Performance analysis of communication-aware task scheduling algorithms for heterogeneous computing

This paper addresses the performance analysis of scheduling communication-aware tasks onto networks of workstations in heterogeneous computing environments. A heuristic algorithm is here proposed to schedule tasks by exploiting the schedule holes in schedules. It is shown that this algorithm ensures performance within a factor of two times of the optimum for general directed acyclic task graphs. Experimental results present the superiority of the proposed algorithm to those presented in previous literature, and also show that the performance would be affected by the system heterogeneities and the program structure of applications.