Developing Applications with the AMD ROCm Ecosystem
Presented by AMD in collaboration with ENCCS, CSC and the LUMI Supercomputer.
This four half-day online workshop will cover how to develop and port applications to run on AMD GPU and CPU hardware on top AMD-powered supercomputers. Participants will learn about the ROCm software development languages, libraries, and tools. They will also get a developer’s view of the hardware that powers the system. This workshop will focus mostly on how to program applications to run on the GPU.
In this workshop participants will have the opportunity to use the newly installed LUMI-G system.
Prerequisites
Participants will be expected to have prior experience developing HPC applications, and some understanding of recent HPC computer hardware and the Linux operating system.
Agenda
all times CET
Nov 29, 2022, 13:00-17:00 — Programming Environments
- The ROCm Software Development Ecosystem
- HIP (George Markomanolis, AMD)
- Hipify — CUDA to HIP
- Getting Started with OpenMP® Offload Applications on AMD Accelerators (Jose Noudohouenou, AMD)
- Developing Fortran Applications, HIPFort & Flang (Joe Schoonover, Fluid Numerics)
Nov 30, 2022, 13:00-17:00 — Understanding the Hardware
- The AMD MI250X GPUs
- The AMD EPYC Trento CPUs
- AMD Communication Fabrics
- Memory Systems
- Affinity — Placement, Ordering and Binding
- Roofline Model (Noah Wolfe, AMD)
Dec 1, 2022, 13:00-17:00 — Tools
- Debuggers — rocgdb
- Profilers — from rocprof to Omnitrace and beyond (Jonathan Madsen and Nick Curtis, AMD)
- Additional Tools and Resources
Dec 2, 2022, 13:00-17:00 — Special Topics
- Using OpenMP® (Michael Klemm, AMD)
- Tips and Tricks
Registration
Please register by following this link https://events.prace-ri.eu/event/1440/registrations/1066/
Contact
For any questions contact us at training@enccs.se