Skip to content
This repository has been archived by the owner on Apr 24, 2022. It is now read-only.

improve CUDA device uniqueID by adding pciDomainID as prefix. #2101

Open
wants to merge 1 commit into
base: master
Choose a base branch
from

Conversation

hibagus
Copy link

@hibagus hibagus commented Jan 5, 2021

This is needed for multi-socket system where pciBusID and pciDeviceID can be the same on two different pci domains.
Now, the uniqueID will be the same as what nvidia-smi gives on Bus-Id

An example is below: GPU0 and GPU4 have the same pciDeviceID and pciBusID. They are connected to a different sockets, and thus they are different in pciDomainID. This means that there are more than one GPUs that have the same UniqueID and only one will be used for mining.

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.33.01    Driver Version: 440.33.01    CUDA Version: 10.2     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  XXXXXXXXX  On            | 00000004:04:00.0 Off |                    0 |
| N/A   25C    P0    34W / xxxW |      0MiB / 16160MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   1  XXXXXXXXX   On           | 00000004:05:00.0 Off |                    0 |
| N/A   27C    P0    35W / xxxW |      0MiB / 16160MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   2  XXXXXXXXX   On           | 00000035:03:00.0 Off |                    0 |
| N/A   23C    P0    34W / xxxW |      0MiB / 16160MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   3  XXXXXXXXX   On           | 00000035:04:00.0 Off |                    0 |
| N/A   28C    P0    36W / xxxW |      0MiB / 16160MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

Thanks!

…for multi-socket system where pciBusID and pciDeviceID can be the same on two different pci domains
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant