Computer Engineering and Applications ›› 2011, Vol. 47 ›› Issue (6): 34-37.

• 研究、探讨 • Previous Articles     Next Articles

New optical vector-matrix multiplication

WANG Xianchao1,2,YAO Yunfei1,CHEN Yun1   

  1. 1.School of Mathematics and Computational Science,Fuyang Normal College,Fuyang,Anhui 236041,China
    2.School of Computer Engineering and Science,Shanghai University,Shanghai 200072,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-02-21 Published:2011-02-21

一种新的光学向量-矩阵乘法

王先超1,2,姚云飞1,陈 蕴1   

  1. 1.阜阳师范学院 数学与计算科学学院,安徽 阜阳 236041
    2.上海大学 计算机工程与科学学院,上海 200072

Abstract: This paper implements Optical Vector-Matrix Multiplication(OVMM) in totally parallel on a novel optical computing architecture,Ternary Optical Computer(TOC),by use of Modified Signed-Digit(MSD) number system.For higher efficiency,partial products(PP’s) are generated in parallel via M transformation,and the products of two numbers and the Vector Inner Products(VIP’s) are produced using binary-tree algorithm.Finally,an experiment verifies the feasibility and correctness of OVMM on Ternary Optical computer.It is not necessary to gauge light intensities but to judge whether there is light during decoding.As a result,it eliminates the need for A-Ds at the output,thus that the operating accuracy and speed are both improved significantly.

Key words: Modified Signed-Digit, binary-tree algorithm, ternary optical computer, partial product, vector inner product

摘要: 在三值光计算机上实现光学向量-矩阵乘法。为了利用光的并行性,研究使用MSD数字系统。在该数字系统上应用全并行无进位加法和M变换实现光学向量-矩阵乘法。为了提高效率,利用M变换以全并行的方式生成部分积,而后采用了二叉迭代法来求向量内积。实验证明了其可行性和正确性。由于该系统在信息表达上将光强与光的偏振性结合起来,所以在解码时不需要测量光强,只需判断光的有无即可,提高了运算精度和速度。

关键词: 改进的符号数(MSD), 二叉迭代法, 三值光计算机, 部分积, 向量内积