I-Corps: Customizable and scalable high-performance microprocessor

I-Corps：可定制和可扩展的高性能微处理器

• 批准号: 1720263
• 负责人: Mikko Lipasti
• 金额: $ 5万
• 依托单位: University of Wisconsin-Madison
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-01-15 至 2017-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1720263&HistoricalAwards=false
• 关键词: Corps; Customizable; scalable; performance; microprocessor

The broader impact/commercial potential of this I-Corps project is to enable superior performance and energy efficiency for cost-sensitive embedded processors. As computing devices have become ubiquitous and permeated every aspect of life, the need for greater performance has also followed. However, traditional design techniques to enable high performance microprocessors incur resource costs and typically suffer from poor energy efficiency. These overheads come at a considerable cost as most embedded designs are energy constrained and highly cost sensitive. The byproduct of these competing design constraints has been stagnating progress due to the ineffectiveness of traditional techniques. Through the use of this project's novel processor microarchitecture, commercial designs should observe approximately 30% greater energy efficiency and superior performance without incurring resource overhead. Applications such as networking and storage domains are initial candidates for adopting the design.This I-Corps project focuses on the creation of an energy-efficient microcontroller through the use of a novel processor microarchitecture. At the time most fundamental microprocessor techniques were developed, power was not a problem and transistors were scarce. To satisfy these constraints, designs focused on maximizing utilization of the few execution resources present. As transistors became more plentiful and performance of greater importance, designs utilized these additional transistors to enable complex organizational and scheduling techniques to maximize the operation of the scarce execution resources. This project's architecture reverses this trend by allocating additional transistors to execution resources and greatly simplifying organizational and scheduling overheads. Initial research suggests that considerable energy and area efficiency gains are possible while maintaining a high level of performance in comparison to current commercial designs. Additionally, the novel structure of this microprocessor enables many optimizations that are impractical in conventional designs due to their complex organization and scheduling.

I-Corps项目更广泛的影响/商业潜力是为成本敏感型嵌入式处理器提供卓越的性能和能源效率。随着计算设备变得无处不在并渗透到生活的方方面面，对更高性能的需求也随之而来。然而，实现高性能微处理器的传统设计技术会产生资源成本，并且通常会受到低能效的影响。由于大多数嵌入式设计都受到能源限制且对成本高度敏感，因此这些开销需要付出相当大的代价。由于传统技术的无效，这些相互竞争的设计约束的副产品已经停滞不前。通过使用该项目新颖的处理器微架构，商业设计应该可以在不产生资源开销的情况下观察到大约30%的能源效率和卓越的性能。诸如网络和存储域之类的应用程序是采用这种设计的最初候选者。这个I-Corps项目的重点是通过使用一种新颖的处理器微架构来创造一种节能的微控制器。在最基本的微处理器技术被开发出来的时候，功率不是问题，晶体管也很稀缺。为了满足这些约束，设计的重点是最大化现有的少数执行资源的利用率。随着晶体管越来越多，性能越来越重要，设计利用这些额外的晶体管来实现复杂的组织和调度技术，以最大限度地利用稀缺的执行资源。这个项目的体系结构通过为执行资源分配额外的晶体管和极大地简化组织和调度开销来扭转这种趋势。初步研究表明，与目前的商业设计相比，在保持高水平性能的同时，可以获得可观的能源和面积效率收益。此外，该微处理器的新颖结构使许多优化在传统设计中由于其复杂的组织和调度而不切实际。