• DocumentCode
    499115
  • Title

    Triplet-based architecture and its process migration mechanism

  • Author

    Liu, Bin ; Tong, Zhi-chen ; Gao, Yu-jin

  • Author_Institution
    Coll. of Econ. & Manage., Hebei Univ. of Sci. & Technol., Shijiazhuang, China
  • Volume
    5
  • fYear
    2009
  • fDate
    12-15 July 2009
  • Firstpage
    2754
  • Lastpage
    2758
  • Abstract
    This paper introduced scalable triplet-based computer architecture (TriBA) that is designed according to the decomposition method of complex problems. The specialized process migration mechanism for TriBA (PMMFT), which includes register tables to help record the process and related migration information, is proposed. By retransferring the messages related to the migrated process during migration, PMMFT decreases the process residual dependency and improves the process communication transparency. By the checkpoint-based process migration strategy, PMMFT can avoid the migration failure due to the possible node failure related to the migration process. In experiment, PMMFT is adopted by THINDLBA, a dynamic load balancing algorithm specially designed for TriBA, the results verified the availability of PMMFT.
  • Keywords
    computer architecture; resource allocation; checkpoint-based process migration strategy; complex problems; decomposition method; dynamic load balancing algorithm; process migration mechanism; process residual dependency; register tables; scalable triplet-based computer architecture; Computer aided instruction; Computer architecture; Cybernetics; Data communication; Data processing; Educational institutions; Joining processes; Load management; Machine learning; Multiprocessor interconnection networks; Checkpoint; Process migration; Process register table; Triplet-based architecture;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Machine Learning and Cybernetics, 2009 International Conference on
  • Conference_Location
    Baoding
  • Print_ISBN
    978-1-4244-3702-3
  • Electronic_ISBN
    978-1-4244-3703-0
  • Type

    conf

  • DOI
    10.1109/ICMLC.2009.5212644
  • Filename
    5212644