Performance of Analytic Orbit Propagators on a Hypercube and a Workstation Cluster

摘要

In this paper we discuss the benefit of parallel computing in propagating orbits of objects. Several analytic methods are now in use operationally. We will discuss three such schemes. We demonstrate the benefit of parallelism by using an INTEL iPSC/2 hypercube and by using a cluster of Unix-based workstations running Parallel Virtual Machine (PVM). The software PVM allows a heterogeneous set of networked workstations to appear as a multicomputer. We will show that one can achieve near 100% efficiency on the hypercube.