点击次数：

论文类型：会议论文

发表时间：2015-08-24

收录刊物：EI、CPCI-S、Scopus

页面范围：1826-1831

关键字：DAG scheduling; heterogeneous computing; random graphs generator; task priority; static scheduling

摘要：Efficient application scheduling algorithms are important to obtain high performance in heterogeneous computing systems. However, most of current algorithms are of low efficiency in scheduling. Aiming at this problem, we propose a heterogeneous scheduling algorithm with improved task priority (HSIP) whose functionality relies on two pillars: 1) a task priority strategy based on standard deviation and Communication Cost Weight to make scheduling priority more efficient; 2) an entry task duplication selection policy to make the makespan shorter and improve the idle time slot insertion-based. We evaluate our proposed scheme on some real application DAGs by comparison with Predict Earliest Finish Time (PEFT), SD-Based Algorithm for Task Scheduling (SDBATS), Heterogeneous Earliest Finish Time (HEFT) and Critical Path On a Processor (CPOP). Experimental results show that our proposed scheme performs better than other algorithms in terms of schedule length ratio, speedup and efficiency.