Unlock instant, AI-driven research and patent intelligence for your innovation.

Deep learning training task-oriented processor memory optimization method and device

A technology of deep learning and optimization methods, applied in the field of deep learning, can solve problems such as occupancy and low utilization of computing units, and achieve the effects of avoiding overhead, avoiding video memory occupation, and avoiding storage overflow

Pending Publication Date: 2020-02-04
BEIJING BAIDU NETCOM SCI & TECH CO LTD
View PDF3 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The embodiment of the present application provides a processor video memory optimization method and device for deep learning training tasks, by determining the optimal path for transferring calculation results, and using the optimal path to transfer the calculation results of the first calculation unit to the second calculation unit , while avoiding occupying video memory, avoiding the problem of low utilization of GPU computing units caused by video memory swapping

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
  • Deep learning training task-oriented processor memory optimization method and device
  • Deep learning training task-oriented processor memory optimization method and device
  • Deep learning training task-oriented processor memory optimization method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0038] Exemplary embodiments of the present application are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present application to facilitate understanding, and they should be regarded as exemplary only. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the application. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

[0039] At present, GPU is usually used to complete the training of deep learning models. GPU contains a small number of control units and storage units and a large number of computing units, which has good concurrency, and the number of computing units can reach 5120 or even more. In order to improve the utilization rate of the computing unit of the GPU, it is necess...

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 deep learning training task-oriented processor video memory optimization method and device, and relates to the technical field of artificial intelligence. The optimal path for transmitting the calculation result is determined, and the calculation result of the first calculation unit is transmitted to the second calculation unit by using the optimal path, so that the problem of low utilization rate of the calculation unit of the GPU caused by video memory exchange is avoided while video memory occupation is avoided, and the training speed of most tasks is hardly reduced.

Description

technical field [0001] The embodiments of the present application relate to the technical field of deep learning, and in particular to a method and device for optimizing processor memory for deep learning training tasks. Background technique [0002] At present, deep learning (Deep Learning) is a new field in machine learning research. Its motivation is to establish and simulate the neural network of human brain for analysis and learning, so as to imitate the mechanism of human brain to explain data, such as images, sounds and texts. . Typical applications of deep learning are image recognition and speech recognition. During the application process, it is necessary to train deep learning models, such as face recognition models, speech recognition models, etc. [0003] Typically, a processor, such as a Graphics Processing Unit (GPU), is used to train a deep learning model. GPU has multiple computing units, a small number of control units and storage units. In order to ful...

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): G06N3/063G06N20/00
CPCG06N3/063G06N20/00G06N3/08G06T1/20G06T1/60G06F12/08G06F2212/657G06F2212/1024G06F2212/502G06F2212/454G06F2212/455G06F7/768G06F13/1657
Inventor 王海峰胡晓光于佃海
Owner BEIJING BAIDU NETCOM SCI & TECH CO LTD