TUTNC - A GENERAL-PURPOSE PARALLEL COMPUTER FOR NEURAL-NETWORK COMPUTATIONS

This paper presents the architecture and realization of the TUTNC (Tam pere University of Technology Neural Computer) system. TUTNC is design ed for parallel computation and it is suitable for several artificial neural network (ANN) algorithms. For parallel processing a MIMD multip rocessor architecture is chosen. Processors are connected via a tree s haped communication network, which can be programmed to perform effici ent global operations. The architecture is modular and expandable. The cost of the system is kept low by using commercial DSPs (digital sign al processors) and FPGA (field programmable gate array) chips. A small -scale prototype of four processors has been built. Performance is mea sured using some neural network mappings and estimations are also give n for a larger system. Results show that a good cost/performance ratio can be achieved with this architecture.