We use cookies to improve your experience with our site.

计算机设计师真实环境中的正确选择:龙芯2号处理器的设计经验与教训

Making Effective Decisions in Computer Architects Real-World: Lessons and Experiences with Godson-2 Processor Designs

  • 摘要: 尽管在过去几十年中设计和开发了许多处理器,但体系结构研发团队很难从研究文献中获得详尽的设计经验的总结。本文系统地阐述了作者在设计龙芯2号系列处理器时的设计选择。龙芯2号系列处理器包括250MHz的龙芯2B、450MHz的龙芯2C以及1GHz的龙芯2E。这些处理器均采用64位、四发射、乱序执行的体系结构,分别于2003年、2004年及2005年流片成功,每款处理器的SPEC CPU2000的值都是上一款处理器的三倍。作者从这些设计中获得了第一手的设计经验,为处理器的设计提供了已有的教科书和论文所没有的独特视角和见解。作者在龙芯2号系列处理器的性能改进过程中尝试了数百个结构与设计的优化措施,论文从中总结了十个关键的经验教训,分为四个方面:硅与模拟的关系、设计平衡、性能优化和皮体系结构调整。论文最后总结了设计高性能低能耗处理器的三个最重要的因素:持续改进、面向硅上设计的态度以及对软件和工艺设计的透彻理解。

     

    Abstract: Although the design of many kinds of microprocessors has been underdeveloping for several decades, the computer architecture R\&D communitylacks well documented lessons and experiences about design decisions inthe research literature. In this paper, we systematically present thedesign decisions we made during the designing and prototyping of Godson-2series processors. The 250MHz Godson-2B, 450MHz Godson-2C, and 1GHzGodson-2E processors that implement 64-bit, four-issue, out-of-orderarchitecture were taped out in 2003, 2004, and 2005, respectively. Eachprocessor triples its predecessor in the SPEC CPU2000 rates. Ourfirst-hand experiences and lessons gained from these designs wouldprovide unique perspectives and insights that are not available in anyexisting text books and/or published papers. We summarize 10 criticallessons and experiences based on hundreds of our attempts atarchitectural and design optimizations for performance improvement ofGodson-2 series processors. The issues include silicon-simulationcorrelation, design balancing, performance optimizing, andpico-architecture tuning. We conclude that persistent improvement,attitude towards work-on-silicon design, and insightful understandingof software and fabrication process are the three most importantfactors for designing a high performance processor with low energyconsumption.

     

/

返回文章
返回