Scheduling communication-aware tasks on distributed heterogeneous computing systems

Abstract

This article addresses the impacts of performance on scheduling tasks in heterogeneous computing systems. A heuristic, called the enhanced communication-aware task scheduling algorithm, is proposed to exploit the potential of parallel processing, allowing for system heterogeneity. This article also shows that the scheduling performance is affected by the heterogeneity of computational power and that of communication mechanisms. The experimental results reveal the superiority of the proposed algorithm to those presented previously.