首页 | 本学科首页   官方微博 | 高级检索  
     

高性能通用微处理器体系结构关键技术研究
引用本文:张民选,王永文,邢座程,邓让钰,蒋江,张承义.高性能通用微处理器体系结构关键技术研究[J].计算机研究与发展,2006,43(6):987-992.
作者姓名:张民选  王永文  邢座程  邓让钰  蒋江  张承义
作者单位:国防科学技术大学计算机学院,长沙,410073
基金项目:中国科学院资助项目;国家科技攻关项目
摘    要:X处理器是我国自主设计的基于EPIC思想的高性能通用微处理器.介绍了8级流水线和OLSM执行模型,以很少的硬件代价克服了基本EPIC模型的局限性.设计了一种多分支预测结构,支持多条分支指令的并行执行,并通过判定执行减少分支指令的数目;设计了两级cache存储器,提出DTD低功耗设计方法,并通过前瞻执行隐藏访存的延迟.最后,展望了高性能通用微处理器的发展趋势.

关 键 词:微处理器  体系结构  并行  分支  存储器
收稿时间:07 11 2005 12:00AM
修稿时间:2005-07-112006-01-10

Research on High Performance General Purpose Microprocessor Architecture
Zhang Minxuan,Wang Yongwen,Xing Zuocheng,Deng Rangyu,Jiang Jiang,Zhang Chengyi.Research on High Performance General Purpose Microprocessor Architecture[J].Journal of Computer Research and Development,2006,43(6):987-992.
Authors:Zhang Minxuan  Wang Yongwen  Xing Zuocheng  Deng Rangyu  Jiang Jiang  Zhang Chengyi
Affiliation:School of Computer Science, National University of Defense Technology, Changsha 410073
Abstract:The X processor is an EPIC based high-performance general-purpose microprocessor. Eight-stage pipelines and OLSM execution model are designed, which overcomes the limitations of traditional EPIC execution model with little hardware overhead. A multi-branch predication structure is designed for parallel execution of multiple branch instructions. Predication is used to reduce branch instructions. Two-level cache memory is designed. DTD method is presented for low power design and speculation is used to hide memory latency. Finally, the future of high-performance microprocessors is prospected.
Keywords:microprocessor  architecture  parallelism  branches  memory
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号