are "14,126,882 units" left. I was running some on a Windows Ryzen box and they were taking about 12 hours per task. Presumably, be faster on Linux.
It might be worth it to see if there is a fairly easy way to multi-thread this CPU application after you get the new GPU application(s) up into the beta test.
Especially if you are still collecting data for that task.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Copyright © 2024 Einstein@Home. All rights reserved.
Tom M wrote: I just looked
)
I agree that it would nice to have
i dont think it would improve
)
i dont think it would improve task throughput to run 1x task on 16 cores, or 16x tasks on 1 core.
_________________________________________________________________________
Ian&Steve C. wrote: i dont
)
Good point. It seems to me the only way that multi-threading could speed up throughput would be if the threads have "overlapping" processes. Since I don't know the algorithm I don't even pretend to have an idea of an answer.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!