[FIXED] AssertionError: Torch not compiled with CUDA enabled, when trying to use Torch in conda

Issue

Using the https://pytorch.org/get-started/locally/#anaconda

>> python
>> conda install pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia
...
Python 3.8.18 (default, Sep 11 2023, 13:39:12) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on win32
Type "help", "copyright", "credits" or "license" for more information.
> import torch
> >>> import torch
> >>>
> >>> if torch.cuda.is_available():
> ...     print("CUDA is available.")
> ... else:
> ...     print("CUDA is not available.")
> ...
> CUDA is not available.

GTX 4020, Windows, Updated drivers.

AssertionError: Torch not compiled with CUDA enabled

Been stuck on this for a while now, any ideas?

Tried getting torch installed

Solution

Had to reinstall cuDNN:

• https://developer.nvidia.com/downloads/compute/cudnn/secure/8.9.5/local_installers/11.x/cudnn-windows-x86_64-8.9.5.29_cuda11-archive.zip/
• conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia

Answered By - dogface