CI-P: Planning Simulation Infrastructure Evaluation for Parallel/Distributed Computer Systems

CI-P：并行/分布式计算机系统的规划仿真基础设施评估

• 批准号: 1557244
• 负责人: Nam Sung Kim
• 金额: $ 10万
• 依托单位: University of Illinois at Urbana-Champaign
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-08-21 至 2019-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1557244&HistoricalAwards=false
• 关键词: CI; Planning; Simulation; Infrastructure; Evaluation

The importance of running one or more applications on a computer system comprised of many connected computers such as data centers and supercomputers in a parallel/distributed manner has continued to increase because of the proliferation of valuable and essential applications including various Internet-based services provided by Google, Facebook, Amazon etc. The evaluation of performance and power efficiency of parallel/distributed computer systems running such applications prove to be critical when the high impact of these applications on our society, economy and environment is considered. This necessitates a flexible, detailed, and open-source full-system simulation infrastructure, which is lacking at present. The goal of this project is to lay the groundwork for developing such an evaluation infrastructure for increased computing performance, so as to assure the leading position of the US academia and industry in large-scale parallel/distributed computing.This planning grant will enable developing a system simulator augmented with rudimentary features and models to support simulation of a simple parallel/distributed computer system. It will be distributed to the computer architecture and operating system research communities with the expectation that such a simulator will lay the foundation for developing a more advanced full-system simulation infrastructure that can quickly but precisely simulate more representative parallel/distributed computer system in the future. A tutorial will be arranged to foster the use of the system simulator, and a workshop, which will involve students, researchers from academic institutions and leading industrial/national research labs, will be organized to identify various requirements and desirable features needed by a full blown system simulation infrastructure.

在由诸如数据中心和超级计算机的许多连接的计算机组成的计算机系统上以并行/分布式方式运行一个或多个应用的重要性已经持续增加，这是因为包括由Google、Facebook、Amazon等的性能和功率效率的评估并行/当考虑到这些应用对我们的社会、经济和环境的高度影响时，运行这些应用的分布式计算机系统被证明是关键的。这就需要一个灵活、详细、开源的全系统仿真基础设施，而这正是目前所缺乏的。该项目的目标是为开发这样一个提高计算性能的评估基础设施奠定基础，以确保美国学术界和工业界在大规模并行/分布式计算方面的领先地位。该计划拨款将用于开发一个系统模拟器，该系统模拟器具有基本的功能和模型，以支持模拟简单的并行/分布式计算机系统。它将被分发到计算机体系结构和操作系统研究社区的期望，这样的模拟器将奠定基础，开发一个更先进的全系统仿真基础设施，可以快速，但精确地模拟更有代表性的并行/分布式计算机系统在未来。将安排一个教程来促进系统模拟器的使用，并将组织一个由来自学术机构和领先工业/国家研究实验室的学生、研究人员参加的研讨会，以确定成熟的系统所需的各种要求和理想功能模拟基础设施。