DVFS节能算法的改进研究

Improvement of energy saving algorithm in DVFS

不断增长的能源消耗已经成为制约云计算发展的瓶颈。指出了DVFS节能贪婪算法的局限性,针对降低能耗与保持处理器运算能力的矛盾,提出了一种改进的节能算法GEDT(Globle Energy-Deadline Tradeoff)。该算法通过最优松弛系数的计算,约束处理器在运行不同任务时降低频率的幅度。实验仿真结果显示,在满足任务最迟截止时间的同时,GEDT算法具有较优的节能效果。

Growing energy consumption has become a bottleneck restricting the development of cloud compu- ting. This paper pointed out the limitations of greedy algorithm in DVFS and proposed a improved energy-saving algorithm-GEDT to solve the contradiction between energy reduction and computing capacity maintainability. By the calculation of the optimal relaxation factor, it restricts the frequency reduction amplitude of processors with different tasks. The experimental shows that GEDT algorithm has better energy-saving effect while meeting the deadline for tasks.