Scheduling of precedence constrained tasks on multiprocessor systems

The problem of scheduling a set of precedence constraint tasks onto a finite number of identical processors with and without communication overhead is studied. The objective is to minimize the makespan. In this paper, we are concerned with a priority-list scheduling method. A new policy for ranking the priority of each task is proposed. Under this priority policy, two heuristic algorithms are proposed to solve task scheduling problems with and without communication overheads. Experiments show that our algorithm for solving the problem without communication overhead improves previous result by about 20%; for problems with communication overhead the improvement is about 70% over previous work.<>