一个50 MIPS(峰值)32/ 64b微处理器

R. Conrad, R. Devlin, D. Dobberpuhl, B. Gieseke, R. Heye, G. Hoeppner, J. Kowaleski, M. Ladd, J. Montanaro, S. Morris, R. Stamm, H. Tumblin, R. Witek
{"title":"一个50 MIPS(峰值)32/ 64b微处理器","authors":"R. Conrad, R. Devlin, D. Dobberpuhl, B. Gieseke, R. Heye, G. Hoeppner, J. Kowaleski, M. Ladd, J. Montanaro, S. Morris, R. Stamm, H. Tumblin, R. Witek","doi":"10.1109/ISSCC.1989.48185","DOIUrl":null,"url":null,"abstract":"An RISC (reduced-instruction-set-computer) microprocessor is described that, subject to data dependencies, can issue one 32-b instruction every 20-ns cycle to achieve peak performance of 50 MIPS (million instructions per second) for worst-case process and operating conditions. The chip includes a 64-b by 32-b general-purpose register file, a 22-b by 32-b privileged-register file, a 1 kB eight-way-associative virtual instruction cache, a 2-kB direct-mapped write-through physical data cache, an 8-entry fully associative instruction address translation buffer, a 32-entry fully associative data address translation buffer, a 10-entry by 64-b output data FIFO, 3-entry by 64-b instruction input FIFO, a 2-entry by 64-b data input FIFO, hardware support for multiprocessing, and a heavily pipelined integer execution unit. Although the execution unit has a 32-b datapath, the data cache, external interface, and register file are organized by 64 b to maximize data transfer rates and to allow single-cache issue of all double-precision instructions. The chip is fabricated in a 1.5- mu m drawn n-well double-metal CMOS process. It contains 294353 transistors, of which 135680 are in the cache arrays, measures 14.5 mm*9.5 mm, and is mounted in a 224-pin surface-mount leaded chip carrier. Power dissipation is 9 W at a 20 ns cycle time.<<ETX>>","PeriodicalId":385838,"journal":{"name":"IEEE International Solid-State Circuits Conference, 1989 ISSCC. Digest of Technical Papers","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1989-02-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"8","resultStr":"{\"title\":\"A 50 MIPS (peak) 32/64 b microprocessor\",\"authors\":\"R. Conrad, R. Devlin, D. Dobberpuhl, B. Gieseke, R. Heye, G. Hoeppner, J. Kowaleski, M. Ladd, J. Montanaro, S. Morris, R. Stamm, H. Tumblin, R. Witek\",\"doi\":\"10.1109/ISSCC.1989.48185\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"An RISC (reduced-instruction-set-computer) microprocessor is described that, subject to data dependencies, can issue one 32-b instruction every 20-ns cycle to achieve peak performance of 50 MIPS (million instructions per second) for worst-case process and operating conditions. The chip includes a 64-b by 32-b general-purpose register file, a 22-b by 32-b privileged-register file, a 1 kB eight-way-associative virtual instruction cache, a 2-kB direct-mapped write-through physical data cache, an 8-entry fully associative instruction address translation buffer, a 32-entry fully associative data address translation buffer, a 10-entry by 64-b output data FIFO, 3-entry by 64-b instruction input FIFO, a 2-entry by 64-b data input FIFO, hardware support for multiprocessing, and a heavily pipelined integer execution unit. Although the execution unit has a 32-b datapath, the data cache, external interface, and register file are organized by 64 b to maximize data transfer rates and to allow single-cache issue of all double-precision instructions. The chip is fabricated in a 1.5- mu m drawn n-well double-metal CMOS process. It contains 294353 transistors, of which 135680 are in the cache arrays, measures 14.5 mm*9.5 mm, and is mounted in a 224-pin surface-mount leaded chip carrier. Power dissipation is 9 W at a 20 ns cycle time.<<ETX>>\",\"PeriodicalId\":385838,\"journal\":{\"name\":\"IEEE International Solid-State Circuits Conference, 1989 ISSCC. Digest of Technical Papers\",\"volume\":\"1 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"1989-02-15\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"8\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"IEEE International Solid-State Circuits Conference, 1989 ISSCC. Digest of Technical Papers\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ISSCC.1989.48185\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE International Solid-State Circuits Conference, 1989 ISSCC. Digest of Technical Papers","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ISSCC.1989.48185","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 8

摘要

描述了一个RISC(精简指令集计算机)微处理器,受数据依赖性的影响,可以每20ns周期发出一条32b指令,在最坏的过程和操作条件下达到50 MIPS(每秒百万指令)的峰值性能。该芯片包括一个64-b × 32-b通用寄存器文件、一个22-b × 32-b特权寄存器文件、一个1kb 8路关联虚拟指令缓存、一个2kb直接映射透写物理数据缓存、一个8条全关联指令地址转换缓冲区、一个32条全关联数据地址转换缓冲区、一个10条64-b输出数据FIFO、一个3条64-b指令输入FIFO、一个2条64-b数据输入FIFO、支持多处理的硬件。以及一个高度流水线化的整数执行单元。虽然执行单元有一条32b的数据路径,但数据缓存、外部接口和寄存器文件都是按64b组织的,以最大限度地提高数据传输速率,并允许所有双精度指令的单缓存问题。该芯片采用1.5 μ m n阱双金属CMOS工艺制造。它包含294353个晶体管,其中135680个在缓存阵列中,尺寸为14.5 mm*9.5 mm,安装在224引脚表面贴装的引脚芯片载体中。功耗为9w,周期为20ns。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
A 50 MIPS (peak) 32/64 b microprocessor
An RISC (reduced-instruction-set-computer) microprocessor is described that, subject to data dependencies, can issue one 32-b instruction every 20-ns cycle to achieve peak performance of 50 MIPS (million instructions per second) for worst-case process and operating conditions. The chip includes a 64-b by 32-b general-purpose register file, a 22-b by 32-b privileged-register file, a 1 kB eight-way-associative virtual instruction cache, a 2-kB direct-mapped write-through physical data cache, an 8-entry fully associative instruction address translation buffer, a 32-entry fully associative data address translation buffer, a 10-entry by 64-b output data FIFO, 3-entry by 64-b instruction input FIFO, a 2-entry by 64-b data input FIFO, hardware support for multiprocessing, and a heavily pipelined integer execution unit. Although the execution unit has a 32-b datapath, the data cache, external interface, and register file are organized by 64 b to maximize data transfer rates and to allow single-cache issue of all double-precision instructions. The chip is fabricated in a 1.5- mu m drawn n-well double-metal CMOS process. It contains 294353 transistors, of which 135680 are in the cache arrays, measures 14.5 mm*9.5 mm, and is mounted in a 224-pin surface-mount leaded chip carrier. Power dissipation is 9 W at a 20 ns cycle time.<>
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信