Development of a National Information Directory for Academic Supercomputing Centers

该提案要求支持建立一种正式手段，收集和存储有关美国专门用于大规模科学计算(超级计算)的学术计算资源的性质和用途的信息。请拨经费用于为开展信息收集活动、组织讲习班和相关费用提供工作人员支助。经常有人说，这种信息目录如果便于查阅，并与最新数据一起保存，就能在高性能计算领域为用户、管理员和规划者提供帮助。这项由俄亥俄州超级计算机中心主任查尔斯·本德尔提出的努力，将在学术超级计算中心联盟(CASC)的赞助下进行，并得到该联盟成员的参与。CASC是一个由美国15个州和国家学术超级计算中心组成的联盟。CASC已经成为超级计算中心之间成功的信息交换媒介，随着高性能计算资源的更广泛部署，它在这方面的作用可能会扩大。这项提议的具体目的是为学术超级计算中心和高性能计算资源制定国家信息目录的概念和结构。作为第一步，将进行一项研究，以征求与用户如何看待其不断变化的计算需求和他们在现有高性能计算中心的体验有关的基线信息。将通过国家网络分发研究材料并收集答复。鉴于高性能计算资源的用户和供应商的广泛参与和参与，这一努力很可能产生有价值的信息和有用的产品。

This proposal requests support for establishing a formal means for collecting and storing information on the nature and use of academic computing resources dedicated to large scale scientific computing (supercomputing) in the United States. Funding is requested for staff support for conducting an information gathering activity, an organizational workshop, and related costs. It has been voiced on frequent occasions that such an information directory, if easily accessible, and maintained with current data, could ablely serve to assist the users, administrators and planners in the area of high performance computing. The effort, proposed by Charles Bender, director of the Ohio Supercomputer Center, would be carried out under the auspices and with the involvement of members of the Coalition of Academic Supercomputing Centers (CASC), a consortium of 15 state and national academic supercomputing centers in the United States. CASC has served as a successful medium of information exchange among supercomputing centers and is likely to enlarge its role in this regard as high performance computing resources become more widely deployed. The specific intent of this proposal is to develop the concept and structure of a National Information Directory for academic supercomputing centers and high performance computing resources. As a first step, a study would be conducted to solicit baseline information related to how users view their evolving computational requirements and their experiences at existing high performance computing centers. The study materials would be distributed and responses collected via the national network. Given broad participation and involvement of users and providers of high performance computing resources this effort is likely to produce valuable information and useful products.