Approximate precoding convolution operation method and system based on Booth algorithm

• Main Authors: ZHENG GUOHAO, JIANG JIANFEI, KONG YAOYAO, SHENG WEIGUANG, JING NAIFENG, WANG QIN, MAO ZHIGANG
• Format: Patent
• Language: Chinese; English
• Published: 18.04.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention provides an approximate precoding convolution operation method and system based on a Booth algorithm, and the method comprises the steps: S1, inputting a multiplier and a multiplicand into a multiplier, carrying out the partial integral separation through an approximate precoding module, inputting the separated multiplicand correlation partial product into an adder tree module, and completing the compression, adding the multiplicand correlation partial product and the multiplier correlation partial product after compression to obtain a multiplier result; and S2, carrying out sign bit expansion on the multiplier result, inputting the expanded multiplier result into an accumulation module to carry out shift and addition operation, and carrying out carry compensation on the shift and addition operation result. 本发明提供了一种基于Booth算法的近似预编码卷积运算方法及系统，包括：步骤S1：乘数和被乘数输入乘法器通过近似预编码模块进行部分积分离，将分离后的被乘数相关部分积输入至加法树模块完成压缩，将压缩后的被乘数相关部分积和乘数相关部分积相加得到乘法器结果；步骤S2：将乘法器结果进行符号位扩展，并将扩展后的乘法器结果输入累加模块进行移位和相加操作，并将移位相加操作结果进行进位补偿。