• DocumentCode
    597446
  • Title

    Assessing load-sharing within optimistic simulation platforms

  • Author

    Vitali, Roberto ; Pellegrini, Alessandro ; Quaglia, Francesco

  • Author_Institution
    DIS, High-Performance & Dependable Comput. Syst. (HPDCS) Res. Group DIS - Sapienza, Sapienza Univ. of Rome, Rome, Italy
  • fYear
    2012
  • fDate
    9-12 Dec. 2012
  • Firstpage
    1
  • Lastpage
    13
  • Abstract
    The advent of multi-core machines has lead to the need for revising the architecture of modern simulation platforms. One recent proposal we made attempted to explore the viability of load-sharing for optimistic simulators run on top of these types of machines. In this article, we provide an extensive experimental study for an assessment of the effects on run-time dynamics by a load-sharing architecture that has been implemented within the ROOT-Sim package, namely an open source simulation platform adhering to the optimistic synchronization paradigm. This experimental study is essentially aimed at evaluating possible sources of overheads when supporting load-sharing. It has been based on differentiated workloads allowing us to generate different execution profiles in terms of, e.g., granularity/locality of the simulation events.
  • Keywords
    multiplexing; multiprocessing systems; synchronisation; ROOT-Sim package; load-sharing architecture; load-sharing assessment; modern simulation platforms; multicore machines; open source simulation platform; optimistic simulation platforms; optimistic simulators; optimistic synchronization paradigm; run-time dynamics; Computational modeling; Computer architecture; Kernel; Load modeling; Message systems; Organizations; Synchronization;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Simulation Conference (WSC), Proceedings of the 2012 Winter
  • Conference_Location
    Berlin
  • ISSN
    0891-7736
  • Print_ISBN
    978-1-4673-4779-2
  • Electronic_ISBN
    0891-7736
  • Type

    conf

  • DOI
    10.1109/WSC.2012.6465230
  • Filename
    6465230