Using Declarative Machine Descriptions in a Retargetable Optimizing Compiler

在可重定向优化编译器中使用声明性机器描述

• 批准号: 0311482
• 负责人: Norman Ramsey
• 金额: $ 35万
• 依托单位: Harvard University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2003
• 资助国家: 美国
• 起止时间: 2003-07-01 至 2007-10-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0311482&HistoricalAwards=false
• 关键词: Using; Declarative; Machine; Descriptions; Retargetable

A computer program is written in a high-level *programming* *language*, but to be run, it must be translated into a *machine* *language*. Each kind of machine---desktop computer, cell phone, handheld computer---has a different machine language and needs a different translator. At present, each translator must be written by an expert who specializes *both* in translation *and* in machine language. This project will develop methods by which a translator can be written without such double expertise. An expert who knows only about the machine will write a *declarative* description that says what the machine does. The project's software will analyze this description and mechanically produce components to be used by an expert who knows only about translation. Because this approach relies crucially on the description's being correct, the project will also develop techniques for mechanically checking that the description is consistent with real hardware. The descriptions and the analysis thereof will enable the machine expert's work to be reused in many different translators (and other applications) and the translation expert's work to be readily applied to many different machines. Translators for new machines will be developed more quickly, more reliably, and with less effort than at present.

计算机程序用高级 *编程 * * *语言 *编写，但是要运行，必须将其翻译成 *机器 * *语言 *。 每种机器---桌面计算机，手机，手持计算机---都有不同的机器语言，需要不同的翻译器。 目前，每个翻译人员都必须由专家在翻译中 *和 *用机器语言的专家撰写。 该项目将开发可以在没有这种双重专业知识的情况下编写翻译人员的方法。 仅了解机器的专家会写一个 *声明的 *描述，说明机器的作用。 该项目的软件将分析此描述，并机械地生产仅了解翻译的专家使用的组件。 由于这种方法至关重要的是描述的正确性，因此该项目还将开发机械检查描述与真实硬件一致的技术。 描述及其分析将使机器专家的工作能够在许多不同的翻译器（以及其他应用程序）中重复使用，并且翻译专家的工作很容易应用于许多不同的机器。 新机器的翻译人员将更快，更可靠地开发，并且努力少于目前。