Massive Parallelism in Neural Network Simulation

Andreas Zell, Niels Mache, Markus Huttel, and Michael Vogt

on a MasPar MP-1216, a parallel SIMD computer with 16,384 processors. For multilayer feedforward networks we have obtained sustained rates of up to 348 M CPS and 129 M CUPS with backpropagation, a high mark for this architecture and for general purpose SIMD computers. This paper focuses on the problems of mapping neural networks to parallel hardware, on implementation problems in obtaining high propagation rates on a SIMD machine and on problems with the resuiting learning algorithms.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.