The training event scheduled for 28 to 30 October 2025 targets scientists, researchers, and practitioners looking to enhance their skills in porting, optimizing, and profiling codes on AMD systems. Organized by AMD and NCC France, the course will be conducted online, enabling participants to access AMD infrastructures through the AMD Accelerator Cloud.
The event lasts for three half-days. The first day targets beginners, covering basics like HIP, ROCm, OpenMP offloading, and profiling. The second and third days delve into more advanced subjects such as advanced HIP/OpenMP programming, GPU-aware MPI, performance portability tools, AMD profiling utilities, and recent tools like the trace decoder and sophisticated profilers. Each session is interactive, allowing participants to practice what they learn in addition to understanding the theory.
This training targets researchers and engineers from academia and industry who possess some HPC experience. Participants should be familiar with the Linux command line, HPC software environments, and have the ability to compile and run code on HPC platforms. By the end, attendees will be able to optimize the use of AMD GPUs and CPUs for high-performance tasks, thereby enhancing their efficiency and productivity in scientific computing.