¹ College of Automation, Chongqing University of Posts and Telecommunications Chongqing 400065, China
² School of Computer Science and Technology, Chongqing University of Posts and Telecommunications Chongqing 400065, China

---

Abstract—Task partition is a NP-Hard problem on multi-core processor. To achieve a good MCT (maximum completion time) goal, this paper presents a novel task partitioning method for multi-core processor based on the cohesion and coupling properties of tasks. First, the relations between the cohesion and coupling are analyzed and the computation method of coupling cost based on cohesion is designed with the task scheduling requirements considered. Then, aiming to the problem that the MCT (maximum completion time) is different when the tasks’ coupling is changed and the tasks are rescheduled during the task partitioning process, an evaluating method of MCT with the considering of task scheduling requirements is designed. Last, this paper compartmentalizes tasks based on the MCT to make it minimum. This method reduces the processing of the combinatorial optimization problem in task partitioning process, and builds the relation between task partition and task scheduling. The experimental results testify this method available.

Index Terms—multi-core processor, Task Partition, Cohesion, Coupling

[PDF]

Cite: Jianchun Jiang and Suhua Zeng, " A Novel Task Partitioning Method for Multi-core Processor Based-on Cohesion and Coupling," Journal of Computers vol. 8, no. 5, pp. 1247-1254, 2013.