Research in Parallel Program Design and Architecture Synthesis

并行程序设计与体系结构综合研究

• 批准号: 8914587
• 负责人: Mary Irwin
• 金额: $ 104.17万
• 依托单位: Pennsylvania State Univ University Park
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1990
• 资助国家: 美国
• 起止时间: 1990-01-01 至 1995-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=8914587&HistoricalAwards=false
• 关键词: Research; Parallel; Program; Design; Architecture

This award provides support to develop computing infrastructure for research in four primary areas: parallel program design tools, development of a CAD design system, distributed database design, and visual texture perception. The new infrastructure includes the development of a more powerful UNIX environment, the upgrading of parallel processing capability, the acquisition of a number of workstations, and the provision of hardware and software technicians. The parallel program design tools research focuses on the design of various tools to assist in the development of parallel programs. One aspect is the design of kernel-level mechanisms for distributed real- time programs. Another is the design of operating system mechanisms to support different kinds of parallel programming paradigms. A third aspect is the development of techniques for the static analysis of parallel programs, including deadlock detection. The CAD design system research focuses on the development of a constraint driven, multi-level CAD design system. This system includes CAD tools to support high-level architectural design, behavioral synthesis, logic synthesis, and layout generation. Synthesis for performance is an emphasis. Distributed database design research is concerned with two issues in the design of distributed databases. The first is the design and analysis of highly available databases. Replication methods to tolerate site and communication failures are being explored. Database design using robust data structures to tolerate an intermediate level of faults, including software faults and partial media failures is also being investigated. The second issue is the development of a distributed database system using an object mode. Visual texture perception research is concerned with visual texture perception in humans and machines. The trade-offs of using different types of filters as the front-end for the texture discrimination module in a machine vision system are being explored. VLSI-appropriate architectures for various problems in machine vision are also being studied. Some of these architectures will be prototyped using the CAD tools developed above.

该奖项为在四个主要领域的研究中开发计算基础架构提供了支持：并行程序设计工具，开发CAD设计系统，分布式数据库设计和视觉纹理感知。 新的基础架构包括开发更强大的UNIX环境，并行处理能力的升级，许多工作站的获取以及提供硬件和软件技术人员的提供。 并行程序设计工具研究重点是设计各种工具，以帮助制定并行程序的开发。 一个方面是针对分布式实时程序的内核级机制的设计。 另一个是操作系统机制的设计，以支持各种并行编程范例。 第三个方面是开发平行程序静态分析的技术，包括死锁检测。 CAD设计系统的研究重点是开发约束驱动的多层CAD设计系统。 该系统包括支持高级体系结构设计，行为综合，逻辑合成和布局生成的工具。 表演的综合是一种重点。 分布式数据库设计研究与分布式数据库设计中的两个问题有关。 首先是对高度可用数据库的设计和分析。 正在探索可容忍站点和通信故障的复制方法。 还研究了使用强大数据结构的数据库设计，以耐受中间的故障水平，包括软件故障和部分媒体故障。 第二个问题是使用对象模式开发分布式数据库系统。 视觉纹理感知研究与人类和机器中的视觉纹理感知有关。 正在探索使用不同类型的过滤器作为机器视觉系统中纹理歧视模块的前端的权衡。 还研究了针对机器视觉中各种问题的适合VLSI的架构。 其中一些架构将使用上面开发的CAD工具进行原型。