• DocumentCode
    3016337
  • Title

    CUDA code support in multiagent platform JADE

  • Author

    Zaoralek, L. ; Gajdos, Petr

  • Author_Institution
    Dept. of Comput. Sci., VSB - Tech. Univ. of Ostrava, Ostrava-Poruba, Czech Republic
  • fYear
    2012
  • fDate
    27-29 Nov. 2012
  • Firstpage
    546
  • Lastpage
    551
  • Abstract
    This paper describes a new feature of multi-agent framework which enables execution of GPU kernels. This brings a novel implementation of JADE CUDA Agent that accepts PTX/CUDA-C code and runs this code on GPUs. All such agents can receive PTX code or CUDA source code via ACL messages. Beyond integration of CUDA support, this paper is also focused on performance measurement of proposed solution and evaluation of final results.
  • Keywords
    C language; Java; graphics processing units; multi-agent systems; operating system kernels; parallel architectures; parallel programming; performance evaluation; ACL messages; CUDA source code; GPU kernels; JADE CUDA agent; PTX/CUDA-C code; multiagent platform JADE; mutliagent framework; Computer architecture; Graphics processing units; Instruction sets; Java; Kernel; Performance evaluation; Registers; ACL messages; CUDA; JADE; Multi-Agent System; PTX code;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Intelligent Systems Design and Applications (ISDA), 2012 12th International Conference on
  • Conference_Location
    Kochi
  • ISSN
    2164-7143
  • Print_ISBN
    978-1-4673-5117-1
  • Type

    conf

  • DOI
    10.1109/ISDA.2012.6416596
  • Filename
    6416596