METHOD AND DEVICE FOR MODEL INFERENCE BASED ON GRAPHICS RENDERING PIPELINES, AND STORAGE MEDIUM

Title:

METHOD AND DEVICE FOR MODEL INFERENCE BASED ON GRAPHICS RENDERING PIPELINES, AND STORAGE MEDIUM

Document Type and Number:

WIPO Patent Application WO/2021/022962

Kind Code:

A1

Abstract:

A method and device for model inference based on graphics rendering pipelines, and a storage medium. The method comprises: acquiring an instruction flow in a rendering thread; extracting and saving texture data information from the instruction flow, the texture data information comprising texture data; and inputting the texture data information into graphics processing unit (GPU) rendering pipelines, the GPU rendering pipelines being used for inferring with respect to the texture data based on a GPU model, thus producing an inference result of the texture data, the GPU model being a model running in a GPU. This implements the model inference of the texture data, avoids a conversion of the texture data into another data type, thus reducing I/O memory overhead.

Inventors:

SHI XINDONG (CN)
WANG SHU (CN)
WU JIANGZHENG (CN)
LU MINGWEI (CN)

Application Number:

PCT/CN2020/100759

Publication Date:

February 11, 2021

Filing Date:

July 08, 2020

Export Citation:

Click for automatic bibliography generation Help

Assignee:

HUAWEI TECH CO LTD (CN)

International Classes:

G06T1/20

Foreign References:

CN105741228A	2016-07-06
CN109934197A	2019-06-25
CN109933429A	2019-06-25
CN109002744A	2018-12-14
CN201910730619A	2019-08-08
CN2017082773W	2017-05-02

Other References: