Preemptive weighted completion time scheduling of parallel jobs

Abstract

In this paper we present a new algorithm for the off-line scheduling of parallel and independent jobs on a massively parallel processor system. We also introduce a machine model which is based on existing multiprocessors and accounts for the penalty of preemption. It is shown that the new algorithm achieves a small approximation factor for both weighted completion time and makespan scheduling. To fine tune the algorithm a fairly simple numerical optimization problem is derived. This way different preemption penalties can be considered when determining the approximation factor. Finally, we compare the generated schedule with non-preemptive schedules for the same problem.