AI chip task processing method and device based on x86 architecture

Abstract

The invention relates to an AI chip task processing method and device based on an x86 architecture. The method comprises the steps of changing a kernel configuration file and conducting kernel recompilation on an x86 system based on a changed kernel configuration file; building a compiling environment for the x86 system by utilizing a preset compiler version and the library file; performing equipment registration on the AI chip in the established compiling environment to generate an AI chip driver, and mounting the AI chip driver to the x86 system; obtaining a calculation task, and downloadingthe calculation task to an AI chip driver by utilizing a runtime program of the x86 system; and performing task configuration and enabling on the functional module of the AI chip by utilizing AI chipdrive and the calculation task until the task is completed. According to the scheme provided by the invention, the AI chip can run on the server under the x86 architecture, and can process the calculation task, so that the application range of the AI chip is expanded, and a rare hardware environment is provided for actual AI chip design, test and application.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to an x86 architecture-based AI chip task processing method, device, computer equipment and storage medium. Background technique [0002] As the most potential technology at present, artificial intelligence has been applied in all aspects of life, such as face recognition, disease detection, construction railway flaw detection, traffic optimization, customer flow analysis, etc.; The two advantages of low power consumption are bound to become the mainstream AI computing hardware in the future, and will be deployed on computing centers and servers such as a large number of computer clusters. [0003] At present, mainstream AI chip manufacturers and open source projects mostly use arm-based CPUs to communicate with AI chips. The disadvantage is that it is limited to a single platform architecture, and only supports arm cpu and arm linux to complete driver registration and op...

AI Technical Summary

Problems solved by technology

[0003] At present, mainstream AI chip manufacturers and open source projects mostly use arm-based CPUs to communicate with AI chips. The disadvantage is that it is limited to a single platform architecture, only supports arm cpu and arm linux to complete the registration and operation of the driver, and does not natively support mainstream x86 architecture computers and servers, which greatly limits the scope of use of AI chips

Images