Section: Partnerships and Cooperations

International Initiatives

JLPC

Inria joint-Lab on Extreme Scale Computing:

• Coordinators: Franck Cappello and Marc Snir.

• Other partners: Argonne National Lab, Inria, University of Urbanna Champaign, Tokyo Riken, Jülich Supercomputing Center, Barcelona Supercomputing Center.

• Abstract: The Joint Laboratory is based at Illinois and includes researchers from Inria, and the National Center for Supercomputing Applications, ANL, Riken, Jülich, and BSC. It focuses on software challenges found in extreme scale high-performance computers.

MORSE

Matrices Over Runtime Systems at Exascale

• Inria Associate-Teams program: 2011-2016

• Coordinator: Emmanuel Agullo (Hiepacs)

• Parners: Inria (Runtime & Hiepacs), University of Tennessee Knoxville, University of Colorado Denver and KAUST.

• Abstract: The Matrices Over Runtime Systems at Exascale (MORSE) associate team has vocation to design dense and sparse linear algebra methods that achieve the fastest possible time to an accurate solution on large-scale multicore systems with GPU accelerators, using all the processing power that future high end systems can make available. To develop software that will perform well on petascale and exascale systems with thousands of nodes and millions of cores, several daunting challenges have to be overcome both by the numerical linear algebra and the runtime system communities. With Inria Hiepacs, University of Tennessee, Knoxville and University of Colorado, Denver.

• INESC-ID, Lisbon, Portugal on appplication moldeling.

• UWLAX (Wisconsin) works with us on network topology modeling;

• we collaborate with ICL at University of Tennessee on instrumenting MPI applications and modeling platforms (works on hwloc take place in the context of the Open MPI consortium) and MPI and process placement

• On the industrial side collaborate with Cisco Systems about network topologies and platform models and Intel on modeling many-core platforms and BULL on memory hierarchy modeling.

• ETH Zurich (Switzerland), on topology mapping;

• PPL (U. Illinois at Urbana Champaign) on topology-aware load-balancing (through the Inria-Urbana-Argonne Joint Lab).

• University of Tokyo and Riken on the adaptation of MPI and runtime systems to MIC processors.

• Oak Ridge National Laboratory on high-performance network programming interfaces.

ANR-JST FP3C

Framework and Programming for Post Petascale Computing.

• ANR-JST 2010 Program, 01/09/2010 - 31/03/2014

• Identification: ANR-10-JST-002

• Coordinator: Serge Petiton (Inria Saclay)

• Other partners: CNRS IRIT, CEA DEN Saclay, Inria Bordeaux, CNRS-Prism, Inria Rennes, University of Tsukuba, Tokyo Institute of Technology, University of Tokyo, Kyoto University.

• Abstract: Post-petascale systems and future exascale computers are expected to have an ultra large-scale and highly hierarchical architecture with nodes of many-core processors and accelerators. That implies that existing systems, language, programming paradigms and parallel algorithms would have, at best, to be adapted. The overall structure of the FP3C project represents a vertical stack from a high level language for end users to low level architecture considerations, in addition to more horizontal runtime system researches.

SEHLOC

Scheduling evaluation in heterogeneous systems with hwloc

• STIC-AmSud 2012 Program, 01/2013 - 12/2014 (24 months)

• Coordinator: Brice Goglin

• Other Partners: Universidad Nacional de San Luis (Argentina), Universidad de la Repúpublica (Uruguay).

• Abstract: This project focuses on the development of runtime systems that combine application characteristics with topology information to automatically offer scheduling hints that try to respect hardware and software affinities. Additionally we want to analyze the convergence of the obtained performance from our algorithms with the recently proposed Multi-BSP model which considers nested levels of computations that correspond to natural layers of nowadays hardware architectures.

NextGN

Preparing for Next Generation Numerical Simulation Platforms

• PUF (Partner University Fund) - France USA, 01/2013 - 12-2016 (3 years)

• Coordinator: Franck Capello, Marc Snir and Yves Robert

• Other Partners: Inria, Argonne National Lab and University of Urbanna Chapaign

• This PUF proposal builds on the existing successful joint laboratory between Inria and UIUC that has produced in past three years and half many top-level publications, some of which resulted in student awards; and several software packages that are making their way to production in Europe and USA. The proposal extends the collaboration to Argonne National Laboratory (ANL) and CNRS researchers who will bring their unique expertise and their skills to help addressing the scalability issue of simulation platforms.