EN FR
Homepage Inria website


Section: Partnerships and Cooperations

European Initiatives

FP7 Projects

DEEP-ER
  • Type: COOPERATION

  • Defi: Exascale computing platforms, software and applications

  • Instrument: Integrated Project

  • Objectif: Dynamic Exascale Entry Platform - Extended Reach

  • Duration: October 2013 - September 2016

  • Coordinator: Forschungszentrum Juelich Gmbh (Germany)

  • Partner: Intel Gmbh (Germany), Bayerische Akademie der Wissenschaften (Germany), Ruprecht-Karls-Universitaet Heidelberg (Germany), Universitaet Regensburg (Germany), Fraunhofer-Gesellschaft zur Foerderung der Angewandten Forschung E.V (Germany), Eurotech Spa (Italy), Consorzio Interuniversitario Cineca (Italy), Barcelona Supercomputing Center - Centro Nacional de Supercomputacion (Spain), Xyratex Technology Limited (United Kingdom), Katholieke Universiteit Leuven (Belgium), Stichting Astronomisch Onderzoek in Nederland (The Netherlands) and Inria (France).

  • Inria contact: Stephane Lanteri

  • Abstract: the DEEP-ER project aims at extending the Cluster-Booster Architecture that has been developed within the DEEP project with a highly scalable, efficient, easy-to-use parallel I/O system and resiliency mechanisms. A Prototype will be constructed leveraging advances in hardware components and integrate new storage technologies. They will be the basis to develop a highly scalable, efficient and user-friendly parallel I/O system tailored to HPC applications. Building on this I/O functionality a unified user-level checkpointing system with reduced overhead will be developed, exploiting multiple levels of storage. The DEEP programming model will be extended to introduce easy-to-use annotations to control checkpointing, and to combine automatic re-execution of failed tasks and recovery of long-running tasks from multi-level checkpoint. The requirements of HPC codes with regards to I/O and resiliency will guide the design of the DEEP-ER hardware and software components. Seven applications will be optimised for the DEEP-ER Prototype to demonstrate and validate the benefits of the DEEP-ER extensions to the Cluster-Booster Architecture.

Collaborations with Major European Organizations

  • Prof. Martin Gander: University of Geneva, Mathematics section (Switzerland)

  • Domain decomposition methods (optimized Schwarz algorithms) for the solution of the frequency domain Maxwell equations

  • Dr. Maciej Klemm: University of Bristol, Communication Systems & Networks Laboratory, Centre for Communications Research (United Kingdom)

  • Numerical modeling of the propagation of electromagnetic waves at the nanoscale for biomedical applications