Deep learning-based operation acceleration method, device and system

In order to solve the problems in the prior art, the invention innovatively provides a deep learning-based operation acceleration method. The method comprises the following steps: counting the operation time of each operation layer in an AI algorithm in deep learning; carrying out acceleration opera...

Full description

Saved in:
Bibliographic Details
Main Author HE HAOYI
Format Patent
LanguageChinese
English
Published 22.11.2019
Subjects
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
ELECTRIC DIGITAL DATA PROCESSING
PHYSICS
Online AccessGet full text

Cover

More Information
Summary:In order to solve the problems in the prior art, the invention innovatively provides a deep learning-based operation acceleration method. The method comprises the following steps: counting the operation time of each operation layer in an AI algorithm in deep learning; carrying out acceleration operation on at least one operation layer by each acceleration card, adopting the operation result of thecurrent acceleration card as the input and transmitting the input to the next acceleration card to carry out acceleration operation. The invention further provides an operation acceleration device and system based on deep learning. The problem of slow time caused by processing all operation layers in the AI algorithm in deep learning by a single accelerator card is effectively solved. Therefore,the operation efficiency is effectively improved. The acceleration operation time of each acceleration card is the same. The problem that operation congestion is easily caused if the acceleration operation time of different ac
Bibliography:Application Number: CN201910681089