• DocumentCode
    2255092
  • Title

    Controlling and programming the SPHINX multi-SIMD pyramid machine

  • Author

    Méhat, J. ; Mérigot, A.

  • Author_Institution
    Paris VIII Univ., St. Denis, France
  • fYear
    1988
  • fDate
    10-12 Oct 1988
  • Firstpage
    423
  • Lastpage
    428
  • Abstract
    A control scheme with a very limited overhead for SPHINX, a cellular pyramidal machine primarily designed for image-processing applications, is presented. It is organized as a set of stacked layers of decreasing size interconnected according to a dual network: a mesh-based interneighbor interconnection network within a layer and a binary tree between adjacent layers. The SPHINX processing element relies on bit serial operations and communications. It is formed of an arithmetic and logic unit with enhanced data transfer capabilities, a 256-bits local memory, and a set of special-purpose registers. A low-overhead interlayer method to synchronize communication of independently controlled single-instruction multiple-data (SIMD) processor meshes is presented. It is shown how it can be used to provide multitasking within a mesh to allow opposite data flow to cross safely. Related programming concepts for multi-SIMD machines-control transmission-are presented
  • Keywords
    multiprocessor interconnection networks; parallel architectures; parallel machines; parallel programming; ALU; SPHINX; binary tree; bit serial operations; cellular pyramidal machine; control scheme; control transmission; dual network; enhanced data transfer capabilities; image-processing applications; limited overhead; logic unit; mesh-based interneighbor interconnection network; multitasking; single-instruction multiple-data; stacked layers; Binary trees; Communication system control; Image processing; LAN interconnection; Multiprocessor interconnection networks; Multitasking; Parallel languages; Process control; Process design; Registers;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Frontiers of Massively Parallel Computation, 1988. Proceedings., 2nd Symposium on the Frontiers of
  • Conference_Location
    Fairfax, VA
  • Print_ISBN
    0-8186-5892-4
  • Type

    conf

  • DOI
    10.1109/FMPC.1988.47394
  • Filename
    47394