Abstract
At Caltech, we have successfully used a 64 node concurrent processor for over a year and we extrapolate our experience to study the general applicability of this class of computer, We have implemented over ten scientific problems on our machine - ranging from high energy physics to matrix problems- and find typical speed ups of 50 over a single node. This corresponds to a power of 8 times a VAX11/780. We believe that the small available software base makes concurrent processors currently specialized machines. However we see no reason why they cannot be effectively applied to the large majority of computationally intense problems and so be a general purpose computer in the future when more software is available.

This publication has 4 references indexed in Scilit: