A simple VLSI architecture for neurocomputing
Recent advances in ''neural'' computation models will only demonstrate their true value with the introduction of parallel computer architectures designed to optimise the computation of these neural networks. General-purpose neurocomputers is one approach which provides a framework for executing neural networks in much the same way that traditional computers address the problems of ''number crunching'', for which they are best suited. This framework must include a means of programming (i.e. operating system and programming languages) and the hardware must exploit the properties of VLSI, for parallelism and communication. At University College London, the authors have designed and are currently implementing in CMOS, a primitive processing element for building a parallel MIMD Neurocomputer, configured from arrays of these elements connected to a host computer (Figure 1). The goal of the Neurocomputer is to support a range of Connectionist algorithms spanning both neural networks and semantic networks. The primitive processing element is a self-contained microprocessor composed of three basic units: a 16 instruction set processor, a communication unit and a small local memory. Each processing element has a neuron name, used for message routing. The communication unit provides two bi-directional parallel links (implemented by four unidirectional) and a simple protocol to support a logical bus for routing message packets.
- Research Organization:
- Dept. of Computer Science, Univ. College London, London WC1E 6BT (GB)
- OSTI ID:
- 6275327
- Report Number(s):
- CONF-8809132-
- Journal Information:
- Neural Networks; (United States), Vol. 1:1; Conference: 1. International Neural Network Society annual meeting, Boston, MA, USA, 6 Sep 1988
- Country of Publication:
- United States
- Language:
- English
Similar Records
Coherent VLSI design environment. Final letter report for period ending 30 September 1987
Neurocomputing: Picking the human brain
Related Subjects
ARRAY PROCESSORS
PROGRAMMING LANGUAGES
PARALLEL PROCESSING
COMPUTER ARCHITECTURE
ALGORITHMS
COMMUNICATIONS
COMPUTER CALCULATIONS
COMPUTER NETWORKS
COMPUTERIZED SIMULATION
DESIGN
ELECTRONIC CIRCUITS
GEOMETRY
MEMORY DEVICES
MOS TRANSISTORS
NERVOUS SYSTEM
OPTIMIZATION
MATHEMATICAL LOGIC
MATHEMATICS
PROGRAMMING
SEMICONDUCTOR DEVICES
SIMULATION
TRANSISTORS
990210* - Supercomputers- (1987-1989)