Scheduling multiprocessor tasks on hypercubes

The problem to be addressed here is one of scheduling multiprocessor tasks, i.e. tasks which may require more than one processor at a time. This model is quite natural for many parallel applications. It is, however, different than standard approach in the scheduling theory. In this paper, a low-order polynomial-time preemptive scheduling algorithm is proposed for tasks scheduled on uniform processors, when schedule length is the performance measure. The results of computational experiments over the algorithm are reported