Experimental Systems: Baring It all Software: The MIT Raw Machine

实验系统：裸露一切软件：MIT Raw Machine

• 批准号: 9810173
• 负责人: Anant Agarwal
• 金额: $ 100万
• 依托单位: Massachusetts Institute of Technology
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1998
• 资助国家: 美国
• 起止时间: 1998-10-01 至 2001-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9810173&HistoricalAwards=false
• 关键词: Experimental; Systems; Baring; Software; MIT

EIA9810173 Anant Agarwal MIT Experimental Systems: Baring It All To Software: The MIT RAW Machine A Raw Machine is an architecture that may be suitable for billion-transistor chips. It combines a replicated set of processors with integrated configurable logic, coupled by point to point interconnect, and a compiler system that discovers and schedules instruction-level parallelism across the tiles. A Raw architecture eliminates the traditional instruction set interface and instead exposes the raw details of the hardware directly to the compiler. This project is evaluating a Raw architecture through simulation and compiler development. Research questions to be addressed include the balance of area devoted to memory, computation, and communication on the chip; implementation of a parallelizing compiler; extent to which static communication can be used by the compiler and run-time system; comparison with similar approaches such as multiscalar architectures and VLIW; and efficiency with which runtime software techniques can enhance the static Raw architecture.

EIA9810173 Anant Agarwal MIT 实验系统：将一切都暴露在软件中：麻省理工学院的RAW机器 原始机器是一种可能适用于十亿晶体管芯片的架构。 它将一组复制的处理器与集成的可配置逻辑相结合，通过点对点互连耦合，以及一个编译器系统，该编译器系统可以发现和调度跨瓦片的并行性。 Raw架构消除了传统的指令集接口，而是直接向编译器公开硬件的原始细节。 这个项目通过模拟和编译器开发来评估Raw架构。 要解决的研究问题包括：致力于内存，计算和芯片上的通信区域的平衡;并行化编译器的实现;在何种程度上可以使用的编译器和运行时系统的静态通信;比较类似的方法，如多标量体系结构和VLIW;和效率与运行时软件技术可以提高静态原始体系结构。