ROCm/AMDMIGraphX

AMD's graph optimization engine.

What it solves

AMDMIGraphX is a graph inference engine designed to accelerate the execution of machine learning models on AMD hardware. It optimizes the process of running a trained model (inference) to ensure higher performance and efficiency.

How it works

It functions as an inference engine that leverages the ROCm ecosystem. It supports reading ONNX files (via Protobuf) and provides both C/C++ and Python APIs for integration. The engine utilizes several low-level AMD libraries, including MIOpen, rocBLAS, and HIP, to execute operations on the GPU.

Who it’s for

This tool is primarily for developers and machine learning engineers who need to deploy and optimize ML models for high-performance inference on AMD GPUs.

Highlights

  • Graph-based acceleration: Specifically designed to speed up ML model inference.
  • Broad integration: Works with the ROCm software stack.
  • Flexible APIs: Offers both Python bindings and C/C++ APIs for application development.
  • Multiple build options: Supports installation via binaries, CMake, or Docker containers.

Related

  • Project
    microsoft/onnxruntime

    A cross-platform machine learning accelerator that optimizes inference and training for models from various frameworks like PyTorch and TensorFlow.

  • Project
    rapidsai/cugraph

    cuGraph is an open‑source RAPIDS library that provides GPU‑accelerated graph algorithms with Python (NetworkX‑style) and low‑level C/C++ APIs, enabling fast, end‑to‑end graph analytics on the GPU.

  • Project
    InfiniTensor/InfiniTensor

    InfiniTensor is an open‑source C++ runtime that imports ONNX graphs, applies generic graph rewrites, and executes them on CPU or on custom accelerator back‑ends (InfiniOps/InfiniRT). It offers Python bindings, a test suite, and a modular design that lets hardware vendors plug in their own kernels.

  • Project
    amd/RyzenAI-SW

    A suite of tools and runtime libraries for optimizing and deploying AI inference on AMD Ryzen AI-based PCs to maximize hardware acceleration.

  • Project
    inisis/OnnxSlim

    OnnxSlim is a Python package that simplifies ONNX models by removing unnecessary operators, keeping accuracy while speeding up inference. Install via `pip install onnxslim`, then run `onnxslim input.onnx output.onnx` or use its `onnxslim.slim()` API. It’s integrated into many AI toolchains (TensorRT‑Model‑Optimizer, HuggingFace optimum, Ultralytics, etc.) and has reached millions of downloads.