Fine grain parallel processors and real-time applications: MIMD controller/SIMD array

Abstract

AB ST R ACTWe discuss important issues in the applications of massively parallel architectures for real-time cognition. One of these is real-time control of the array processor. Systems accommodating massive data parallelism are faced with the critical issue of how to incorporate adaptive programming decisions without losing machine efficiency. This involves re-examination of functions to be performed by the software versus those performed by the hardware. Our discussion is based on a MlMD controller that we have constructed for a SlMD array.We then discuss the efficiency of a SlMD architecture as a computational model with reference to several functions important in vision applications and that have been benchmarked on our system. Applications that we have implemented involving real-time control are also described.