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!
Been running only Meerkat x5 units for a few days now.
Once in a while there's a unit running running lonely with around 800s-900s being a bit off, they were validated.
Didn't see any of those on the screen myself.
I'm running BRP7 Meerkat test app.
Not running petris app.
Ends with error 700 (and 1008), seems to be cuda illegal memory acces?
[09:54:00][31629][ERROR] Error during CUDA device->host time series length transfer (error: 700)
[09:54:00][31629][ERROR] Demodulation failed (error: 1008)!
09:54:00 (31629): called boinc_finish(1008)
Don't know what really is happening there, lets post it in a Meerkat thread.
There is a bug and trouble section of the website. Perhaps open a thread there?
OBTW, is that gpu plugged directly into the MB?
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!
This system appears to be getting 1.6M on brp7 tasks running 2x?
If this could be scaled to a 7 or 8 GPU system we could be talking about up to 12.8M.
I wonder what he is doing differently?
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!
Understood. Is there anything 'extra' I need to do for the MPS to be active? Or is it automatic with the NVIDIA driver and the custom app running?
EDIT: I ran a test overnight of trying to run the app 2 at a time on my 3080. Somewhat surprisingly (at least to me since usage was at 100% with 1 running) the times doubled. So.... it didn't overall help but also didn't hurt for this app.
Thank you. That is food for
)
Thank you. That is food for thought.
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!
Been running only Meerkat x5
)
Been running only Meerkat x5 units for a few days now.
Once in a while there's a unit running running lonely with around 800s-900s being a bit off, they were validated.
Didn't see any of those on the screen myself.
Today I had a compute error without me tinkering:
https://einsteinathome.org/workunit/766701870
I'm running BRP7 Meerkat test app.
Not running petris app.
Ends with error 700 (and 1008), seems to be cuda illegal memory acces?
Don't know what really is happening there, lets post it in a Meerkat thread.
Greetings
Peter, There is a bug and
)
Peter,
There is a bug and trouble section of the website. Perhaps open a thread there?
OBTW, is that gpu plugged directly into the MB?
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!
That'd be a better place
)
That'd be a better place indeed, thx.
Card is normal plugged to MB, no pci4 issues with cables etc.
https://einsteinathome.org/ho
)
https://einsteinathome.org/host/6767055
This system appears to be getting 1.6M on brp7 tasks running 2x?
If this could be scaled to a 7 or 8 GPU system we could be talking about up to 12.8M.
I wonder what he is doing differently?
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
)
I don't think it is anything specifically unique except consistent operation. A good CPU and GPU.
It doesn't look like speeds of tasks are anything different from what is typical (although I can't see the wall clock time to run).
But, I could be wrong so mountkidd, correct me if I am!
Tom M
)
he's running CUDA MPS and some custom tuning. like I am. the 3080Ti does about 1.7M like this.
more PCIe bandwidth is required when you do this though.
_________________________________________________________________________
How do you run/use Cuda MPS?
)
How do you run/use Cuda MPS?
It only works on
)
It only works on Linux
https://docs.nvidia.com/deploy/mps/index.html
_________________________________________________________________________
Understood. Is there anything
)
Understood. Is there anything 'extra' I need to do for the MPS to be active? Or is it automatic with the NVIDIA driver and the custom app running?
EDIT: I ran a test overnight of trying to run the app 2 at a time on my 3080. Somewhat surprisingly (at least to me since usage was at 100% with 1 running) the times doubled. So.... it didn't overall help but also didn't hurt for this app.