首页> 外国专利> In-memory matrix multiplication and its use in neural networks

In-memory matrix multiplication and its use in neural networks

机译:内存中矩阵乘法及其在神经网络中的应用

摘要

A method for an associative memory array of the present invention comprises the steps of storing each column of a matrix in an associated column of the associated memory array, storing each bit in the row j of the matrix in an R-matrix-row-j row of the array -; Storing a vector in each of the associated columns, wherein bit j from the vector is stored in an R-vector-bit-j that is the row of the array. The method also includes concurrently activating R-vector-bit-j and R-matrix-row-j rows of the vector-matrix pair to simultaneously receive the result of the Boolean function on all associated columns, Using the result on the associated column to calculate the product between the pairs of rows, and writing the product to R-product-j that is low in the array.
机译:用于本发明的关联存储器阵列的方法包括以下步骤:将矩阵的每一列存储在关联的存储器阵列的关联列中;将矩阵的行j中的每一位存储在R矩阵行j中。数组的行-;将向量存储在每个关联的列中,其中向量的位j存储在数组的行R-vector-bit-j中。该方法还包括同时激活向量矩阵对的R-vector-bit-j行和R-matrix-row-j行,以同时在所有关联列上接收布尔函数的结果。计算两对行之间的乘积,并将乘积写入数组中较低的R-product-j。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号