期刊文献+

一种支持SIMD指令的流水化可拆分乘加器结构 被引量:2

A Pipelined Splittable Multiply-accumulate Unit Architecture
下载PDF
导出
摘要 乘加器是媒体数字信号处理器的关键运算部件。该文结合32位数字信号处理器芯片MD32开发(“863”计划)实践,提出了一种流水化可拆分的乘加器硬件实现结构,通过对乘法操作的流水处理实现了200MHz工作频率下的单周期吞吐量指标,通过构造可拆分的数据通道实现了对SIMD乘法指令的支持,支持4个通道16位媒体数据的并行乘法,大大提升了处理器的媒体处理性能。文中对所提出的乘加器体系结构,给出了理论依据和实验结果,通过MD32的流片实现得到了物理验证。 A pipelined splittable multiply-accumulate unit architecture designed for media digital signal processor is presented in this paper. The MAC operations are pipelined to achieve single-cycle 32-bit × 32-bit MAC throughput at 200 MHz. Splittable data-path architecture is employed to support 4-way 16-bit × 16-bit single-instruction-multiple-data (SIMD) MAC instructions, The MAC architecture is physically verified in silicon DSP chip.
作者 李东晓
出处 《计算机工程》 CAS CSCD 北大核心 2006年第7期264-266,共3页 Computer Engineering
基金 国家自然科学基金资助项目(90307002) 国家"863"计划基金资助项目(2002AA1Z1140)
关键词 乘加器 SIMD 流水化 可拆分 Multiply-accumulate (MAC) SIMD Pipelined Splittable
  • 相关文献

参考文献4

  • 1郑伟,姚庆栋,张明,蒋志迪,李东晓,赖莉亚,周莉.一种支持SIMD指令的低功耗分裂式ALU设计[J].计算机工程,2004,30(17):175-177. 被引量:1
  • 2郑伟,姚庆栋,张明,刘鹏,李东晓.一种高性能、低功耗乘法器的设计[J].浙江大学学报(工学版),2004,38(5):534-538. 被引量:8
  • 3Liao Yuyun,Roberts D B.A High-performance and Low-power 32-bit Multiply-accumulate Unit with Single-instruction-multiple-data (SIMD) Feature[J].IEEE Journal of Solid-State Circuits,2002,37(7):926-931.
  • 4Farooqui A A,Oklobdzija V G.General Data-path Organization of a MAC Unit for VLSI Implementation of DSP Processors[C].Proc.of ISCAS,1998,2:260-263.

二级参考文献9

  • 1[3]John M, Lewis G. Low Power Asynchronous Digital Signal Processing [PhD Thesis]. University of Manchester, 2000- 10
  • 2[4]Chandrakasan A P, Brodersen R W. Low Power Digital CMOS Design. Kluwer Academic Publishers, 1995
  • 3LEWIS M,GEORGE J.Low power asynchronous digital signal processing [D].Manchester: University of Manchester,2000.
  • 4BOOTH A D.A signed binary multiplication technique [J].The Quarterly Journal of Mechanics and Applied Mathematics,1951,4: 236-240.
  • 5HUANG Tso-bing,JAN Jeng-hsiun,TSAI Ming-yu,et al.Partition methodology for the final adder in a tree-structure parallel multiplier generater [A].Proceedings of Asia-Pacific Conference on Circuits and Systems [C].Singapore ..IEEE ,2002: 471-474.
  • 6GOLDOVSKY A,PATEL B,SCHULTE M,et al.Design and implementation of 16 by 16 low-power two's complement multiplier [A].Proceedings of IEEE International Symposium on Circuits and Systems [C].Geneva: IEEE,2000: 345-348.
  • 7ERCEGOVAC M D,LANG T.On-the-fly conversion of redundant into conventional representations [J].IEEE Transactions on Computer,1987,C36(7): 895-897.
  • 8CIMINIERA L,MONTUSCHI P.Catty-save multiplication schemes without final addition [J].IEEE Transactions on Computers,1996,C45(9): 1050-1055.
  • 9孙海平,李伟,王锐,高明伦.基于资源共享的ALU设计[J].微电子学与计算机,2001,18(5):16-20. 被引量:5

共引文献7

同被引文献8

引证文献2

二级引证文献1

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部