GPU with marked index greater than 10

[liqid_mark]

I'm having issues with a server with more than 10 GPUs. I'm getting WUs but anything over an index value greater than 10 is being assigned to GPU 1.

Code: Select all

22:15:29:WU11:FS11:Running FahCore: /usr/bin/FAHCoreWrapper /var/lib/fahclient/cores/cores.foldingathome.org/v7/lin/64bit/Core_22.fah/FahCore_22 -dir 11 -suffix 01 -version 705 -lifeline 4950 -checkpoint 
15 -gpu-vendor nvidia -opencl-platform 0 -opencl-device 10 -cuda-device 10 -gpu 10

does this

Code: Select all

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    1      9886      C   ...org/v7/lin/64bit/Core_22.fah/FahCore_22   323MiB |
+-----------------------------------------------------------------------------+

Is there anything I can do to resolve this? This server has 20 Quadro RTX 8000s.

[Joe_H]

This is a known issue with the folding client, not sure if there is a way around using just the client software. There was another posting about this problem within the last 2-3 days if I recall correctly.

[liqid_mark]

Thank you Joe_H, sorry for having a duplicate post!

[bruce]

There's a reason we still call ourself FoldingAtHome. Our goal has alwasy been to make use of home computers to solve big problems ... and I think we're succeeding. Nobody has 10 GPUs in their home computer. I have no doubts that there are many supercomputers around with lots of GPUs running in parallel, but we're not equiped to handle them.

[Joe_H]

No problem, some items are hard to search on the forum because they occur so frequently like "GPU" and "10". If I or another moderator happen to find it, we can either merge your post with the topic, or post a link.

[liqid_mark]

Thanks for the reply bruce, I appreciate it. I'll try to setup another server in the datacenter for this and split the GPUs between them. Thank you!

[_r2w_ben]

This is a known issue with the folding client, not sure if there is a way around using just the client software. There was another posting about this problem within the last 2-3 days if I recall correctly.

Here's the recent thread, which also has a link to the issue.