The DEEP-SEA project focuses on developing a programming environment for future European exascale systems, adapting all levels of the software stack to support highly heterogeneous compute and memory configurations. It aims to simplify the management and programming of heterogeneous computing architectures, enabling applications to run optimally across various hardware configurations. The project integrates general-purpose CPUs, accelerators, and memory types, and supports dynamic resource allocation and application malleability, making it suitable for large-scale simulations, data analytics, and machine learning tasks. DEEP-SEA looks at all relevant parts of the system: nodes, clusters, applications, system software and tools.

To provide solutions that are usable on as many HPC systems as possible and that scale to exascale and beyond, DEEP-SEA uses the co-design approach with real-world applications from seven areas: Space Weather (xPic, AIDApy), Weather Forecast (IFS @ECMWF), Seismic imaging (RTM, BSIT), Molecular dynamics (GROMACS), Computational fluid dynamics (Nek5000), Neutron Monte Carlo transport for nu-clear energy (PATMOS), Earth System Modelling (TSMP). To support the co-design approach, an early-access programme has been established, giving access to a software stack for heterogeneous compute and memory systems that allows scientists and developers to make best use of all available resources.