Relative performance comparisons for the Group Explicit class of methods on MIMD, SIMD and pipelined vector computers

Abstract

The study of the potentiality and suitability of the Group Explicit (GE) class of methods for parallel computers will be completed with the reporting of some results obtained from the implementation of the Group Explicit with Ungrouped ends (GEU) method on two different types of computer systems: the ICL DAP and the CRAY-1S. Finally, a diagrammatical description of the parallel behaviour on a MIMD system of all the GE schemes experimented with will be presented in comparison with the Standard Explicit method.