XPS: EXPL: Enabling An Ecosystem of Parallel Programming Abstractions

XPS：EXPL：启用并行编程抽象生态系统

• 批准号: 1628929
• 负责人: Eric Van Wyk
• 金额: $ 29.58万
• 依托单位: University of Minnesota-Twin Cities
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-09-01 至 2021-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1628929&HistoricalAwards=false
• 关键词: XPS; EXPL; Enabling; Ecosystem; Parallel

All modern computers, from smart phones to supercomputers, have multiple processing units. Since writing programs that effectively use these resources is notoriously difficult, many programming languages, displaying a wide variety of linguistic abstractions, have been created. The "right" set of abstractions depends on many factors, ranging from the nature of the application at hand, the sophistication of the programmer, and the degree of performance desired and effort required to achieve it. For this research, the intellectual merits are in the creation of tools and language specifications that support an ecosystem of parallel programming language abstractions. These are to be developed by independent parties and safely and automatically imported into a programmer's compiler so that programmers can pick and choose the language features that they feel best meet their needs. The project's broader significance and importance are in a transformative methodology and supporting tools for writing parallel programs, one which dramatically simplifies the development of new parallel programming and domain-specific language features because it allows researchers to focus their efforts on the design and implementation of new abstractions. By making it easier for programmers to use and experiment with new features, distributed as composeable language extensions, they can write performant and correct parallel programs with less effort.The proposed approach will use a specification of the C programming language based on context free grammars (for specifying concrete syntax) and attribute grammars (for specifying semantic analysis, optimization, and code generation). This defines the host language into which parallel programming language extensions are imported. Language extensions are also specified in these formalisms as they easily compose with the host language and other extensions. Modular analyses of the language extensions ensure that the composition of the programmer's chosen extensions will be successful and result in a working compiler that will not terminate abnormally. Of interest is the degree to which new and existing language abstractions can be realized in this composable approach to language extension.

所有现代计算机，从智能手机到超级计算机，都有多个处理单元。 由于编写有效地使用这些资源的程序是非常困难的，因此已经创建了许多显示各种语言抽象的编程语言。 “正确”的抽象集取决于许多因素，从手头的应用程序的性质，程序员的复杂性，以及所需的性能和努力程度，以实现它。对于这项研究，智力的优点是在创建工具和语言规范，支持生态系统的并行编程语言抽象。 这些将由独立的各方开发，并安全自动地导入程序员的编译器，以便程序员可以挑选他们认为最能满足他们需求的语言功能。 该项目的更广泛的意义和重要性在于编写并行程序的变革性方法和支持工具，这大大简化了新并行编程和特定领域语言功能的开发，因为它允许研究人员将精力集中在新抽象的设计和实现上。通过使程序员更容易使用和试验新的功能，作为可组合的语言扩展，他们可以编写性能和正确的并行programs with less efforts.建议的方法将使用一个规范的C编程语言的上下文无关的语法（用于指定具体的语法）和属性语法（用于指定语义分析，优化和代码生成）的基础上。这定义了导入并行编程语言扩展的宿主语言。 语言扩展也在这些形式主义中指定，因为它们很容易与宿主语言和其他扩展组合。 语言扩展的模块化分析确保程序员选择的扩展的组合将是成功的，并导致编译器不会异常终止。感兴趣的是在何种程度上可以实现新的和现有的语言抽象，在这种组合的方法来语言扩展。