Gath-Geva clustering algorithm for high performance computing (HPC) monitoring

Supercomputer is a fundamental means to perform complex and huge computations. Simultaneously, it is one of the most energy consuming infrastructures. The diversity of applications executed within a HPC system makes it difficult to control resource utilization and identify the behaviour of these app...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2019 Third International Conference on Intelligent Computing in Data Sciences (ICDS) S. 1 - 6
Hauptverfasser: El Motaki, Saloua, Yahyaouy, Ali, Gualous, Hamid, Sabor, Jalal
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.10.2019
Schlagworte:
Application identification
Clustering algorithms
Gath-Geva clustering
HPC
Machine learning algorithms
Measurement
Monitoring
Partitioning algorithms
Quality of service
Resilience
Resource management
Shape
Supercomputer
Supercomputers
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Supercomputer is a fundamental means to perform complex and huge computations. Simultaneously, it is one of the most energy consuming infrastructures. The diversity of applications executed within a HPC system makes it difficult to control resource utilization and identify the behaviour of these applications while running. To effectively alleviate this concern, scientists have appealed to use machine learning techniques for HPC monitoring and diagnosis. This work focuses on the employment of Gath-Geva clustering algorithm to identify applications and their behavioural similarities while running on HPC system. The choice of this algorithm is based on its ability to be adapted to data structures of arbitrarily shaped, sized and dense data.
DOI:10.1109/ICDS47004.2019.8942390