基于多粒度设计和并行加速的可重构高性能乘法器

Feng Jing, Zijun Liu, Xiaojun Ma, Guo Yang, Guo Peng, Donglin Wang
{"title":"基于多粒度设计和并行加速的可重构高性能乘法器","authors":"Feng Jing, Zijun Liu, Xiaojun Ma, Guo Yang, Guo Peng, Donglin Wang","doi":"10.1109/ICSESS.2017.8342979","DOIUrl":null,"url":null,"abstract":"This paper proposes a reconfigurable high performance multiplier (RHPM) based on multi-granularity design and parallel acceleration. Capable of supporting multiple precisions for different processing requirements, the RHPM can perform one 32×32, two 16×16, or four 8×8 bit unsigned/signed multiplication, or one 16×16, or two 8×8 bit complex number multiplication. The structures of the partial product generator and the partial product accumulator are improved in the paper, so as to reuse most of the hardware resources. Compression can be completed automatically by means of recording the validity of every bit in the partial product array which accelerates the computation dramatically. The RHPM is implemented with TSMC 28nm technology, exhibiting a 0.68s of the critical path delay, while consuming only 0.6281mW in power. Results show its significant superiority in terms of performance and power efficiency compared with our previous work or other similar products.","PeriodicalId":179815,"journal":{"name":"2017 8th IEEE International Conference on Software Engineering and Service Science (ICSESS)","volume":"37 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":"{\"title\":\"A reconfigurable high-performance multiplier based on multi-granularity design and parallel acceleration\",\"authors\":\"Feng Jing, Zijun Liu, Xiaojun Ma, Guo Yang, Guo Peng, Donglin Wang\",\"doi\":\"10.1109/ICSESS.2017.8342979\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"This paper proposes a reconfigurable high performance multiplier (RHPM) based on multi-granularity design and parallel acceleration. Capable of supporting multiple precisions for different processing requirements, the RHPM can perform one 32×32, two 16×16, or four 8×8 bit unsigned/signed multiplication, or one 16×16, or two 8×8 bit complex number multiplication. The structures of the partial product generator and the partial product accumulator are improved in the paper, so as to reuse most of the hardware resources. Compression can be completed automatically by means of recording the validity of every bit in the partial product array which accelerates the computation dramatically. The RHPM is implemented with TSMC 28nm technology, exhibiting a 0.68s of the critical path delay, while consuming only 0.6281mW in power. Results show its significant superiority in terms of performance and power efficiency compared with our previous work or other similar products.\",\"PeriodicalId\":179815,\"journal\":{\"name\":\"2017 8th IEEE International Conference on Software Engineering and Service Science (ICSESS)\",\"volume\":\"37 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2017-11-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"1\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2017 8th IEEE International Conference on Software Engineering and Service Science (ICSESS)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ICSESS.2017.8342979\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2017 8th IEEE International Conference on Software Engineering and Service Science (ICSESS)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICSESS.2017.8342979","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1

摘要

提出了一种基于多粒度设计和并行加速的可重构高性能乘法器(RHPM)。RHPM能够支持不同处理要求的多种精度,可以执行一个32×32、两个16×16或四个8×8位的无符号/有符号乘法,或者一个16×16或两个8×8位的复数乘法。本文对部分产品产生器和部分产品累加器的结构进行了改进,实现了大部分硬件资源的重用。通过记录部分积数组中每个比特的有效性,可以自动完成压缩,大大加快了计算速度。rpm采用台积电28纳米技术实现,关键路径延迟0.68秒,功耗仅为0.6281mW。结果表明,与我们以前的工作或其他同类产品相比,它在性能和功率效率方面具有显著的优势。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
A reconfigurable high-performance multiplier based on multi-granularity design and parallel acceleration
This paper proposes a reconfigurable high performance multiplier (RHPM) based on multi-granularity design and parallel acceleration. Capable of supporting multiple precisions for different processing requirements, the RHPM can perform one 32×32, two 16×16, or four 8×8 bit unsigned/signed multiplication, or one 16×16, or two 8×8 bit complex number multiplication. The structures of the partial product generator and the partial product accumulator are improved in the paper, so as to reuse most of the hardware resources. Compression can be completed automatically by means of recording the validity of every bit in the partial product array which accelerates the computation dramatically. The RHPM is implemented with TSMC 28nm technology, exhibiting a 0.68s of the critical path delay, while consuming only 0.6281mW in power. Results show its significant superiority in terms of performance and power efficiency compared with our previous work or other similar products.
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