Virtual Institute — High Productivity Supercomputing


EuroHPC (2021-2024)

DEEP – Software for Exascale Architectures

The DEEP-SEA project will deliver the programming environment for future European Exascale systems. The project adapts all levels of the software stack to support highly heterogeneous compute and memory configurations. The software stack allows code optimization across existing and future architectures and systems. It includes low-level drivers, computation and communication libraries, resource management, and programming abstractions with associated runtime systems and tools.

The project focuses on systems that follow the design principles of the Modular Supercomputing Architecture (MSA), where different components, such as standard CPUs and different types of accelerators, including GPUs, constitute modules of a single system. From these, a subset can be allocated for a specific workload. Applications can leverage the parts of the MSA that result in an optimal energy–performance ratio. Large-scale simulations, data analytics, and machine learning – jobs have different requirements to run at the optimal combination of energy use and delivered compute performance. However, the complexity of such systems makes it hard to allocate the resources optimally.

Parallelization is already non-trivial in a traditional HPC environment – with the rising complexity of heterogeneous resources, this is hardly manageable by users and application developers. Simplifying the management and programming of heterogeneous computing and memory architectures is, therefore, the prime goal of DEEP-SEA. As part of DEEP-SEA, project members use and extend tools from the VI-HPS suite, including Extrae/Paraver, Extra-P, Scalasca, and Score-P.  




The DEEP-SEA project receives funding from the EuroHPC Joint Undertaking (JU), which is supported by the European Union’s Horizon 2020 research and innovation programme and Germany, France, Spain, Greece, Belgium, Sweden, United Kingdom, Switzerland.

Logos of the funding agencies