Welcome to the HIP for GPU training course

This course is delivered through a combination of presentations and reinforcement activities, like MCQ's and hands-on exercises via guided instructions.

Dedicated hardware has been ring-fenced to work on the exercises which are a key part of this course.

The estimated time for completion is approximately 6 hours.


What you will learn

  • Understand the core concepts of GPU programming with HIP.
  • Port applications from CUDA to HIP.
  • Allocate and manage GPU memory effectively.
  • Work with streams, synchronisation, and shared memory.
  • Write, launch, and optimise GPU kernels.
  • Develop efficient algorithms for GPU execution.
  • Explore optimisation techniques for AMD GPU architectures.
  • Use portable build systems for GPU-enabled applications.
  • Work with tools and libraries within the ROCm ecosystem.


Prerequisites

Last modified: Monday, 20 July 2026, 2:57 PM