Improve HPC skills for the next generation of ocean models

This action will foresee to educate researchers to increase ther HPC skills for the next generation of ocean models developed and mantained in the framework of EDITO-Model Lab initiative. The objective of the action will be to guarantee the access to a set of consolidated existing numerical models for hydrodynamics and waves and the next generation of ocean model configurations optimised for European HPC hardware.
Different numerical models will be used with global and limited area configurations. Global, regional-to-coastal configurations, already tested and validated from researchers.The set of models make a better use of heterogeneous computing platforms.
Expensive model workloads (such as the computation of diagnostics) are able to be offloaded to accelerators (GPUs) to make elements of the model ready to the production for new architectures such as the EuroHPC systems (e.g. Leonardo machine).
The main focus will be to train researchers to improve their knoledge of the technical features of the EDITO-Model Lab ocean models with emphasis on HPC languages and tools usage, optimization and tuning in the HPC arena, GPU porting and performance assessment techniques. Different learning modules will be created and mantained within CINECA training courses in HPC (online or in presence).