AMD ROCm Programming Guide

AMD ROCm™ is a software stack, composed primarily of open-source software, that provides the tools for programming AMD Graphics Processing Units (GPUs), from low-level kernels to high-level end-user applications.

The AMD ROCm Programming Guide introduces the core concepts, APIs, and best practices for programming with ROCm and the HIP programming language. It provides hands-on guidance for writing GPU kernels, managing memory, optimizing performance, and integrating HIP with the broader AMD ROCm ecosystem of tools and libraries.

Getting started

• Introduction

• What is ROCm?

• What is HIP?

• Introduction to the HIP programming model

• Hardware implementation

• SAXPY Tutorial - Hello HIP

Core HIP programming

• Introduction to core HIP programming

• HIP kernel programming

• Using HIP runtime API

  • Initialization

  • Memory management

  • Error handling

  • Asynchronous concurrent execution

  • Cooperative groups

  • HIP graphs

  • Call stack

  • OpenGL interoperability

  • External resource interoperability

• GPU programming patterns

  • Two-dimensional kernels: matrix multiplication tutorial

  • Atomic operations: histogram tutorial

  • CPU-GPU cooperative computing: K-means clustering tutorial

  • Stencil operations: image convolution tutorial

  • Multi-kernel programming: breadth-first search tutorial

Performance optimization techniques

• Performance guidelines

• Optimizing performance

  • Highly parallel workload: image gamma correction

  • Fixed-size kernels: image gamma correction

  • Reduction

  • Tiling and reuse: matrix multiplication

  • Tiling and coalescing: matrix transpose

• Multi-GPU programming

ROCm platform

• ROCm install

  • Installation prerequisites

  • Quick start installation guide

  • Installation via native package manager

  • Post-installation instructions

Known issues are listed and can be reported on the on the AMD ROCm Programming Guide GitHub repository.

To contribute to the documentation, see Contributing to ROCm docs for contribution guidelines.

You can find licensing information on the Licensing page.