基于近似Booth4编码的新型低功耗乘法器
A Novel Low Power High Speed Multiplier Based on Booth4 Encoding
陆雨龙 1李少珍 1向石涛1
作者信息
- 1. 武汉轻工大学电气与电子工程学院,湖北 武汉 430000
- 折叠
摘要
随着物联网应用的不断智能化和高集成化,对新型高速及低功耗运算单元的需求日益迫切,因此终端设备需要不断降低功耗和提升运算速率.针对运算单元功耗高的问题,在保证精准度的前提下,对 Booth4 编码进行近似优化处理,并结合符号补偿技术和乘法系数优化的方法,提升传统 Booth4 乘法器的运算速率,降低乘法器的功耗.与传统Booth4 乘法器相比,该新型Booth4 乘法器的功耗和延时分别降低了 16.9%和 22.9%.为了验证新型 Booth4乘法器的实用性,利用其对图像 Lena和 Gameraman进行图像滤波处理,结果显示图像质量参数为优.
Abstract
The development of more and more intelligent and highly integrated Internet of Things,urges the demand for new high-speed and low-power computing units.Therefore terminal devices need to continuously reduce power consumption and increase operation rate.In order to decrease the high power consumption of calculation units and confirm their premise and accuracy,Booth4 code is approximated,optimized and combined with the symbol compensation technology and multiplication coefficient optimization method.All results show that the operation rate of the conventional Booth4 multiplier is improved and the power consumption of the multiplier is reduced.Compared with the conventional Booth4 multiplier,the new Booth4 multiplier proposed in this paper reduces 16.9%and 22.9%in terms of power consumption and delay,respectively.The new Booth4 multiplier is applied to image filtering on images Lena and Gameraman for verification,whose results show superior parameters of image quality.
关键词
Booth算法/乘法器/近似/高斯滤波/低功耗Key words
Booth algorithm/multiplier/approximation/Gaussian filter/low power consumption引用本文复制引用
出版年
2024
NETL
NSTL
万方数据