Method and device for processing data

Abstract

The embodiment of the invention discloses a method and a device for processing data. A specific embodiment of the method comprises the steps of obtaining a to-be-executed instruction, wherein the to-be-executed instruction comprises a data identifier; decoding the to-be-executed instruction to obtain a first address of an operand indicated by the data identifier in the memory; aligning the first address according to a preset data bit width of the memory to obtain a second address; and reading / writing the operands in the memory according to the second address. According to the embodiment, the storage utilization rate of the memory is improved.

Description

technical field [0001] The embodiments of the present application relate to the field of computer technologies, and in particular, to methods and devices for processing data. Background technique [0002] In recent years, with the rise and development of model algorithms represented by deep learning, neural network models have been widely used in various fields, such as speech recognition, image recognition, natural language processing and other fields. There are a large number of computationally intensive operators in the neural network model, such as matrix calculation, convolution, pooling, activation, normalization, and so on. Because these operations are very time-consuming, the computing capability of a traditional CPU (Central Processing Unit, central processing unit) cannot meet the demand, thus making heterogeneous computing become the mainstream. And therefore developed a variety of special processors for neural networks, such as GPU (Graphics Processing Unit, gra...

AI Technical Summary

Problems solved by technology

Because these operations are very time-consuming, the computing power of the traditional CPU (Central Processing Unit, central processing unit) is difficult to meet the demand, thus making heterogeneous computing become the mainstream

Images