• DocumentCode
    642645
  • Title

    Live demonstration: Ethernet communication linking two large-scale neuromorphic systems

  • Author

    Partzsch, Johannes ; Mayr, Christian ; Vogginger, Bernhard ; Schuffny, Rene ; Rast, A. ; Plana, L. ; Furber, Steve

  • Author_Institution
    Dept. of Parallel VLSI Syst. & Neural Circuits, Tech. Univ. Dresden, Dresden, Germany
  • fYear
    2013
  • fDate
    8-12 Sept. 2013
  • Firstpage
    1
  • Lastpage
    1
  • Abstract
    With neuromorphic VLSI hardware rapidly moving towards large-scale, possibly immovable systems capable of implementing brain-scale neural models in hardware, there is an emerging need to be able to integrate multi-system combinations of sensors and cortical processors over distributed, multisite configurations. In a recent paper, we proposed a UDP-based AER spiking interface that permits direct bidirectional spike communications over standard networks [1]. Our chosen test case for the demonstration is communication between a large-scale digital neuromorphic system, SpiNNaker [2] and wafer-scale mixed-signal neuromorphic system, BrainScaleS [3]. Since SpiNNaker processors are entirely general-purpose, it is possible to use any given core on a chip for management and system functions. We implemented the UDP interface as a packet interceptor within the Monitor on the chip connected directly to the Ethernet interface. On the BrainScales side, a dedicated FPGA board handles all external communication and configuration of the mixed-signal hardware [4]. The UDP pulse packets use the Ethernet contained on the FPGA board.
  • Keywords
    VLSI; biomedical communication; brain; field programmable gate arrays; local area networks; transport protocols; FPGA board; SpiNNaker processors; UDP interface; UDP-based AER spiking interface; brain-scale neural models; cortical processors; direct bidirectional spike communications; ethernet communication; external communication; hardware systems; integrate multisystem combinations; large-scale neuromorphic systems; latency performance; mixed-signal hardware; neuromorphic VLSI hardware; packet interceptor; refractoriness; sensors; spike-train statistics; synfire chain; timing spread; visual processing chain; wafer-scale mixed-signal neuromorphic system; Educational institutions; Electronic mail; Hardware; Neuromorphics; Program processors; Semiconductor device modeling;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Circuit Theory and Design (ECCTD), 2013 European Conference on
  • Conference_Location
    Dresden
  • Type

    conf

  • DOI
    10.1109/ECCTD.2013.6662196
  • Filename
    6662196