Matrix operation method and device, computer equipment and storage medium

Abstract

The invention relates to a matrix operation method and device, computer equipment and a storage medium. The method comprises the steps of obtaining a convolution kernel generated according to matrix elements of any row in a first matrix, wherein the convolution kernel comprises a row identifier; obtaining a convolution matrix generated according to the matrix elements of any column in the second matrix, the convolution matrix comprising column identifiers, and the number of the matrix elements of each row of the first matrix being the same as the number of the matrix elements of each column ofthe second matrix; inputting the convolution kernel and the convolution matrix to a convolution accelerator, and outputting a convolution value; writing a convolution value into a product matrix according to the row identifier and the column identifier, wherein the product matrix is used for storing a product result of the first matrix and the second matrix. Matrix multiplication operation is converted into convolution operation, matrix product operation is achieved through a convolution accelerator, and the hardware cost is reduced while the operation rate is guaranteed.

Description

technical field [0001] The present application relates to the field of computer technology, and in particular to a matrix operation method, device, computer equipment and storage medium. Background technique [0002] With the continuous development of computer vision, digital image processing technology also develops. In digital image processing technology, images are stored in matrix, so matrix operation is an important link in image data processing technology. As the matrix data becomes clearer, the memory space required to occupy is larger. When calculating the matrix, the matrix budget becomes relevant. A larger bottleneck for algorithms, especially in embedded platforms. [0003] In the existing technology, in order to quickly realize matrix operations, it is realized by adding a digital signal processor (digital signal processing, DSP). DSP is a theory and technology for processing signals by means of numerical calculations. DSP can realize matrix operations, but it i...

AI Technical Summary

Problems solved by technology

[0003] In the existing technology, in order to quickly realize matrix operations, it is realized by adding a digital signal processor (digital signal processing, DSP). DSP is a theory and technology for processing signals by means of numerical calculations. DSP can realize matrix operations, but it is expensive and costly. high

Images