search

Azure / hpcpack

The repo to track public issues for Microsoft HPC Pack product.
MIT License
29 stars 11 forks source link

GPU Information Win10 1909 #8

Closed thilbig closed 3 years ago

thilbig commented 3 years ago

We have upgraded a workstation node with a NVIDIA RTX2080 to Windows 1909. Since this upgrade the Cluster Manger of the HPC Pack 2016 U3 didn't show any GPU information about this node. We found out that Windows 10 1909 comes with an newer NVIDIA driver where the nvidia-smi.exe isn't in place where the HPC Pack search for it. It's now in C:\Windows\System32 and not in C:\Program Files\NVIDIA Corporation\NVSMI. Is there any way to fix this issue so that the HPC Pack software search for the nvidia-smi.exe to report the gpu to the Cluster Manager? A workaround is to uninstall the newer driver and install an older one.

Here is the output from nvidia-smi

C:\Users\admin-hpc>nvidia-smi.exe
Mon Nov 16 16:52:56 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 432.00       Driver Version: 432.00       CUDA Version: 10.1     |
|-------------------------------+----------------------+----------------------+
| GPU  Name            TCC/WDDM | Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce RTX 2080   WDDM  | 00000000:04:00.0  On |                  N/A |
| 30%   32C    P8    17W / 225W |    541MiB /  8192MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
thilbig commented 3 years ago

Got my answer here. I have to add a registry key ‘NvmlPath’ under HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\HPC with a string value for C:\Windows\System32, and then restart HpcManagement service.