U.S.-Austria Cooperative Research on Virtual Shared Memory for Distributed-Memory Machines (Computer Science)

美国-奥地利关于分布式内存机虚拟共享内存的合作研究（计算机科学）

• 批准号: 8912776
• 负责人: Ken Kennedy
• 金额: $ 1.68万
• 依托单位: William Marsh Rice University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1990
• 资助国家: 美国
• 起止时间: 1990-04-15 至 1993-10-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=8912776&HistoricalAwards=false
• 关键词: U.S.; Austria; Cooperative; Research; Virtual

This award supports Professor Ken Kennedy and others in his group at Rice University to maintain collaboration in computer science research with Professor Hans Zima now of the University of Vienna. The Rice University team has been working for several years on a programming environment for distributed-memory machines as part of the "ParaScope" project. Simultaneously, Professor Zima has been building a similar static data partitioning system, called SUPERB, for the SUPRENUM project. During a sabbatical visit at Rice, the collaborators developed an enhanced system that moves toward dynamic partititioning that increases adaptability. Their design uses annotations to a FORTRAN program written in a data partitioning language. The user provides the annotation. The underlying compilation system then translates the FORTRAN program to a distributed memory program that employs the data assignment strategy specified by the annotations. Lack of efficient compiling programs for parallel and distributed hardware stands in the way of effective use of the new distributed computing resources by scientists in many disciplines. The principal research of the Rice group in this area is supported by an NSF grant to a Science and Technology Center there. Maintaining active collaboration with Professor Zima, now that he has returned to Europe, will be an inexpensive and valuable enhancement to the Texas-based work.

该奖项支持莱斯大学的肯·肯尼迪教授和他的团队中的其他人与现在维也纳大学的汉斯·齐玛教授在计算机科学研究方面保持合作。 作为“ParaScope”项目的一部分，莱斯大学团队多年来一直致力于分布式内存机器的编程环境。 与此同时，Zima 教授一直在为 SUPRENUM 项目构建一个类似的静态数据分区系统，称为 SUPERB。 在莱斯大学的一次休假访问期间，合作者开发了一个增强的系统，该系统朝着动态分区的方向发展，从而提高了适应性。 他们的设计使用了对用数据分区语言编写的 FORTRAN 程序的注释。 用户提供注释。 然后，底层编译系统将 FORTRAN 程序转换为分布式内存程序，该程序采用注释指定的数据分配策略。 缺乏针对并行和分布式硬件的高效编译程序阻碍了许多学科的科学家有效利用新的分布式计算资源。 莱斯大学小组在这一领域的主要研究得到了国家科学基金会向那里的科学技术中心拨款的支持。 Zima 教授现已返回欧洲，与他保持积极的合作将为德克萨斯州的工作带来廉价且有价值的增强。