Network-on-Chip Architectures for Neural Networks
Authors:
- Dmitri VainbrandDVDmitri Vainbrand
- Ran GinosarRGRan Ginosar
- 1 January 2010
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 135-144
Abstract
Providing highly flexible connectivity is a major architectural challenge for hardware implementation of reconfigurable neural networks. We perform an analytical evaluation and comparison of different configurable interconnect architectures (mesh NoC, tree, shared bus and point-to-point) emulating variants of two neural network topologies (having full and random exponential configurable connectivity). We derive analytical expressions and asymptotic limits for performance (in terms of bandwidth) and cost (in terms of area and power) of the interconnect architectures considering three communication methods (unicast, multicast and broadcast). It is shown that multicast mesh NoC provides the highest performance/cost ratio and consequently it is the most suitable interconnect architecture for configurable neural network implementation. Simulation results successfully validate the analytical models and the asymptotic behavior of the network as a function of its size.Keywords
This publication has 19 references indexed in Scilit:
- Connection-centric network for spiking neural networks2009
- A GALS Infrastructure for a Massively Parallel MultiprocessorIEEE Design & Test of Computers, 2007
- A generic reconfigurable neural network architecture as a network on chip2004
- A multi-sender asynchronous extension to the AER protocol2002
- Networks on chips: a new SoC paradigmComputer, 2002
- Orientation-selective aVLSI spiking neuronsNeural Networks, 2001
- Route packets, net wires2001
- Point-to-point connectivity between neuromorphic chips using address eventsIEEE Transactions on Circuits and Systems II: Express Briefs, 2000
- Neural Networks1996
- A communication scheme for analog VLSI perceptive systemsIEEE Journal of Solid-State Circuits, 1995