This project provides a user-friendly installer and manager for Ollama installations optimized for AMD GPUs, leveraging the likelovewant/ollama-for-amd library. It targets users seeking a simplified setup and maintenance process for running large language models on AMD hardware, offering automated installation, ROCm library updates, and common error fixes.

How It Works

The tool offers a graphical interface to automate the installation and updating of Ollama with AMD-specific ROCm libraries. It intelligently detects Ollama installations, allows selection of GPU models for optimized library versions, and includes a one-click solution for the prevalent 0xc0000005 runtime error. Proxy support is integrated for users facing network restrictions.

Quick Start & Requirements

Installation is straightforward via a downloadable .exe from the Releases page or by building from source.

• Install: Download Ollama-For-AMD-Installer.exe from the Releases Page or git clone the repository and run pip install -r requirements.txt.
• Prerequisites: Python 3.10 or higher. Administrator privileges are required for installation.
• Docs/Demo: No explicit links provided in the README for docs or demos.

Highlighted Details

• Automated installation of the latest Ollama for AMD releases.
• Streamlined updates for ROCm libraries tailored to specific AMD GPU models.
• Built-in fix for the common 0xc0000005 runtime error.
• Flexible installation path detection and optional proxy support.

Maintenance & Community

The author is committed to maintaining the project but has transitioned to NVIDIA hardware. Development and testing now rely on an AMD APU (6800H) and community feedback, particularly for discrete GPU (dGPU) related issues. Contributions via issues or pull requests are actively welcomed.

Licensing & Compatibility

This project is licensed under the permissive MIT License, allowing for broad use, modification, and distribution, including within commercial or closed-source applications.

Limitations & Caveats

The author's primary development hardware is now NVIDIA, meaning testing on dedicated AMD GPUs is limited. The project's continued effectiveness for dGPU users is dependent on community contributions and feedback for issue reporting and resolution.