ITR/SY (CISE): Software Improvement Through Binary Rewriting

ITR/SY (CISE)：通过二进制重写改进软件

• 批准号: 0113633
• 负责人: Saumya Debray
• 金额: $ 37.29万
• 依托单位: University of Arizona
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2001
• 资助国家: 美国
• 起止时间: 2001-08-01 至 2005-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0113633&HistoricalAwards=false
• 关键词: ITR; SY; CISE; Software; Improvement

ABSTRACTProposal #0113633U of ArizonaAndrews, GregoryA binary rewriting system is a software system that transforms a binary (executable) program into a different but functionally equivalent binary program. This project is developing binary rewriting techniques for flexible link-time and run-time code optimizations. The aim is to develop a unified binary rewriting infrastructure that is able to handle a wide variety of applications---sequential, parallel, distributed,and mobile---hardware architectures---from RISC to CISC---and optimization criteria---including execution time, power consumption, and communication bandwidth.Existing techniques for compile-time code optimization suffer from several limitations: they are unable to cross the dividing line between application code and libraries; they cannot take advantage of commonly encountered values along the critical path if such values cannot be guaranteed to be compile-time constants;and they typically focus only on improving execution time. To overcome these limitations, this project is investigating the following topics: (1) low-level cost models that can be used for cost-benefit analyses of different optimization metrics; (2) efficient computation of value profiles and their use forlow-level code specialization; and (3) techniques that reduce the overheads associated with communication libraries used by parallel and mobile applications.

格雷戈里亚利桑那州安德鲁斯的提议#0113633U二进制重写系统是将二进制(可执行)程序转换成不同但功能等价的二进制程序的软件系统。该项目正在开发用于灵活的链接时和运行时代码优化的二进制重写技术。目标是开发一个统一的二进制重写基础设施，它能够处理各种应用程序-顺序、并行、分布式和移动-硬件体系结构-从RISC到CISC-以及优化标准-包括执行时间、功耗和通信带宽。现有的编译时代码优化技术受到几个限制：它们不能跨越应用程序代码和库之间的分界线；如果不能保证关键路径上经常遇到的值是编译时常量，它们就不能利用这些值；并且它们通常只专注于改善执行时间。为了克服这些限制，该项目正在研究以下主题：(1)可用于不同优化指标的成本效益分析的低级成本模型；(2)价值概况的有效计算及其在低级代码专门化中的使用；(3)减少并行和移动应用程序所使用的通信库的相关管理费用的技术。