Preparing for Next Generation Numerical Simulation Platforms: NextGN
This project is supported by the Partnership University Fund.
Computational science, or numerical simulation, is considered as the third pillar of science. The continued increase in the performance of computing systems allows for more accurate simulations, thus advancing science and engineering. However, clock speeds are not increasing; the increase in computer performance comes from the use of a larger number of cores. However, the large number of cores, the large power consumption and the increasing failure rates will be major obstacles to the continued increase in performance.
The NextGN PUF project, associating Inria, Argonne National Laboratory, UIUC and CNRS, will lay the foundations for a broad and persistent France-US collaboration effort in this area. It will develop new research activities and form new generations of researchers for Exascale machines and HPC Clouds. The objective is to address the main obstacles to scalable simulation platforms by 1) reducing dramatically the communication cost and 2) make sure that simulations complete with correct results. The project is organized in 7 technical work packages, each addressing a specific challenge. Four applications domains are considered: Climate, Earthquake, Molecular dynamics, Astrophysics.
To achieve its objectives, the project will support an unprecedented volume of students and faculty visits in support of joint research, 2 research workshops per year (one in France and one in USA) with strong student participation and 1 summer school per year opened to the students of the collaboration. The project will also investigate the creation of dual/joint degrees and credit exchanges between University of Illinois at Urbana Champaign and several Universities in France (ENS Lyon, UJF Grenoble, U. Paris 11, ENS Cachan/Bretagne).