摘要
给出了一种基于2的补码的卷积运算算法及其硬件实现体系结构。该算法结构使用了一个特殊的移位累加单元代替并行补码乘法,降低了卷积系数为负数时的计算复杂度。
One convolution a lgorithm based on Radix complement code and architecture has been given in this article.It re-moves complexity when convolution coefficients are negative.It Uses one special shift and accumulator instead of the par-allel comple-ment code multiplica-tion.It decreases the complexity of parallel computa-tion whe n weights are negative.
出处
《微电子学与计算机》
CSCD
北大核心
2003年第1期70-72,共3页
Microelectronics & Computer
