期刊文献+

一种低功耗可配置乘加器的设计

Design of a Reconfigurable Low Power Multiply-Accumulator
下载PDF
导出
摘要 提出一种可配置的64位乘加器。根据计算模式的不同,该乘加器能够1次完成1个64×64、2个32×32、4个16×16和8个8×8的有/无符号乘加计算。在部分积(pp)产生电路中插入模式相关的选择器,并在最终树和最后的加法器中插入模式相关的进位消除电路,来实现乘加器的可配置。通过对部分积重新进行编排,避免了在部分积压缩树中插入进位消除电路。在部分积压缩树中,采用一种低功耗4∶2压缩器,有效降低了功耗和面积。最后,对乘加器的速度、面积和功耗等性能进行了分析。 A reconfigurable 64-bit multiply-accumulator(MAC) was presented.Depending on different computation mode,the MAC was capable of performing one 64×64,two 32×32,four 16×16 or eight 8×8 signed/unsigned multiply-accumulate operations.The MAC was "vectorized" by inserting mode-dependent selectors into partial product(pp) generation and inserting mode-dependent kills in carry chain of the final tree and final adder.In this work,insertion of kills into carry chain of the product reduction tree was avoided by relocating the partial product.Using low-power 4∶2 compressor in partial product tree,both power and area were reduced efficiently.Performances of the MAC,such as speed,power and area,were analyzed.
出处 《微电子学》 CAS CSCD 北大核心 2011年第2期255-259,共5页 Microelectronics
关键词 乘加器 可配置乘加器 4∶2压缩器 Multiply-accumulator Reconfigurable multiply-accumulator 4∶2 compressor
  • 相关文献

参考文献8

  • 1KRITHIVASAN S, SCHULTE MJ. Multiplier architectures for media processing [C]///37^th Conf Sign, Syst, and Comp. Asilomar, CA, US.& 2003, 2: 2193-2197. 被引量:1
  • 2TANG K C, WU A K M, FONG A S, et al. Integrated partition imeger execution unit for multimedia and conventional applications [C]/// IEEE Int Conf Elec, Circ, and Syst. Lisbon, Portugal. 1998, 2: 103-107. 被引量:1
  • 3LEE C G, STOODLEY M G. Simple vector microprocessors for muhimedia applications [C]// Proc 31 Ann ACM/ IEEE Int Symp Microarchitec. Dallas, TX, USA. 1998.. 25-36. 被引量:1
  • 4KESHAB K P. Approaches to low--power implementations of DSP systems [J]. IEEE Trans Circ and Syst I: Fundam Theo and Appl, 2001, 48(10): 1214-1224. 被引量:1
  • 5VEERAMACHANENI S, KRISHNA K, AVINASH L, et al. Novel architectures for high-speed and lowpower 3-2, 4-2 and 5-2 compressors [C] // IEEE 20th Int Conf VLSI Des. Bangalore, India. 2007: 324-329. 被引量:1
  • 6FAYED A, ELGHARBAWY W, BAYOUMI L. A data merging technique for high-speed low-power multiply accumulate units [C] /// IEEE Int Conf Acous, Speech, and Sign Process. Hillsboro, OR, USA. 2004, 51 145-148. 被引量:1
  • 7DANYSH A, TAN D. Architecture and Implementation of a vector/SIMD multiply-accumulate unit[J]. IEEE Trans Comp, 2005, 54(3).. 284-293. 被引量:1
  • 8SCHMOOKLER M, PUTRINO M, ROTH C, et al. A low-power, high-speed implementation of a power PC microprocessor vector extension [C] // 14^th IEEE Symp Comp Arithrn. Adelaide, SA, Australia. 1999: 12-19. 被引量: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