NGS: StreamIt: A Language and a Compiler for Streaming Applications

NGS：StreamIt：流应用程序的语言和编译器

• 批准号: 0305453
• 负责人: Saman Amarasinghe
• 金额: $ 50万
• 依托单位: Massachusetts Institute of Technology
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-03-01 至 2007-02-28
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0305453&HistoricalAwards=false
• 关键词: NGS; StreamIt; Language; Compiler; Streaming

ABSTRACTACI-0305453; PI: AMARASINGHEThere is evidence that streaming media applications are already consuming most of the cycles on consumer machines. StreamIt is a language and complier specifically designed for modern stream programming. Its goal is to raise the abstraction level in the streaming domain., providing a natural, high-level syntax that conceals architectural details without sacrificing performance, and provide a language and compiler that can abstract away cognitive complexities of programming next generation, heterogeneous computing grids. The proposal will extend the current StreamIt language to include novel features such as variable rates, multi-dimensional streams, a messaging model and adaptable stream graphs so that the language can support a wide variety of applications in the streaming domain, and will also develop a large application base to evaluate the effectiveness of the language and the compiler. In addition the project will undertake a substantial compiler effort to create a robust and powerful compiler. With novel optimizations such as a space-time multiplexer, a stream graph partitioner, a phased scheduler a constraint scheduler, and liner transformer, as well as compilation technology necessary for heterogeneous computational platforms, and for substrates whose available resources can vary dynamically. Furthermore, the project will develop many optimizing compiler backends for different communication exposed architectures as well as next generation computing grids

有证据表明，流媒体应用程序已经消耗了消费者机器上的大部分周期。 StreamIt是一种专门为现代流编程设计的语言和编译器。 它的目标是提高流媒体领域的抽象层次。提供了一种自然的高级语法，该语法隐藏了体系结构细节而不牺牲性能，并提供了一种语言和编译器，该语言和编译器可以抽象出编程下一代异构计算网格的认知复杂性。 该提案将扩展当前的StreamIt语言，以包括可变速率，多维流，消息传递模型和自适应流图等新功能，以便该语言可以支持流媒体领域的各种应用程序，并且还将开发一个大型应用程序库以评估语言和编译器的有效性。 此外，该项目将进行大量的编译工作，以创建一个健壮和强大的编译器。 随着新的优化，如时空多路复用器，流图分区，分阶段调度器的约束调度器，和线性Transformer，以及编译技术所需的异构计算平台，并为基板的可用资源可以动态变化。 此外，该项目还将为不同的通信暴露架构以及下一代计算网格开发许多优化编译器后端