search

NickLucche / stable-diffusion-nvidia-docker

GPU-ready Dockerfile to run Stability.AI stable-diffusion model v2 with a simple web interface. Includes multi-GPUs support.
MIT License
359 stars 43 forks source link

CUDA error: Unknown error during inference #1

Open NickLucche opened 2 years ago

NickLucche commented 2 years ago

This is likely caused by an incorrect cuda version (e.g. nvidia-smi reports gpu driver 11.7, the container uses 11.3). Hotfix:

# get inside container
docker exec -it stable-diffusion bash
# upgrade pytorch-cuda package
conda install pytorch torchvision cudatoolkit=11.6 -c pytorch -c conda-forge
# exit container 
ctrl + D
# restart container
docker restart stable-diffusion
mchaker commented 2 years ago

I haven't encountered this. Is it still an issue?

NickLucche commented 2 years ago

I encountered this on a 2070super card, I'll verify that once again but I think this is still an issue that's kind of hard to fix, unless we let the user choose their own pytorch+cuda package.

mchaker commented 2 years ago

What are the full repro instructions? I get a 2080ti next week and can test this