• DocumentCode
    2397579
  • Title

    Reconfiguration and routing in defective WSI processor arrays

  • Author

    Blight, David C. ; McLeod, Robert D.

  • Author_Institution
    Dept. of Electr. & Comput. Eng., Manitoba Univ., Winnipeg, Man., Canada
  • fYear
    1995
  • fDate
    18-20 Jan 1995
  • Firstpage
    298
  • Lastpage
    307
  • Abstract
    In this paper we discuss an alternative to the traditional approach of reconfiguration and adaptive routing. We first reconfigure a faulty processor array, not with the goal of producing a regular topology, but rather to increase the bandwidth of the network (increase the conductivity of a networks defined by percolation theory). This reconfiguration will provide us with a network which is disordered and irregular but has sufficient connectivity to allow the implementation of message passing techniques (such as store and forward or wormhole techniques). In the second part of our techniques we implement modified adaptive routing algorithms which can provide deadlock free, livelock free and starvation free routing between most elements in the reconfigured array. In the final part of our approach, we utilize a topology mapping to map the physical addresses of processor elements to logical coordinates, and an inverse mapping to map logical coordinates to physical addresses. This allows regular and irregular topologies to be implemented on the processor array, and provide reliable and effective communication between required processing elements
  • Keywords
    fault tolerant computing; message passing; multiprocessor interconnection networks; network routing; parallel architectures; reconfigurable architectures; wafer-scale integration; WSI processor arrays; adaptive routing; conductivity; deadlock free routing; defects; faults; inverse mapping; livelock free routing; message passing; network bandwidth; percolation theory; reconfiguration; starvation free routing; store and forward; topology mapping; wormhole techniques; Adaptive arrays; Bandwidth; Concurrent computing; Costs; Logic arrays; Network topology; Parallel processing; Routing; Telecommunication network reliability; Wafer scale integration;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Wafer Scale Integration, 1995. Proceedings., Seventh Annual IEEE International Conference on
  • Conference_Location
    San Francisco, CA
  • Print_ISBN
    0-7803-2467-6
  • Type

    conf

  • DOI
    10.1109/ICWSI.1995.515464
  • Filename
    515464