RISC-V co-processing system and method based on convolutional neural network

Abstract

The invention relates to the technical field of data processing, and discloses an RISC-V co-processing system and method based on a convolutional neural network, and the system comprises a memory, an RISC-V main processor and a co-processor. The memory is used for storing data, the memory is connected with the RISC-V main processor and the coprocessor, and the RISC-V main processor and the coprocessor can read and write the data in the memory; the RISC-V main processor is used for sending an RISC-V function instruction; and the coprocessor is used for receiving the RISC-V function instruction, building a convolutional neural network according to the received RISC-V function instruction, reading the image data from the memory, performing operation processing on the image data based on the convolutional neural network to obtain output data, and storing the output data into the memory. According to the invention, the problems of low operation efficiency and large storage space occupation when the existing uniprocessor system carries out convolution operation are solved.

Description

technical field [0001] The present invention relates to the technical field of data processing, and specifically refers to a RISC-V co-processing system and method based on a convolutional neural network Background technique [0002] The existing satellite on-orbit processing system is based on hardware optimization of fixed algorithms, which is difficult to design, poor in versatility, and difficult to upgrade functions. At the same time, due to resource and cost constraints, the real-time data processing capability of the track is limited. The above two points lead to the fact that the speed and time for internal end users to obtain intelligence products can only be at the hour level, which is still far behind the international advanced level, and cannot meet the growing military and civilian satellite detection needs. With the rapid development of the integrated chip industry, artificial intelligence, deep learning and real-time recognition based on embedded systems have...

AI Technical Summary

Problems solved by technology

[0005] Based on the above technical problems, the present invention provides a RISC-V co-processing system and method based on convolutional neural network, which solves the problems of low operation efficiency and large storage space when the existing single processor system performs convolution operation

Images