Advancing the capability of high performance applications means a faster rate of scientific discovery from the world’s most intricate models and simulations. While directive-based programming paradigms such as OpenMP are quick and easy to add to an existing application, it is often difficult to find the best optimization strategy. In this webinar, Florent Lebeau (Field Application Engineer at Arm), will demonstrate how by focussing on multi-threading and on-node optimization, developers can help maximize application efficiency on many-core High Performance Computing (HPC) systems.
Topics covered in the webinar include:
- Multi-threading to compute across all available cores and deliver best performance
- How to identify and minimize thread level synchronization
- How to place threads to maximize cache occupancy and memory bandwidth