ByronLeeeee/Ollama-For-AMD-Installer

This project simplifies the installation process of likelovewant's library, making it easier for users to manage and update their AMD GPU-compatible Ollama installations.

What it solves

This tool simplifies the installation and management of community-driven AMD GPU support for Ollama on Windows. It automates the complex process of injecting ROCm libraries into the official Ollama application to enable hardware acceleration for AMD graphics cards and APUs.

How it works

The installer provides a graphical user interface (GUI) that handles the following:

  • Hardware Detection: Automatically identifies the user's AMD GPU to select the correct ROCm architecture profile.
  • Automated Deployment: Downloads and pairs matched versions of the Ollama application and ROCm libraries (e.g., ROCm 7.1.1 or 6.4.2) to ensure compatibility.
  • Injection Process: Replaces system DLLs within the Ollama installation directory with community-provided AMD libraries.
  • Recovery and Safety: Creates backups of affected files before injection, allowing users to roll back changes if the installation fails.
  • Troubleshooting: Includes a one-click fix for common runtime errors (like 0xc0000005) and a Vulkan mode configuration utility.

Who it’s for

Windows users with AMD GPUs or APUs who want to run Ollama with hardware acceleration but find the manual installation of ROCm libraries difficult.

Highlights

  • Automated Workflow: Supports both full installation of matched app/library pairs or injecting libraries into an existing Ollama setup.
  • Smart Downloader: Includes a proxy tester and mirror selection to handle unstable network conditions during GitHub downloads.
  • Hardware-Aware: Built-in scanner to automatically detect AMD GPU profiles.
  • Safety Features: Automatic SHA-256 verification of downloads and a multi-level backup/restore system for injected files.

Related

  • Project
    stefankoegl/kdtree

    A simple Python implementation of a kd-tree for constructing, modifying, and searching multi-dimensional points, specifically for nearest-neighbor queries.

  • Project
    langchain4j/langchain4j-spring

    A collection of Spring Boot starters that simplify the integration of LangChain4j into Java applications, reducing boilerplate for AI capabilities.

  • Project
    KittenCN/stock_prediction

    A PyTorch-based experimental framework for stock time-series prediction supporting multiple deep learning architectures like Transformers and Diffusion models.

  • Project
    dorjeduck/llm.mojo

    A port of Andrej Karpathy's llm.c to the Mojo language, demonstrating that Mojo can achieve C-like performance for low-level LLM training.