Arithmetic device for neural network, chip, equipment and related method

Abstract

Provided are an arithmetic device for a neural network, a chip and equipment, wherein the arithmetic device includes a control unit and a multiplication-accumulation unit, the multiplication-accumulation unit includes a filter register and a plurality of calculation units, and the filter register is connected with the plurality of calculation units; the control unit is used for generating controlinformation and sending the control information to the calculation units; the filter register is used for caching a filter weight value to be subjected to multiplication-accumulation operation; and the calculation units are used for caching input characteristic values to be subjected to multiplication-accumulation operation, and performing multiplication-accumulation operation on the filter weightvalue and the input feature values according to the received control information. All the calculation units are controlled through one control unit, and thus the design complexity of the control unitcan be lowered; and one filter register is shared by the plurality of calculation units, and thus a required cache size can be reduced.

Description

technical field [0001] The present application relates to the field of neural networks, and more specifically, to a computing device, chip, equipment and related methods for neural networks. Background technique [0002] A deep neural network is a machine learning algorithm that is widely used in computer vision tasks such as object recognition, object detection, and semantic segmentation of images. A deep neural network consists of an input layer, several hidden layers, and an output layer. The output of each layer in a deep neural network is the sum of the products of a set of weight values ​​and their corresponding input feature values ​​(i.e. multiply-accumulate). The output of each hidden layer is also called the output feature value, which serves as the input feature value of the next hidden layer or output layer. [0003] A deep convolutional neural network is a deep neural network in which at least one hidden layer operation is a convolution operation. In the curr...

AI Technical Summary

Problems solved by technology

Among them, the GPU-based calculation process requires more data movement operations in the entire calculation process, resulting in relatively low energy efficiency in data processing.

However, based on the operation process of the neural network special processor, the instruction set architecture of the neural network special processor requires complex control logic to complete tasks such as instruction fetching and decoding, resulting in a large chip area required for the control logic. In addition, the neural network Dedicated processors require tool chain support such as compilers, making development difficult

Images