Artificial intelligence accelerator card scheduling method and device based on domestic CPU and OS

A technology of artificial intelligence and scheduling method, applied in multi-programming device, resource allocation, program control design and other directions, can solve problems such as technical difficulty, and achieve the effect of improving data processing rate

Pending Publication Date: 2021-12-24
INSPUR QILU SOFTWARE IND
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Due to the particularity of the domestic CPU and operating system software and hardware environment at the instruction set and operating system level, it is still technically difficult to use GPU for artificial intelligence reasoning

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Artificial intelligence accelerator card scheduling method and device based on domestic CPU and OS
  • Artificial intelligence accelerator card scheduling method and device based on domestic CPU and OS
  • Artificial intelligence accelerator card scheduling method and device based on domestic CPU and OS

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] combined with figure 1 , this embodiment proposes an artificial intelligence acceleration card scheduling method based on domestic CPU and OS, the implementation process involves a server based on domestic CPU and OS, multiple artificial intelligence acceleration cards deployed on the server, deployed on the cluster management tool zookeeper or nacos dispatch center;

[0041] First, the server is used as the client to push the video stream, and the artificial intelligence accelerator card is used as the tcp server to monitor the client port;

[0042] Subsequently, a tcp connection pool is established, and the dispatch center monitors the remaining number of connections in the tcp connection pool;

[0043] Finally, when the client needs to process data, it initiates a request to the dispatch center, and the dispatch center assigns a tcp connection to the client according to the dispatch policy selected by the user, and the client can send the collected image to the corr...

Embodiment 2

[0055] combined with figure 1 , present embodiment proposes a kind of artificial intelligence acceleration card scheduling device based on domestic CPU and OS, its structure comprises:

[0056] The server based on the domestic CPU and OS is used as the client to push the video stream, and is used to send the collected images to the artificial intelligence accelerator card;

[0057] Multiple artificial intelligence accelerator cards are deployed on the server to monitor the client port;

[0058] The scheduling center is used to allocate tcp connections to the client according to the scheduling strategy selected by the user, so that the image collected by the client is sent to the corresponding artificial intelligence accelerator card through the tcp connection.

[0059] During the working process of the above scheduling device, what needs to be added is that the initial tcp connection pool size of each artificial intelligence accelerator card is the number of cameras. By defau...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an artificial intelligence acceleration card scheduling method and device based on a domestic CPU and an OS, and relates to the technical field of resource scheduling. In order to better schedule the acceleration cards, the adopted scheme relates to a server based on a domestic CPU and an OS, a plurality of artificial intelligence acceleration cards deployed in the server, and a scheduling center deployed in a cluster management tool. Firstly, a server is taken as a client for pushing a video stream, and an artificial intelligence acceleration card is taken as a tcp server for monitoring a client port; then, a tcp connection pool is established, and the dispatching center monitors the remaining connection number of the tcp connection pool; and finally, when the client needs to process data, the client initiates a request to a scheduling center, the scheduling center allocates tcp connection to the client according to a scheduling strategy selected by a user, and the client can send the acquired image to the corresponding artificial intelligence acceleration card through the allocated tcp connection. According to the invention, the image data processing efficiency of the artificial intelligence acceleration card can be improved.

Description

technical field [0001] The invention relates to the field of technical resource scheduling, in particular to an artificial intelligence acceleration card scheduling method and device based on domestic CPU and OS. Background technique [0002] With the development of domestic enterprise informatization and deep learning, many enterprises and governments are using more and more artificial intelligence technologies in their actual business. Due to the particularity of the domestic CPU and operating system software and hardware environment at the instruction set and operating system levels, it is still technically difficult to use GPUs for artificial intelligence reasoning. For example, face recognition technology is often used by companies and government agencies for functions such as punching cards, access control, and mask detection. With the increase in the number of cameras, how to quickly and accurately identify them has become an urgent problem to be solved. Contents of...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06F9/50H04L12/26H04L29/06H04L29/08
CPCG06F9/5027H04L69/163H04L43/10H04L67/60
Inventor 姜玉哲孙桂刚李超
Owner INSPUR QILU SOFTWARE IND
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap