Cannot install GPU on Windows 2016 server (on GCP)

Question

I have created an instance on google cloud compute engine, running a Windows 2016 server with 1 x NVIDIA Tesla P100 GPU.

I manually installed the nvidia driver: 385.54-tesla-desktop-winserver2016-international.exe

When I verify with nvidia-smi.exe the powershell closes straight away, so I cannot read the report. Dxdiag also shows no GPU details.

I am trying to run a photogrammetry software (3DF Zephyr) and I get the error "opengl 2.0 not supported by your system". I assume I have not installed the GPU correctly. But what am I doing wrong?

Answer 1

I had trouble getting the P100 instances working on unbuntu GCP VMs. I finally tried updating the driver to the 390 series and it worked(!!)

For ubuntu, the process to update the driver is

sudo apt-get -y purge nvidia* sudo add-apt-repository ppa:graphics-drivers sudo apt-get -y update sudo apt-get -y install nvidia-390

Perhaps this is the issue in the windows VMs too.

(I did not have this problem with the K80 VMs)