• DocumentCode
    2176031
  • Title

    A Polynomial Algorithm to Prove Deadlock-Freeness of Wormhole Networks

  • Author

    Taktak, Sami ; Encrenaz, Emmanuelle ; Desbarbieux, Jean-Lou

  • Author_Institution
    Syst. on Chip Dept., Pierre & Marie Curie Univ., Paris, France
  • fYear
    2010
  • fDate
    17-19 Feb. 2010
  • Firstpage
    121
  • Lastpage
    128
  • Abstract
    Deadlocks are an important issue in wormhole networks and a lot of works have been done on deadlock avoidance. Sufficient and necessary deadlock-freeness conditions have been proposed and used to build deadlock-free wormhole networks. But none of these works provide an efficient way to verify if a given network is deadlock-free. The present article proposes a new sufficient and necessary condition associated with a polynomial algorithm to check if a given network is deadlock-free. This new algorithm identifies escape paths from channels that are involved in circular dependencies. Identifying escape paths can be easily done by studying strongly connected components of the dependency graph instead of studying individually each cycle. The proposed algorithm has been implemented in a tool for automatic detection of deadlocks in wormhole networks (ODI). ODI has been used to establish the deadlock-freeness of complex realistic networks and networks with defective routers.
  • Keywords
    graph theory; multiprocessor interconnection networks; polynomials; system recovery; automatic deadlock detection; deadlock avoidance; deadlock freeness; dependency graph; escape paths; polynomial algorithm; wormhole networks; Adaptive systems; Laboratories; Network topology; Network-on-a-chip; Polynomials; Production; Routing; Sufficient conditions; System recovery; System-on-a-chip; Network; deadlock; dependency graph; routing function; wormhole;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Parallel, Distributed and Network-Based Processing (PDP), 2010 18th Euromicro International Conference on
  • Conference_Location
    Pisa
  • ISSN
    1066-6192
  • Print_ISBN
    978-1-4244-5672-7
  • Electronic_ISBN
    1066-6192
  • Type

    conf

  • DOI
    10.1109/PDP.2010.19
  • Filename
    5452501