Following a short break in the Gray Scott Thursdays webinar series, the Laboratoire d’Annecy de Physique des Particules, in collaboration with the CC-FR Competence Centre, will host a webinar on GPU programming with Kokkos on 21 May 2026.
Kokkos is a modern C++ performance portability library designed to simplify the development of applications targeting diverse hardware architectures, including multicore CPUs and GPUs. By providing abstractions for parallel execution and memory management, Kokkos enables developers to write a single source code that can efficiently run on different accelerator platforms such as NVIDIA, AMD, and Intel GPUs.
This session will focus on using Kokkos for GPU computing and will introduce the key concepts required to port and optimize applications for accelerator-based systems. Participants will explore execution spaces, memory layouts, parallel kernels, and data movement strategies, while learning how Kokkos helps developers achieve portability and performance across heterogeneous architectures.
The training is intended for researchers, developers, and HPC users interested in modern C++ approaches for portable GPU programming and performance-oriented software development.