Convolutional neural network acceleration engine, convolutional neural network acceleration system and method

Abstract

The invention discloses a convolutional neural network acceleration engine, a convolutional neural network acceleration system and a convolutional neural network acceleration method, and belongs to the field of heterogeneous computing acceleration, wherein the physical PE matrix comprises a plurality of physical PE units, and the physical PE units are used for executing row convolution operation and related partial sum accumulation operation; the XY interconnection bus is used for transmitting the input feature image data, the output feature image data and the convolution kernel parameters from the global cache to the physical PE matrix, or transmitting an operation result generated by the physical PE matrix to the global cache; the adjacent interconnection bus is used for transmitting anintermediate result between the same column of physical PE units; the system comprises a 3D-Memory, and a convolutional neural network acceleration engine is integrated in a memory controller of eachVault unit and used for completing a subset of a convolutional neural network calculation task; the method is optimized layer by layer on the basis of the system. According to the invention, the performance and energy consumption of the convolutional neural network can be improved.

Description

technical field

[0001] The invention belongs to the field of heterogeneous computing acceleration, and more specifically relates to a convolutional neural network acceleration engine, a convolutional neural network acceleration system and a method. Background technique

[0002] With the popularization of intelligent computing, including speech recognition, object detection, scene marking and automatic driving, etc., the prediction accuracy of the deep neural network model is required to be higher and higher, and the design of the deep neural network model (DCNN) tends to be deeper and deeper. And the scale is getting bigger and bigger, and the computing platform needs to provide enough computing power and storage capacity for this.

[0003] For applications such as deep neural networks, it brings many challenges to the computing platform: the number of layers and parameter shapes of different neural network models are different, which requires high hardware flexibility; diff...

Images