Installation Instructions (AMD Gpu's)

Important

AMD support is more prone to issues than Nvidia. Please report any issues here on GitHub or on Discord!

Not every AMD-Card is supported! Check the AMD Compatibility Chart!

Warning

To use an AMD-GPU on Windows, you need WSL (Windows Subsystem for Linux).

Note

The scripts are tested on Ubuntu 24.04 WSL.

1. If you use Windows, install WSL, the compatible drivers and the WSL usecase, as described in the official WSL Documentation from AMD.

2. Download the newest release from the releases tab in this GitHub repository.

3. Unzip the contents of the release and place them where you want to run the server from.

4. Download any models you want to use from the models page and unzip and copy the contents (the config and models folders) into the AI Server folder created in the previous step.

5. Install Conda by running source ./install/install-conda-linux.sh

6. Run source ./install/install-linux.sh --amd

7. If you use WSL: Open a Powershell as Admin, and create a Proxy for the Port, where the server runs. This is only necessary, if stash does not run on the same PC as the AI Server. Use this Command: netsh interface portproxy add v4tov4 listenport=<YourPort> listenaddress=0.0.0.0 connectport=<YourPort> connectaddress=(wsl hostname --distribution <YourDistro> -I). For Distribution, use "ubuntu" if you use the default. You can find more information here.

8. (If using a patreon exclusive model) On the first time running the server, it will open a browser window to login with patreon to get your license. If there is no browser installed in your environment, it will give you a link to download a license. Save the license file to your models folder.

9. Proceed with installing the Cove extension(s).

10. To start the server again in the future, use source ./start.sh

Updating Instructions

To update from a previous version run (linux) source ./install/update.sh --amd