Modeling and numerical simulation using scientific computing and intensive computing is a multidisciplinary strategic challenge so as to develop new scientific knowledge as well as maintain the scientific competitiveness and attractiveness of universities and research organizations. The rise of this awareness over the last few years led to the evolution of the national landscape in this area. In addition to the major international (Tier-0, PRACE, WLCG) and national (Tier-1, GENCI, TGCC, CINES, IDRIS, CC-IN2P3…) infrastructures that host and operate supercomputers, scientific research is in a growing need of proximity centers (Tier-2 level) called mesocentres, dedicated to scientific computation and high-performance computing.
The role of these mesocentres is fourfold :
- to provide a computing infrastructure, a proximity support and an interdisciplinary scientific environment allowing the development of both new simulation tools and knowledge in a highly competitive field,
- to allow scientists to prepare for the production environments of higher levels,
- to offer an opening towards scientific computing to disciplines that are still new to this field because of the scientific environment and proximity support,
- to respond to the necessity of pooling computing capabilities within an eco-responsible vision together with high-level engineers for an effective proximity support.
In general, laboratories cannot perform these tasks because of the means that this requires.
The challenges that simulation is facing today are the following:
- to address problems at the smallest scales of matter,
- to deal with industrial problems in high-dimensional systems,
- to take into account multi-scale and multi-physical problems.
These new challenges have in common a considerable increase in problem size, which amounts to an increase of the computing capabilities required to solve them. To meet these challenges, it is essential to resort to parallel and intensive computing. Intensive parallel computation constitutes a complex field of research in order to be able to simulate problems of large sizes.
The objective of the mesocentre is to gather skills around a common simulation tool to create a stimulating environment that helps researchers tackle the challenges of parallel computing. The foundations of this project rest upon the decision of two institutions, CentraleSupelec and ENS Paris-Saclay, to pool their computing resources within the Paris-Saclay University, host them at IDRIS and set up a common support team.
This decision was motivated by two reasons: the first one concern economies of scale, extensibility of the system and eco-responsible pooling. The second reason is to allow the experimentation, the adjustment and then the consolidation of the establishment of shared computing resources between institutions, with a particular attention given to (i) the organization of an efficient and proximity support structure, gathering engineers and technicians currently in charge of the HPC facilities on the two establishments and (ii) the setting up of a joint scientific animation based on the exchange of skills and know-how between the different disciplines.
The mesocentre is intended to increase its computing capabilities and open itself to researchers from other institutions as well as industrial companies.
It is funded by the Contrat Plan État Région (CPER).
Benefiting from the environment of the Paris-Saclay University, the mesocentre aims to create a space for scientific exchanges and to contribute to the emergence of high-performance computing strategies and projects between different actors of the Paris-Saclay University.