Distributed Computing and Asynchronism - CDA
Head : Didier EL BAZ Secretary : Justine PRANEUF
Figure 1: Reconfiguration steps of a distributed system based on Smart Blocks (ANR Smart Blocks project, common work with Femto-ST)
Team CDA (Head: Didier El Baz) produces scientific research results in the fields of high performance computing, parallel computing and distributed computing with application to combinatorial optimization and numerical simulation problems. These fields of research have known a profound mutation these years with the generalization of the concept of parallelism, the evolution of network technologies and the beginning of the convergence of parallel and distributed computing.
The recent evolution of Computer Science has changed our vision of computing, programming and using machines. The size and complexity of applications and problems to treat have led to a quest for high performance that has imposed the concept of parallelism in processors and computers architectures. This phenomenon can be seen almost everywhere, i.e. from supercomputers to desktop and laptops. Now, the architecture of processors is based on the use of many cores and specialized units like GPUs are used for general purpose parallel computing. Simultaneously, networks technologies have known an important rise, parallel and distributed computing have started to converge and new concepts like cloud computing, cluster computing and peer-to-peer computing have emerged.
Companies in the domain of aerospace industry, defense and energy as well as scientific researchers are greatly interested in these new approaches that are green and economic and that permit one to solve efficiently difficult or large size problems and real time problems.
High performance computing, parallel computing and distributed computing are considered by many international institutions as priority topics. In particular, high performance computing is one of the ten Strategic Axes for Innovation of the CNRS in France. High Performance computing is also a priority topic of the "New Industrial France" book of the Department of Economy. Grid computing are among the priority scientific topics of the French Ministry for Higher Education and Research. In Europe, High Performance computing is among the main topics of the Distributed European Infrastructure for Supercomputing Applications (DEISA) programs of FP7 and Partnership for Advanced Computing in Europe (PRACES) projects. In the US, many important programs support High Performance Computing and Grid Computing among which we can quote the TeraGrid project of the National Science Foundation (NSF), the Advanced Scientific Computing Research (ASCR) project of the Department of Energy (DOE) and the High Performance Computing Modernization Program of the Department of Defence (DoD) via DARPA. Recently, important fundings for Hexascale Computing have been announced by the White house. In Japan, project Kei-Soku led by RIKEN supercomputing center supports High Performance Computing. These circumstances as well as the recent revival of the french supercomputing industry are very propitious to research works led by scientific researchers of team CDA who have worked for a long time on parallel and distributed computing.