in nvidia-smi I have cuda 12.8.
From Pytorch, I have downloaded 12.6 One and I have the latest Nvidia drivers also. I am on Win 11 PC , intel chip v100 2x-32Gb
→ Also if somewhere in some env I install torch version 12.4 / 11.8 will my CUDA 12.8 mess things up?
→ Is cuDNN also required if I am working with LLMs?
Your locally installed CUDA toolkit won’t be used as PyTorch binaries ship with their own CUDA runtime. The driver requirements are explained here.
1 Like