关键词: 任务调度 调度长度 Fork-Join DAG

Abstract:

This paper proposes a new algorithm based on task duplication for scheduling Fork-Join task graphs on busbased clusters. It computes the scheduling length by partitioning the task set, and allocates the tasks to the used processors as far as possible without increasing the scheduling length in order to save the number of processors. The time complexity of the proposed algorithm is higher than the existing algorithm, but its scheduling performance is    the most superior.

Key words: task scheduling, scheduling length, Fork-Join;DAG