Topic mining and forecasting on patent map for GPU technology

In recent years, applications in fields of commercial computing, scientific computing and artificial intelligence have been booming. Compared with traditional computing platforms, GPUs have advantages for data centers such as high concurrent processing capacity and high bandwidth transmission effici...

Full description

Saved in:
Bibliographic Details
Published in:The Journal of supercomputing Vol. 81; no. 5; p. 710
Main Authors: Hou, Ke, Wu, Mengying, Wu, Wenjun, Huang, Linhao
Format: Journal Article
Language:English
Published: New York Springer US 09.04.2025
Springer Nature B.V
Subjects:
Artificial intelligence
Bandwidths
Blockchain
Cloud computing
Cluster analysis
Clustering
Compilers
Computer centers
Computer Science
Concurrent processing
Cooling
Design optimization
Efficiency
Graphic design
Graphics processing units
High performance computing
Innovations
Interpreters
Machine learning
Patent applications
Processor Architectures
Programming Languages
R&D
Research & development
Statistical analysis
Technology
Transmission efficiency
Trends
Technology forecasting
Patent map
GPU
Machine learning
ISSN:1573-0484, 0920-8542, 1573-0484
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In recent years, applications in fields of commercial computing, scientific computing and artificial intelligence have been booming. Compared with traditional computing platforms, GPUs have advantages for data centers such as high concurrent processing capacity and high bandwidth transmission efficiency. It can significantly improve system performance and has received widespread attention. However, GPUs also face unprecedented challenges in high-performance computing, architectural innovation, energy efficiency and thermal optimization. To accelerate the process of GPU performance improvement, it is very important to carry out patent analysis, evolution path research and development trend prediction of GPU technology. Firstly, this paper provides an in-depth analysis of the development status of GPU technology based on patent maps. Secondly, the BERTopic model is used to cluster the patent technology topics to obtain four technology topics in the field of GPU technology. Finally, Logistic model is used to analyze and predict the technology life cycle of the identified technology themes. The results show that from 2005 to 2023, the number of GPU patent applications generally shows an upward trend, experiencing three stages of slow growth, rapid growth and relatively stable fluctuation. The scope of GPU technology applications covers a wide range of fields from high-end gaming and professional graphic design to artificial intelligence and cloud computing. The four technology themes obtained through BERTopic model clustering analysis are: graphics processing and GPU foundation, CPU and GPU collaborative computing, GPU hardware design and optimization, and GPU comprehensive application and cross-fertilization. The initial three themes were analyzed and predicted in terms of the technology life cycle. It is anticipated that within the forthcoming years, these themes will attain maturity or decline, correspondingly, and technological innovation activities will persist.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1573-0484
0920-8542
1573-0484
DOI:10.1007/s11227-025-07215-9