Model reasoning method and device based on graphic rendering pipeline and storage medium

A graphics rendering and reasoning method technology, applied in the computer field, can solve a large number of IO memory overhead and other problems, to achieve the effect of reducing IO memory overhead and avoiding conversion

Pending Publication Date: 2021-02-09
HUAWEI TECH CO LTD
View PDF3 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Therefore, in the existing technology, the inference calculation of the TensorFlow Lite framework on the GPU is based on the Compute shader. The fixed input and output data formats must be in the SSBO format. There is a large amount of IO memory overhead for format conversion of Texture type data

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
  • Model reasoning method and device based on graphic rendering pipeline and storage medium
  • Model reasoning method and device based on graphic rendering pipeline and storage medium
  • Model reasoning method and device based on graphic rendering pipeline and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0099] It should be understood that although the terms first, second, etc. may be used in the embodiments of the present invention to describe image sizes and storage formats, these image sizes and storage formats should not be limited to these terms. These terms are only used to distinguish image sizes or saving formats from one another. For example, without departing from the scope of the embodiments of the present application, the first image size can also be called the second image size, similarly, the second image size can also be called the first image size, the storage format and image size The expressions are similar and will not be repeated here.

[0100] Mobile terminals With the continuous iteration of hardware devices, more and more mobile applications have integrated AI functions, making mobile terminals more intelligent and providing users with more natural and humanized services. In terms of images and videos, applications such as face recognition, scene recogn...

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 provides a model reasoning method and device based on a graphic rendering pipeline and a storage medium. The method comprises the steps of obtaining an instruction stream in a renderingthread; extracting and storing texture data information from the instruction stream, wherein the texture data information comprises texture data; inputting the texture data information into a GPU rendering pipeline of a graphic processor, wherein the GPU rendering pipeline is used for conducting reasoning based on a GPU model on the texture data to obtain a reasoning result of the texture data, wherein the GPU model is a model running in a GPU. Model reasoning of the texture data is achieved, conversion between the texture data and other forms of data types is avoided, and then IO memory expenditure is reduced.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a model reasoning method, device, storage medium and the like based on a graphics rendering pipeline. Background technique [0002] With the continuous iteration of hardware devices on mobile terminals, more and more mobile applications integrate artificial intelligence (AI) functions, making mobile terminals more intelligent and providing users with more natural and humanized services. In terms of images and videos, applications such as face recognition, scene recognition, text recognition, and image super-resolution provide mobile terminals with core functional support such as secure unlocking, reading retrieval, and video beautification. The basis of AI functions usually depends on the reasoning platform of the neural network. Then, in terms of hardware computing power, it also changes from the central processing unit (Central Processing Unit, CPU), graphics processi...

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): G06T1/20G06T11/40G06N5/04
CPCG06T1/20G06T11/40G06N5/04G06N3/063G06N3/045G06T1/60G06T3/40G06T7/40
Inventor 石鑫栋王术吴江铮陆明伟
Owner HUAWEI TECH CO LTD
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