Collaborative Proposal: CRI: CRD: Computer-aided Design Tool and Compiler Repository for Reconfigurable Computing

由于增加的逻辑容量，诸如现场可编程门阵列（FPGA）的可重构逻辑器件已经成为许多数字系统中越来越重要的组件。这些器件的可编程结构为设计人员提供了将并行设计快速映射到硅的机会，同时将电路定制为所需的功能。该电路的成功使用在很大程度上依赖于复杂的计算机辅助设计工具和编译器的设计映射流。随着设计人员用于可重构逻辑器件的编程模型变得越来越多样化，单独和联合优化工具的需求变得更加重要。人们普遍认为，在这个可重构计算领域的研究人员将需要方便地访问代表性的工具集，描述如何将它们联合收割机到映射流的文档，并在不久的将来收集代表性的基准，以维持研究进展。项目完成后，设计人员将可以访问组织良好的文档工具和编译器存储库，包括如何将多个工具协调到完整的设计映射流程中的文档。该基础设施将使研究人员能够优化现有工具以实现新的设计目标，并开发新的、更先进的工具，这些工具将使用当前的基础设施作为基础。基础设施中的所有资源都将托管在位于阿默斯特的马萨诸塞州大学和位于麦迪逊的威斯康星州大学的网络储存库中。该项目的资金将用于购买所需的计算机硬件和基于网络的资源分发材料，并支持将组织基础设施的研究生研究人员。完成后，该基础设施将向公众开放，并将向各个研究领域的研究人员广泛宣传。该提案概述了实现这一目标的一系列开发步骤，包括购买现有的CAD工具和编译器，构建基于Web的基础设施，将单个工具集成到一系列端到端映射流程中，以及制定基础设施部署战略。拟议的开发工作与研究生和本科生的学习机会密切相关。这些学生将有机会熟悉最先进的设计工具和先进的网页设计和数据库建设技术。与学术界和商业界研究人员的初步接触表明了浓厚的兴趣。预计随着项目的进展，外部各方将协助提供CAD工具和基准方面的技术指导和支持。随着项目接近完成，将征求学术界和工业界合作伙伴的财团继续支持基础设施。

As a result of increased logic capacity, reconfigurable logic devices, such as field programmable gate arrays (FPGAs), have become increasingly important components in many digital systems. The programmable structure of these devices offer designers the opportunity to quickly map parallel designs to silicon while customizing circuitry to exactly the functionality desired. The successful use of this circuitry relies heavily on design mapping flows of intricate and complex computer-aided design tools and compilers. As the programming models used by designers to target reconfigurable logic devices become more diverse, the need to optimize tools both individually and jointly becomes more important. It is widely recognized that researchers in this reconfigurable computing field will require easy access to representative tool sets, documentation describing how to combine them into mapping flows, and a collection of representative benchmarks in the near future to sustain research progress.The need for greater accessibility to freely-available reconfigurable computing CAD tools forms the cornerstone of this proposal. Upon completion of the project, designers will have access to a well-organized repository of documented tools and compilers including documentation on how to coordinate multiple tools together into complete design mapping flows. This infrastructure will allow researchers to optimize existing tools for new design goals and develop new, more advanced tools which use the current infrastructure as a basis. All resources in the infrastructure will be hosted in web-based repositories at the University of Massachusetts, Amherst and the University of Wisconsin, Madison. Funding for this project will be used to acquire needed computer hardware and web-based resource distribution materials, and to support graduate student researchers who will organize the infrastructure. Upon completion, the infrastructure will be made publicly-available and will be widely publicized to researchers in a variety of research fields.This proposal outlines a series of development steps to achieve this goal including the acquisition of existing CAD tools and compilers, the construction of a web-based infrastructure, the integration of the individual tools into a series of end-to-end mapping flows, and the development of an infrastructure deployment strategy. The proposed development work is closely tied to learning opportunities for graduate and undergraduate students. These students will have the opportunity to become familiar with state-of-the-art design tools and advanced web-design and database building techniques. Preliminary contact with researchers in both academic and commercial settings has indicated strong interest. It is expected that as the project progresses external parties will assist in offering technical direction and support in terms of CAD tools and benchmarks. As the project nears completion, a consortium of academic and industry partners will be solicited to continue support for the infrastructure.