Architectural Implications of GNN Aggregation Programming Abstractions

Graph neural networks (GNNs) have gained significant popularity due to the powerful capability to extract useful representations from graph data. As the need for efficient GNN computation intensifies, a variety of programming abstractions designed for optimizing GNN Aggregation have emerged to facil...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE computer architecture letters Jg. 23; H. 1; S. 125 - 128
Hauptverfasser: Qi, Yingjie, Yang, Jianlei, Zhou, Ao, Qiao, Tong, Hu, Chunming
Format: Journal Article
Sprache:Englisch
Veröffentlicht: New York IEEE 01.01.2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Schlagworte:
characterization
execution patterns
Graph neural networks
Graph neural networks (GNNs)
Graphical representations
Graphics processing units
Indexes
Kernel
Organizations
Programming
programming abstractions
Taxonomy
ISSN:1556-6056, 1556-6064
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Graph neural networks (GNNs) have gained significant popularity due to the powerful capability to extract useful representations from graph data. As the need for efficient GNN computation intensifies, a variety of programming abstractions designed for optimizing GNN Aggregation have emerged to facilitate acceleration. However, there is no comprehensive evaluation and analysis upon existing abstractions, thus no clear consensus on which approach is better. In this letter, we classify existing programming abstractions for GNN Aggregation by the dimension of data organization and propagation method. By constructing these abstractions on a state-of-the-art GNN library, we perform a thorough and detailed characterization study to compare their performance and efficiency, and provide several insights on future GNN acceleration based on our analysis.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1556-6056
1556-6064
DOI:10.1109/LCA.2023.3326170