• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

计算机工程与科学

• 高性能计算 • 上一篇    下一篇

用于高性能计算的作业调度能效性研究综述

郑文旭,潘晓东,马迪,汪浩   

  1. (国防科技大学计算机学院,湖南 长沙 410073)
  • 收稿日期:2018-10-29 修回日期:2019-03-18 出版日期:2019-09-25 发布日期:2019-09-25
  • 基金资助:

    国家数值风洞项目(NNW2018-ZT6B13)

Overview on the energy efficiency of job
scheduling for high performance computing

ZHENG Wen-xu,PAN Xiao-dong,MA Di,WANG Hao   

  1. (School of Computer,National University of Defense Technology,Changsha 410073,China)
  • Received:2018-10-29 Revised:2019-03-18 Online:2019-09-25 Published:2019-09-25

摘要:

由于科学研究与商业应用等对高性能计算的需求与日俱增,高性能计算的性能和系统规模得到迅速发展。但是,急剧增长的功耗严重限制了高性能计算系统的设计和使用,使得低功耗技术成为高性能计算领域的关键技术。作为整个系统的核心组件,作业调度系统立足有限的系统资源,对用户提交的应用进行作业-资源分配,其能效性对于整个高性能计算系统的能耗控制与调节起到至关重要的作用。首先介绍主要的能量效率技术和常用的作业调度策略,然后对当前高性能计算作业调度能效性进行分析,并讨论了其面临的挑战及未来发展方向。
 

关键词: 高性能计算, 作业调度, 能量有效性

Abstract:

Due to the increasing demand for high performance computing in scientific research and commercial applications, the performance and system scale of high performance computing are developing rapidly. However, the rapid increase in power consumption severely limits the design and use of high-performance computing (HPC) systems, which makes low-power technologies become the key technology in the HPC field. As the core component of the whole system and based on limited system resources, the job scheduling system distributes job resources to the applications submitted by the users. Its energy efficiency plays an important role in the control and regulation of the energy consumption of the whole HPC system. We first introduce the main energy efficiency techniques and common job scheduling strategies, then analyze the energy efficiency of current HPC job scheduling, and discuss the challenges it faces and future development directions.
 

Key words: high performance computing (HPC), job scheduling, energy efficiency