Method and device for multiplying matrix and vector

Abstract

An embodiment of the invention discloses a method and a device for multiplying a matrix and a vector, the method comprises the following steps of: obtaining first indication information of a matrix element; reading a matrix element value of a non-zero element from a preset matrix according to the first indication information, and determining a first positional mark of the read matrix element value; acquiring second indication information of a vector element; reading a vector element value of a second positional mark corresponding to the first positional mark from input vector data according tothe second indication information; and obtaining a multiplication value of the matrix element value and the vector element value. By adopting the embodiment of the invention, the complexity of data processing can be reduced, the power consumption of data processing can be reduced, and the data processing efficiency can be improved.

Description

technical field [0001] The invention relates to the field of data processing, in particular to a method and device for multiplying a matrix and a vector. Background technique [0002] Due to the excellent performance of convolutional neural networks in image recognition, image classification, audio recognition and other data processing applications, convolutional neural networks have become one of the hot topics of various academic research. However, there are a large number of floating-point number multiplication and addition operations in the convolutional neural network, including matrix and vector multiplication operations, which require a large amount of calculation and take a long time, which makes the hardware energy consumption of the convolutional neural network large. Therefore, how to reduce the amount of floating-point calculations in convolutional neural networks has become one of the current technical problems to be solved urgently. [0003] In the prior art, ...

AI Technical Summary

Problems solved by technology

The matrix and vector operations in the prior art need to judge whether the value of the matrix element is zero in real time, and record the position of the non-zero element in real time. The realization of real-time judgment and recording is complex, the operation is complicated, the data processing efficiency is low, and the applicability is poor.

Images