大数据 ›› 2021, Vol. 7 ›› Issue (5): 31-39.doi: 10.11959/j.issn.2096-0271.2021047

• 专题:国产环境下的大数据处理系统 • 上一篇    下一篇

从格点量子色动力学应用看国产超算环境的基础软件

宫明1,2, 蒋翔宇1,2, 陈莹1,2, 刘朝峰1,2   

  1. 1 中国科学院高能物理研究所,北京 100049
    2 中国科学院大学,北京 100049
  • 出版日期:2021-09-15 发布日期:2021-09-01
  • 作者简介:宫明(1981- ),男,博士,中国科学院高能物理研究所副研究员,主要研究方向为格点量子色动力学
    蒋翔宇(1996- ),男,中国科学院大学博士生,主要研究方向为格点量子色动力学
    陈莹(1970- ),男,博士,中国科学院高能物理研究所研究员,主要研究方向为格点量子色动力学
    刘朝峰(1977- ),男,博士,中国科学院高能物理研究所研究员,主要研究方向为格点量子色动力学
  • 基金资助:
    国家重点研发计划资助项目(2017YFB0203202);国家自然科学基金资助项目(11775229);国家自然科学基金资助项目(11935017);国家自然科学基金资助项目(12075253)

Software infrastructures for Chinese supercomputers from the perspective of lattice QCD applications

Ming GONG1,2, Xiangyu JIANG1,2, Ying CHEN1,2, Zhaofeng LIU1,2   

  1. 1 Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049, China
    2 University of Chinese Academy of Sciences, Beijing 100049, China
  • Online:2021-09-15 Published:2021-09-01
  • Supported by:
    The National Key Research and Development Program of China(2017YFB0203202);The National Natural Science Foundation of China(11775229);The National Natural Science Foundation of China(11935017);The National Natural Science Foundation of China(12075253)

摘要:

格点量子色动力学(LQCD)是用数值模拟方法研究基本粒子的重要科学领域,因其巨大的数据量和计算规模而成为国际上超级计算机的主要科研应用之一。随着国产新一代超级计算机的发展,LQCD的计算软件由于其传统编程模型的限制,面临着更新换代的关键节点。从格点量子色动力学的视角出发,分析大规模科学应用软件对底层基础软件的需求特点,面向国产超算平台的发展方向,提出适配于大规模高效异构计算和大数据处理的新编程模型,为国产超算环境的基础软件建议了一个有潜力的发展方向。

关键词: 格点量子色动力学, 高性能计算, 大数据处理, 基础软件, 编程模型

Abstract:

Lattice QCD is a frontier scientific field for studying elementary particles by numerical simulation methods, which has become one of the major scientific research applications of supercomputers.With the rapid development of Chinese supercomputers, the LQCD softwares need to be refactored due to the limitation of its traditional programming model.The characteristics of scientific applications on super computers from the perspective of lattice QCD were reviewed.A novel programming model targeted to Chinese super computers was proposed to adapt large-scale scientific applications with big data processing, which is a promising development direction for the basic softwares of Chinese supercomputing ecosystem.

Key words: lattice QCD, HPC, big data processing, basic software, programming model

中图分类号: 

No Suggested Reading articles found!