Method and device for deep neural network computing acceleration

Abstract

Embodiments of the present invention propose a method, device, terminal, and computer-readable storage medium for computing acceleration of a deep neural network. The method includes: sampling each input vector that needs to be input into the matrix model to obtain multiple sampling vectors; The set quantization parameters are used to quantify the product of each sampling vector to obtain multiple quantization points; the matrix model is divided into multiple matrix blocks according to the quantization parameters; each quantization point and each matrix block are calculated to obtain multiple pre-computation tables; The calculation table calculates each input vector to obtain the calculation result of the matrix model. In the embodiment of the present invention, the pre-calculation table of the same matrix model only needs to be established once, and all input vectors that need to be calculated through the matrix model can use the pre-calculation table for table look-up calculation, which effectively saves the input vector and the matrix model. The calculation process can also maintain the original calculation effect of the matrix model.

Description

technical field [0001] The present invention relates to the technical field of data processing, in particular to a method, device, terminal and computer-readable storage medium for accelerating deep neural network calculations. Background technique [0002] Methods for speeding up deep neural networks in the prior art include matrix operations, pruning (pruning) algorithms, and binary networks. Among them, matrix-vector multiplication in matrix operations is memory-constrained, thus limiting the prediction speed of deep neural networks during computation. However, the accuracy loss of the binary network is relatively large. The pruning pruning algorithm requires a high degree of matrix sparsity, and the retraining process is complicated. Therefore, none of the existing calculation methods can well realize the calculation acceleration of the neural network. [0003] The above information disclosed in this Background section is only for enhancement of understanding of the b...

AI Technical Summary

Problems solved by technology

Among them, the matrix-vector multiplication in the matrix operation is memory-constrained, thus limiting the prediction speed of the deep neural network during the calculation process

However, the accuracy loss of the binary network is large

The pruning pruning algorithm requires a high degree of matrix sparsity, and the retraining process is complicated

Therefore, none of the existing calculation methods can well realize the calculation acceleration of the neural network.

Images