Hey everyone,
as part of my bachelor thesis, I am working on deploying an local LLM, but unfortunately I am having issues with my NVIDIA Driver.
When I type “nvidia-smi” I get the error message, that it couldnt communicate with the driver. I’ve found several solutions in different forums, but unfortunately none of them worked for me.
Ive deployed a VM in the vmWare Environment. The VM contains 4 Tesla T4s, each with 16 GB VRAM in a 16Q Grid.
The Ubuntu Version is 24.04 and the Kernel Version is 6.8
The installed NVIDIA Driver Version is 550.67
The NVIDIA Bug Report is a 2.3 GB File, so I cant upload it here, I will send a OneDrive Share Link.
I’ve also tried it before on a VM with Ubuntu 22.04.
I would be very happy, if someone could help me, because it would help me with my bachelor thesis
No, I don’t think so (at least not on the VM, it is installed on the ESXi Host. (I havent really worked with a linux machine yet).
I’ve seen that there is an Grid Driver, but couldn’t find much to install it. I installed the normal NVIDIA Driver.
On VM before with Ubuntu 22.04 I was able to run nvidia-smi and it also didn’t have the grid driver. But on that machine problems started after an apt update.
Could you help me to install the proper driver? I’ve searched on a few websites, but couldnt find much.
There were these instructions in vmWare, but commands are not working on Ubuntu
You first need to download the proper -grid client driver from the vgpu portal where you also acquired the vmware host driver. It is not publically available.
