EAGER: A Novel Approach to Application Specific Instruction Processor Synthesis from Polychronous Specifications

EAGER：一种根据多时规范合成应用特定指令处理器的新方法

• 批准号: 1435281
• 负责人: Sandeep Shukla
• 金额: $ 8万
• 依托单位: Virginia Polytechnic Institute and State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-08-15 至 2016-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1435281&HistoricalAwards=false
• 关键词: EAGER; Novel; Approach; Application; Specific

Complex embedded systems contain a multitude of sensors, actuators, and a number of micro-controllers or embedded processors, which require complex multi-threaded programming with synchronizations that are conditional. The task of implementing such software manually is error-prone leading to crucial errors, and hence require expensive validation. Due to inherent difficulties of post-implementation verification, this project takes the approach of synthesizing systems that are correct by construction. This project uses a polychronous data-flow specification language which lends itself naturally to multithreaded software synthesis. The theory and tools developed in this project will enable software/hardware synthesis from the same specification which could have a significant impact for a wide array of safety-critical embedded systems.This project provides a new semantics to polychronous specification languages in the form of conditional partial orders, and utilizes this semantics to develop algorithms for application specific processor synthesis. To synthesize the processor, the investigators identify the instruction sets required, and optimize their implementations, while the synthesis of the micro-code programs to implement the functionalities are derived by schedule synthesis.

复杂的嵌入式系统包含许多传感器，执行器以及许多微型控制器或嵌入式处理器，这些处理器需要复杂的多线程编程，并具有有条件的同步。手动实施此类软件的任务是容易出错的导致关键错误，因此需要昂贵的验证。 由于实施后验证的固有困难，该项目采取了合成通过构造正确的系统的方法。该项目使用一种多量化的数据流规范语言，该语言自然而然地适合多线程软件合成。该项目中开发的理论和工具将从相同的规范中启用软件/硬件合成，这可能会对一系列安全至关重要的嵌入式嵌入式系统产生重大影响。该项目为有条件的部分订单的形式提供了一种新的语义，并以条件性的局部订单的形式提供了这种语义，并利用此语义来开发应用算法的应用程序特定处理器特定的处理器特定的处理器合成。为了综合处理器，研究人员确定所需的指令集并优化其实现，而Micro代码程序实施功能的合成是由计划合成得出的。