• DocumentCode
    3599537
  • Title

    The EM-X parallel computer: architecture and basic performance

  • Author

    Kodama, Yuetsu ; Sakane, Hirohumi ; Sato, Mistsuhisa ; Yamana, Hayato ; Sakai, Shuichi ; Yamaguchi, Yoshinori

  • Author_Institution
    Electrotech. Lab., Ibaraki, Japan
  • fYear
    1995
  • Firstpage
    14
  • Lastpage
    23
  • Abstract
    Latency tolerance is essential in achieving high performance on parallel computers for remote function calls and fine-grained remote memory accesses. EM-X supports interprocessor communication on an execution pipeline with small and simple packets. It can create a packet in one cycle, and receive a packet from the network in the on-chip buffer without interruption. EM-X invokes threads on packet arrival, minimizing the overhead of thread switching. It can tolerate communication latency by using efficient multi-threading and optimizing packet flow of fine grain communication. EM-X also supports the synchronization of two operands, direct remote memory read/write operations and flexible packet scheduling with priority. The paper describes distinctive features of the EM-X architecture and reports the performance of small synthetic programs and larger more realistic programs.
  • Keywords
    message passing; parallel architectures; parallel machines; performance evaluation; pipeline processing; processor scheduling; remote procedure calls; scheduling; shared memory systems; EM-X parallel computer; architecture; communication latency; direct remote memory read/write operations; execution pipeline; fine grain communication; fine-grained remote memory accesses; flexible packet scheduling; interprocessor communication; latency tolerance; multithreading; on-chip buffer; operands; optimized packet flow; packet arrival; packets; performance; priority; remote function calls; small synthetic programs; thread switching; threads; Communication switching; Computer architecture; Concurrent computing; Delay; High performance computing; Network-on-a-chip; Packet switching; Pipelines; Read-write memory; Yarn;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Computer Architecture, 1995. Proceedings., 22nd Annual International Symposium on
  • ISSN
    1063-6897
  • Print_ISBN
    0-89791-698-0
  • Type

    conf

  • Filename
    524545