ROCm/legacy-rocm-build

AMD ROCm™ Software - GitHub Home

What it solves

ROCm provides an open-source software stack that allows developers to program AMD GPUs for high-performance computing (HPC), AI, scientific computing, and computer-aided design (CAD). It bridges the gap between low-level GPU kernels and end-user applications, enabling GPU acceleration for complex workloads.

How it works

ROCm is powered by HIP, a C++ runtime API and kernel language that allows developers to write portable GPU code with an interface similar to NVIDIA CUDA. The stack includes a comprehensive suite of drivers, development tools, and APIs, including:

  • Math and Compute Libraries: Specialized libraries for linear algebra, FFTs, and random number generation (e.g., rocBLAS, rocFFT).
  • Runtimes and Compilers: Tools like LLVM and the ROCr Runtime to build and run GPU workloads.
  • Integration: Direct compatibility with major machine learning frameworks such as PyTorch and TensorFlow.
  • Tooling: Dedicated profiling, debugging, and monitoring tools (e.g., ROCgdb, AMD SMI).

Who it’s for

It is designed for developers and researchers working in AI training and inference, scientific computing, and high-performance computing who use AMD GPU hardware.

Highlights

  • CUDA-like Interface: Uses HIP to enable portable GPU programming.
  • Libraray Ecosystem: Includes a vast array of math, communication, and media libraries.
  • ** uma Framework Support**: Integrates with PyTorch, TensorFlow, and JAX.
  • Broad Hardware Support: Compatible with CDNA and RDNA architectures across various Linux distributions.

Related

  • Project
    ROCm/rocm-examples

    A collection of examples and tutorials for the ROCm platform, providing code for HIP runtime functionality, GPU-accelerated applications, and AI model execution.

  • Project
    JuliaGPU/AMDGPU.jl

    A Julia package that enables programming and executing computations on AMD GPUs using the ROCm platform.

  • Project
    ROCm/rocBLAS

    rocBLAS is a Basic Linear Algebra Subprograms library implemented in HIP and optimized for AMD GPUs to provide high-performance linear algebra operations.

  • Project
    ROCm/rocm-libraries

    A consolidated super-repository for ROCm libraries that streamlines the development, build, and integration of high-performance compute components for AMD GPUs.

  • Project
    ROCm/rocm-systems

    A consolidated super-repository for ROCm systems projects that streamlines the development, CI, and integration of libraries and tools required for building PyTorch.