Hardware neural network conversion method, computing device, compiling method and neural network software and hardware collaboration system

Abstract

The invention provides a hardware neural network conversion method which converts a neural network application into a hardware neural network meeting the hardware constraint condition, a computing device, a compiling method and a neural network software and hardware collaboration system. The method comprises the steps that a neural network connection diagram corresponding to the neural network application is acquired; the neural network connection diagram is split into neural network basic units; each neural network basic unit is converted into a network which has the equivalent function with the neural network basic unit and is formed by connection of basic module virtual bodies of neural network hardware; and the obtained basic unit hardware networks are connected according to the splitting sequence so as to generate the parameter file of the hardware neural network. A brand-new neural network and quasi-brain computation software and hardware system is provided, and an intermediate compiling layer is additionally arranged between the neural network application and a neural network chip so that the problem of adaptation between the neural network application and the neural network application chip can be solved, and development of the application and the chip can also be decoupled.

Description

technical field [0001] The present invention generally relates to the technical field of neural network, and more specifically relates to the technology of realizing software neural network by a neural network chip. Background technique [0002] In recent years, deep learning technology has made breakthroughs, and has achieved high accuracy in many fields such as image recognition, language recognition, and natural language processing. However, deep learning requires massive computing resources, and traditional general-purpose processors are already slow. It is difficult to meet the computing needs of deep learning, and it has become an important development direction to hardwareize deep learning and design dedicated chips for it. At the same time, with the development of brain science, compared with the traditional Von Neumann computer, the brain has the characteristics of ultra-low power consumption, high fault tolerance, etc., and has significant advantages in processing ...

AI Technical Summary

Problems solved by technology

[0007] However, hardware also restricts the degree of freedom of neural network applications that it can support, which also brings about an important problem: it is difficult to use such chips to run actual neural network applications

However, this type of scheme puts forward too many constraints on the application, and cannot be well combined with existing applications, and it is difficult to achieve the same effect as the current state-of-the-art neural network on complex tasks.

[0032] It can be seen that the existing neural network hardware is usually directly connected with the neural network application. Either the hardware is too simple, which restricts the degree of freedom of the application, or the hardware has a high degree of freedom and is relatively complicated, making it difficult to improve the degree of integration and integration. efficiency problem

Images