Task Scheduling on a Multiprocessor System with Independent Memories

This paper considers a model of a computing system with several independent but identical processors, each with a private memory of limited, and possibly different, storage capacity. The tasks are assumed to have known resource demands expressed as processing times and memory requirements. Several scheduling strategies are evaluated by worst-case performance bounds and simulation results. Both preemptive and nonpreemptive scheduling are considered. An optimal preemptive algorithm is given to find the shortest schedule for a task set with no precedence constraints.