Many Ways to GPU Programming

In collaboration with NVIDIA and UCL, DiRAC is proud to introduce a new self-paced training course: ‘Many Ways to GPU’. 

Designed for researchers with some programming experience on HPC, this course offers a comprehensive introduction to GPU programming and provides the practical skills needed to transition research workflows from CPU-based to GPU-accelerated computing. 

 

By the end of the course, you will be able to: 

  • Understand what a GPU is and how it differs from a CPU 

  • Use NVIDIA’s powerful profiling tool, Nsight Systems 

  • Write basic C++ or Fortran programs using standard syntax 

  • Integrate OpenACC directives into C/C++ or Fortran code 

  • Apply OpenMP directives within C/C++ or Fortran code 

  • Develop CUDA kernels for use in C/C++ or Fortran applications 

  •  

The course includes: 

  • Interactive Jupyter notebooks featuring worked examples and hands-on exercises to build your confidence 

  • Built-in self-assessments to help you track your progress and reinforce key learning outcomes 

We hope this carefully designed course enhances your research and accelerates your journey into GPU development options.

Skill Level: Beginner