• DocumentCode
    3248577
  • Title

    Jetter: a multi-pattern parallel I/O benchmark

  • Author

    Cao, Liqiang ; Luo, Hongbing ; Zhang, Baoyin

  • Author_Institution
    High Performance Comput. Centre, Inst. of Appl. Phys. & Comput. Math., Beijing
  • fYear
    2008
  • fDate
    Sept. 29 2008-Oct. 1 2008
  • Firstpage
    459
  • Lastpage
    463
  • Abstract
    This paper proposes a new multi-pattern parallel I/O benchmark called Jetter, which evaluates parallel I/O throughput with either the contiguous I/O pattern or the non-contiguous I/O pattern, in either the share-one-file model or the file-per-process model, by either the POSIX interface or the MPI-I/O interface. Jetter helps end users make sense of the pattern performance law, and helps them develop efficient applications in a platform. We have evaluated the parallel I/O bandwidth in a 32 CPU shared memory computer with Jetter. The results show that I/O pattern determines throughput. Optimizing I/O model, interface, etc. in a pattern will improve bandwidth 2 or 3 times.
  • Keywords
    input-output programs; message passing; parallel programming; Jetter; MPI-I/O interface; POSIX interface; file-per-process model; multipattern parallel I/O system; parallel I/O throughput; pattern performance law; share-one-file model; Application software; Bandwidth; Computational modeling; Computer interfaces; Concurrent computing; High performance computing; Mathematical model; Physics; System performance; Throughput;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Cluster Computing, 2008 IEEE International Conference on
  • Conference_Location
    Tsukuba
  • ISSN
    1552-5244
  • Print_ISBN
    978-1-4244-2639-3
  • Electronic_ISBN
    1552-5244
  • Type

    conf

  • DOI
    10.1109/CLUSTR.2008.4663808
  • Filename
    4663808