Scheduling of precedence-constrained parallel program tasks on multiprocessors

Abstract

The classical multiprocessor scheduling problem is the problem of scheduling the tasks of a precedence-constrained task graph (representing a parallel program) onto the processors of a multiprocessor in a way that minimizes the completion time. This problem has proven difficult both in theory and in practice. In this paper we present a heuristic algorithm for the multiprocessor scheduling problem considering non-negligible inter-task (inter-processor) communication. The algorithm incorporates a better ‘tie-resolution’ and considers contention in the communication channels of the multiprocessor system, thereby producing realistic schedules. Our performance studies of the algorithm show that it displays promising results. Finally, we propose a simple and novel idea of ‘reverse scheduling’ and demonstrate its effectiveness.