Parallel computation of partial differential equations: A modelling approach

The use of multiprocessors allowing the parallel execution of a large number of independent processes, provides a promising approach to the need of high computation speeds for some important numerical applications. However, this speed gain is limited by the need of inter-processor information exchange, namely by the necessity of synchronisation points within the parallel computation. Furthermore, communication times themselves introduce a reduction in the speed of parallel computation. The purpose of this paper is to investigate parallel computation in a specific application area: the numerical solution of large partial differential equation systems.