ComfyUI Installer

Easily install ComfyUI (in a python venv) on Linux. Tested on Arch + AMD GPU.

Note: This is not the official ComfyUI icon.

Quick Start

You will need to have python, pyenv and pip on your system. Make sure to install for your GPU Vendor (AMD/Nvidia):

git clone https://github.com/HAMM3REXTREME/ComfyUI-Installer
cd ComfyUI-Installer
./install-amd.sh or ./install-nvidia.sh
./menu-entry.sh # Optionally add menu entry
./launch.sh

Installation

1. To install ComfyUI using this script, clone this repo and cd into it:
   git clone https://github.com/HAMM3REXTREME/ComfyUI-Installer && cd ComfyUI-Installer

2. After that's done, run the install script for your GPU vendor (AMD or Nvidia). This might take a while.
   To install for AMD: ./install-amd.sh
   To install for Nvidia: ./install-nvidia.sh
   Tip: You can optionally run ./menu-entry.sh in order to make a desktop menu entry.

3. When the install script has finished, you just need to copy/paste your models into their proper directories:
   Put your SD checkpoints (the huge ckpt/safetensors files) in: ComfyUI/models/checkpoints
   Put your VAE in: ComfyUI/models/vae

Once you've done that, launch ComfyUI using: ./launch.sh

Updating

Updating ComfyUI

Simply cd into the ComfyUI folder and run git pull:
cd ComfyUI && git pull

Upgrading python venv packages

Make sure to check out the actual ComfyUI repo for the most up to date information.

1. Run source venv/bin/activate to activate the Python virtual environment.
   (The installer creates a python venv named 'venv' in the base folder by default)

2. Upgrade torch, use the command for your GPU vendor (similar to the install script):
   For AMD: pip install --upgrade torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm6.2
   For Nvidia: pip install --upgrade torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu124
   Some version numbers might be out of date here.

3. Upgrade ComfyUI dependencies by running this command:
   pip install --upgrade -r ComfyUI/requirements.txt

After this you should have everything updated and can proceed to running ComfyUI.

Troubleshooting

If you get the "Torch not compiled with CUDA enabled" error:

Uninstall torch with:

pip uninstall torch

And install it again with the command (for Nvidia) above.

For AMD cards not officially supported by ROCm:

Try running it with this command if you have issues:

For 6700, 6600 and maybe other RDNA2 or older: HSA_OVERRIDE_GFX_VERSION=10.3.0 python main.py

For AMD 7600 and maybe other RDNA3 cards: HSA_OVERRIDE_GFX_VERSION=11.0.0 python main.py

You can add these changes (and other args you want) to launch.sh for convenience.