Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Neural network computing deployment method and device, storage medium and computer equipment

A neural network and computer program technology, applied in the fields of storage media, computer equipment, computing deployment methods and devices for neural networks, and can solve problems such as reducing execution efficiency

Pending Publication Date: 2021-04-20
SPREADTRUM COMM (SHANGHAI) CO LTD
View PDF4 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] Traditionally, when deploying divided operators to computing units, the AI ​​computing framework is usually deployed in units of operators. At this time, the host (Host) is required to participate in the deployment of each operator, resulting in Frequent input / output (Input / Output, referred to as I / O) overhead may reduce execution efficiency

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
  • Neural network computing deployment method and device, storage medium and computer equipment
  • Neural network computing deployment method and device, storage medium and computer equipment
  • Neural network computing deployment method and device, storage medium and computer equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0025] As mentioned in the background technology, when the host deploys the operators split from the neural network to the corresponding computing units, the host needs to participate in the deployment of each operator, resulting in frequent I / O overhead between the host and the computing units, which may Reduce execution efficiency.

[0026] In order to solve the above problems, an embodiment of the present invention provides a calculation deployment method of a neural network, the neural network includes a plurality of operators, the method includes: obtaining the calculation characteristics of the current calculation unit, the calculation characteristics include the Describe the support capability of the current computing unit for each operator; obtain one or more operators from the neural network according to the computing characteristics of the current computing unit; fuse the obtained one or more operators, to generate a sub-network, and the sub-network is used to be dep...

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 a neural network computing deployment method and device, a storage medium and computer equipment, the neural network comprises a plurality of operators, the method comprises the following steps: acquiring the computing characteristics of a current computing unit, wherein the computing characteristics comprise the supporting capacity of the current computing unit for each operator; obtaining one or more operators from the neural network according to the calculation characteristics of the current calculation unit; and fusing the obtained one or more operators to generate a sub-network, the sub-network being used for being deployed to the current computing unit. Therefore, when the host deploys the operators segmented from the neural network to the corresponding computing units, the I / O overhead between the host and the computing units can be reduced, and the execution efficiency is improved.

Description

technical field [0001] The present invention relates to the field of computer technology, and in particular to a neural network computing deployment method and device, a storage medium, and computer equipment. Background technique [0002] With the development of artificial intelligence (AI for short), neural networks have also been more and more widely used. Neural network refers to artificial neural network. In the field of machine learning and cognitive science, it is a mathematical model or computational model that imitates the structure and function of biological neural network. An operator is the smallest computing unit that constitutes an artificial neural network, and a neural network is usually composed of multiple operators in multiple layers. The operators in the neural network can be divided and deployed to different computing units. Wherein, the calculation unit refers to a module or device that can execute the divided operator or operator set. [0003] Tradi...

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): G06N3/063G06N3/04
CPCG06N3/04G06N3/063
Inventor 罗佳
Owner SPREADTRUM COMM (SHANGHAI) CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com