Fifth International Conference on Simulation Tools and Techniques

Research Article

A Framework for Exploration of Parallel SystemC Simulation on the Single-chip Cloud Computer

Download41 downloads
  • @INPROCEEDINGS{10.4108/icst.simutools.2012.247751,
        author={Christoph Roth and Simon Reder and Oliver Sander and Michael H\'{y}bner and J\'{y}rgen Becker},
        title={A Framework for Exploration of Parallel SystemC Simulation on the Single-chip Cloud Computer},
        proceedings={Fifth International Conference on Simulation Tools and Techniques},
        publisher={ICST},
        proceedings_a={SIMUTOOLS},
        year={2012},
        month={6},
        keywords={single-chip cloud computer parallel simulation systemc},
        doi={10.4108/icst.simutools.2012.247751}
    }
    
  • Christoph Roth
    Simon Reder
    Oliver Sander
    Michael Hübner
    Jürgen Becker
    Year: 2012
    A Framework for Exploration of Parallel SystemC Simulation on the Single-chip Cloud Computer
    SIMUTOOLS
    ICST
    DOI: 10.4108/icst.simutools.2012.247751
Christoph Roth1,*, Simon Reder1, Oliver Sander1, Michael Hübner1, Jürgen Becker1
  • 1: Karlsruhe Institute of Technology
*Contact email: christoph.roth@kit.edu

Abstract

Since the number of cores integrated on a single die is expected to increase steadily, new memory and communication architectures as well as programming methods are needed and currently explored by manufacturers. In this context, Intel Labs developed the Single-chip Cloud Computer (SCC), a 48-core experimental processor, serving as a platform for many-core software research. Within this paper a framework targeting the investigation of SystemC kernel parallelization on the SCC is presented. The framework provides the basis for implementation of different synchronization schemes while combining distributed and shared memory programming models and exploiting multiple distinct address spaces. As a case study, a synchronous parallelization scheme is preliminarily evaluated by means of several simulation models of different accuracy. Results of the analysis give a first evidence of the applicability of the synchronous parallelization method on the homogeneous non-cache coherent manycore architecture of the SCC for detailed system simulation.