AMD ROCm™ documentation

Applies to Linux and Windows

2024-07-22

4 min read time

Welcome to the ROCm docs home page! If you’re new to ROCm, you can review the following resources to learn more about our products and what we support:

• What is ROCm?

• Release notes

You can install ROCm on our Radeon™, Radeon™ PRO, and Instinct™ GPUs. If you’re using Radeon GPUs, we recommend reading the Radeon-specific ROCm documentation.

For hands-on applications, refer to our ROCm blogs site.

Our documentation is organized into the following categories:

• Linux

  • Quick start guide

  • Linux install guide

  • Package manager integration

  • Install Docker containers

  • ROCm & Spack

• Windows

  • Windows install guide

  • Application deployment guidelines

• Deep learning frameworks

  • PyTorch for ROCm

  • TensorFlow for ROCm

  • JAX for ROCm

• Compatibility matrix

• System requirements (Linux)

• System requirements (Windows)

• Third-party support

• User/kernel space

• Docker

• OpenMP

• Precision support

• ROCm on Radeon GPUs

• API libraries

  • Artificial intelligence

  • C++ primitives

  • Communication

  • Math

  • Random number generators

  • HIP runtime

• Tools

  • Development

  • Performance analysis

  • System

• Hardware specifications

• Using ROCm for AI

• Using ROCm for HPC

• Fine-tuning LLMs and inference optimization

• System optimization

  • AMD Instinct MI300X

  • AMD Instinct MI200

  • AMD Instinct MI100

  • AMD Instinct RDNA2

• AMD Instinct MI300X tuning guides

  • System tuning

  • Workload tuning

• System debugging

• GPU-enabled MPI

• Using compiler features

  • Using AddressSanitizer

  • Compiler disambiguation

  • OpenMP support in ROCm

• Setting the number of CUs

• GitHub examples

• GPU architecture

  • MI100

  • MI250

  • MI300

• GPU memory

• File structure (Linux FHS)

• GPU isolation techniques

• Using CMake

• ROCm & PCIe atomics

• Inception v3 with PyTorch

• Inference optimization with MIGraphX