A Stochastic Task Scheduling Algorithm Based on Importance-ratio of Makespan to Energy for Heterogeneous Parallel Systems

In the past years, most task scheduling algorithms consider either to minimize the application´s completion time or to minimize the application´s energy consumption subject to a deadline constraint. Few of them can achieve these two performance goals at the same time. In this paper, we propose an energyaware importance-ratio-based stochastic task scheduling (EISTS) algorithm, which makes a good balance between the makespan and energy consumption and achieves high system weighted improvement based on the importance-ratio of makespan toenergy consumption for heterogeneous parallel systems. Compared with other existing algorithms, our algorithm can achieve shorter application´s completion time when the importance-ratio of makespan to energy consumption is high. On the contrary, our algorithm consumes less energy. To prove its validity, we compare our proposed algorithms with two existing scheduling algorithms(SHEFT and ECS) in terms of system weighted improvement, makespan, and energy consumption respectively. The results of simulations show that the system weighted improvement of our algorithm and those two compared algorithms are 37%, 17%, and 18% respectively, which clearly demonstrates that ourEISTS algorithm outperforms them significantly. In addition, our algorithm is also very competitive in terms of both makespan and energy consumption.