Hi,
I’m struggling for almost a week to run ML algorithm on GPU. I’m using Pytorch and constantly receiving error “AssertionError: Torch not compiled with CUDA enabled”. Also in Python
import torch
torch.cuda.is_available()
False
What is working set should be look like? CUDA 10.0 + cuDNN 7.5 should work but what about display driver and python version?
The other issue that I can’t uninstall completely CUDA (12.1) and Nvidia driver. I deleted it from Windows programs but still receiving the same versions when I use “nvidia-smi” command: “Driver Version: 531.61 CUDA Version: 12.1”
Thanks in advance
nvidia-smi is installed as part of the driver package and the Cuda version it displays is the version of Cuda that was used to compile both the driver and nvidia-smi.