Performance Portability Training Series

May 2023 – Early 2024

This series, offered by OLCF and NERSC, features training sessions on various performance portable programming solutions to help ease developer transitions between current and emerging high-performance computing (HPC) systems.


Overview

For large-scale simulation, the portability of a codebase is becoming more important due to the variety of architectures being introduced in current and emerging high performance computing (HPC) systems. For example, the NERSC Perlmutter, OLCF Frontier, and ALCF Aurora feature NVIDIA-, AMD-, and Intel-based GPUs, respectively. Performance portable programming solutions have emerged to help ease developer transitions between such systems. Such solutions, for example, use a single interface to interact with multiple underlying programming models (e.g., CUDA, HIP, OpenMP, etc.). This training series aims to inform users on currently available performance portable programming solutions.


Sessions

SessionDate
Advanced SYCL Techniques & Best PracticesMay 23, 2023
HIP Training SeriesAugust - October 2023
Introduction to OpenMP Offload Part 1 : Basics of OffloadSeptember 29, 2023
Introduction to OpenMP Offload Part 2: Optimization and Data Management 2023October 6, 2023
RAJAOctober 10, 2023
Performance Portability for Next-Generation Heterogeneous SystemsFebruary 26, 2024
AMReX TutorialMarch 14, 2024
Kokkos TrainingApril 25-26, 2024
OpenMP Training SeriesMay - October, 2024
Other Solutions TBD

If you have any questions about this training series, please contact John Holmen (holmenjk@ornl.gov) for more information.