Communication-Aware Affinity Scheduling Heuristics in Multicore Systems

This article presents the application of heuristic algorithms to solve the affinity scheduling problem in multicore computing systems. Affinity scheduling is a technique that allows efficient utilization of heterogeneous computing systems, by assigning a set of tasks to cores, taking into account specific efficiency and quality-of-service criteria. The heuristics proposed in this article are useful methods to solve realistic instances of the communication-aware affinity scheduling problem, which account for the different speed of communication and data transfer between tasks executing in different cores on a multicore system. The experimental analysis demonstrates that the proposed heuristics outperform the results computed using traditional scheduling techniques up to 12.3% when considering both the communication and synchronization times between tasks.