This thread has always been useful because it helps reduce the proliferation of additional topic threads.
My question is I have been noticing one of my systems going into extreme back off for no apparent reason. Like 12 hours out. I don't have 12 hours of gamma ray #1 gpu tasks. So it is running dry.
I am experimenting with running 3 R5700 cards. And having some BSOD issues that I don't think are relevant here (I hope).
Any ideas about what is going on?
It is https://einsteinathome.org/host/12835716
Tom M
ps. If a moderator feels the urge to "consolidate" a bunch of low volume message threads into this message thread I won't object.
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.
I don't think a big thread of
)
I don't think a big thread of random posts will do any good. It buries information in a non-findable mess.
DanNeely wrote: I don't
)
It also confuses things if you don't quote everything and it's hard to find by the OP.
Tom M wrote: This thread has
)
Do you leave a cpu core or 2 or even 3 free for the gpu's to use?
mikey wrote: Tom M
)
This machine (the one that experienced that major back-off) has 1 cpu thread per gpu task. Back up to 9 gpu tasks as of this morning. I also run Boinc Manager at 87.5% used cpu cores to keep 2 cpu threads clear so the cpu isn't over committed.
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!
Tom M wrote: mikey
)
One problem could be that you are counting your HT cpu cores in that "1 cpu thread per gpu task" plus it looks like you are running cpu workunits as well. You could just be running out of resources on that 8 core cpu.
mikey
Anyone got in theories about
)
Anyone got in theories about why the bottom end of the top 50 list of computer systems is now showing RAC's under 1,000,000?
I just noticed because I finally got some things working reliably again so I can start driving my RAC's up again.
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!
Tom M wrote: Anyone got in
)
COVID-19 research?
GW on GPUs pay much less per
)
GW on GPUs pay much less per hour than Pulsars and GWs are relatively available.
I was looking at my E@H tasks
)
I was looking at my E@H tasks on the website for this machine: Moonshot
And noticed the tasks seem to be downloading in clumps. Earlier the log reported possible server down.
I have been running my buffer at 0.1 days on Gamma-Ray gpu tasks. I am beginning to wonder if I should open up the amount of work to keep from getting hit with what appears to be no tasks available (for one reason or another).
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!