• DocumentCode
    1917004
  • Title

    Poster: Matrices over Runtime Systems at Exascale

  • Author

    Agullo, Emmanuel ; Bosilca, George ; Bramas, Berenger ; Castagnede, Cedric ; Coulaud, Olivier ; Darve, Eric ; Dongarra, Jack ; Faverge, Mathieu ; Furmento, Nathalie ; Giraud, Luc ; Lacoste, Xavier ; Langou, Julien ; Ltaief, Hatem ; Messner, Matthias ; Nam

  • Author_Institution
    Hiepacs Project, INRIA, Talence, France
  • fYear
    2012
  • fDate
    10-16 Nov. 2012
  • Firstpage
    1332
  • Lastpage
    1332
  • Abstract
    The goal of Matrices Over Runtime Systems at Exascale (MORSE) project is 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. In this poster, we propose a framework for describing linear algebra algorithms at a high level of abstraction and delegating the actual execution to a runtime system in order to design software whose performance is portable accross architectures. We illustrate our methodology on three classes of problems: dense linear algebra, sparse direct methods and fast multipole methods. The resulting codes have been incorporated into Magma, Pastix and ScalFMM solvers, respectively.
  • Keywords
    graphics processing units; linear algebra; mathematics computing; multiprocessing systems; software engineering; GPU accelerator; MORSE project; Magma solver; Pastix solver; ScalFMM solver; abstraction level; graphics processing unit; large-scale multicore system; linear algebra method; matrices over runtime systems at exascale; software design; GPU; HPC; Magma; PaStiX; Runtime System; ScalFMM; multicore;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    High Performance Computing, Networking, Storage and Analysis (SCC), 2012 SC Companion:
  • Conference_Location
    Salt Lake City, UT
  • Print_ISBN
    978-1-4673-6218-4
  • Type

    conf

  • DOI
    10.1109/SC.Companion.2012.168
  • Filename
    6495951