Intelligent application online design evaluation system and method based on FPGA

Abstract

The invention discloses an intelligent application online design evaluation system and method based on an FPGA, which can comprehensively and objectively evaluate the intelligent application processing performance of a non-embedded FPGA and an embedded FPGA from multiple dimensions of power consumption, accuracy, calculation efficiency and the like, and can be used as an FPGA intelligent application design online competition support platform. Moreover, the system supports a user to complete the design, simulation, comprehensive compiling, testing and other operations on a non-embedded FPGA and an embedded FPGA online, and can be used as a training platform for popularizing the FPGA.

Description

technical field [0001] The invention relates to the technical field of online design and evaluation of intelligent applications, in particular to an FPGA-based system and method for online design and evaluation of intelligent applications. Background technique [0002] As the complexity of artificial intelligence application algorithms continues to increase, its requirements for hardware computing power are also increasing. The traditional CPU isomorphic computing method is far from meeting the computing needs of applications. FPGA (Field Programmable Gate Array, Field Programmable Gate Array) As a programmable device, it can obtain better performance per power ratio than GPU, and has higher flexibility than ASIC (Application Specific Integrated Circuit). Therefore, research on FPGA-based machine learning algorithm accelerators has attracted a lot of academic and industry attention. Under the background of self-controllability, using non-embedded FPGA and Phytium, Loongson ...

AI Technical Summary

Problems solved by technology

[0006]However, this solution mainly has the following technical problems: the FPGA chip of this solution does not have an integrated CPU core to form an embedded FPGA, nor is it connected to a server through a PCIe high-speed interface to form a The FPGA cloud server accelerates the computing mode, so the platform cannot support the collaborative and efficient computing of intelligent application scenarios such as image detection and speech recognition

[0009] However, this solution mainly has the following technical problems: Although the platform can test the acceleration performance of the application, it does not correlate with the FPGA power consumption and the accuracy test of the intelligent application processing The scheme design cannot objectively and scientifically evaluate the user's design scheme systematically, and cannot be used as a competition platform to select excellent design teams and design schemes for this field; in addition, this scheme is only for CPU+FPGA heterogeneous coordination acceleration mode, no consideration is given to the intelligent application processing performance of the embedded FPGA

Images