GPU Accelerating Statistical Model Checking for Extended Timed Automata

Oliver S. Bak, Mathias W.B. Christiansen, Oliver V. Eriksen, Sergio Feo-Arenis, Peter G. Jensen, Marcus D. Jensen, Simas Juozapaitis, Kim G. Larsen, Marius Mikučionis*, Marco Muñiz, Andreas Windfeld

*Corresponding author for this work

Research output: Contribution to book/anthology/report/conference proceedingBook chapterResearchpeer-review

Abstract

The core component of Statistical Model Checking (SMC) is the repeated sampling of a given system as to estimate statistical measures. To obtain probabilistic estimates with high confidence a significant number of simulations is required, in particular in the presence of rare events. In this paper we explore the use of Graphical Processing Unit (GPU) for accelerating SMC for Networks of Stochastic Extended Timed Automata (SXTA). We discuss the many challenges and solutions required to achieve significant speedups on a GPU architecture. In collaboration with NVIDIA we develop a prototype tool for parallel SMC using both GPU and multi-core CPU. Experimental results demonstrate trade-offs in the computation time when utilizing either CPU or GPU. In one case we observed the GPU using 20% of the power of the CPU equivalent while delivering a 2.73 time speedup.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
EditorsNils Jansen, Sebastian Junges, Benjamin Lucien Kaminski, Christoph Matheja, Thomas Noll, Tim Quatmann, Mariëlle Stoelinga, Matthias Volk
Number of pages26
Volume15261
PublisherSpringer Science+Business Media
Publication date2025
Pages267-292
ISBN (Print)978-3-031-75774-7
ISBN (Electronic)978-3-031-75775-4
DOIs
Publication statusPublished - 2025
SeriesLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume15261 LNCS
ISSN0302-9743

Bibliographical note

Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Fingerprint

Dive into the research topics of 'GPU Accelerating Statistical Model Checking for Extended Timed Automata'. Together they form a unique fingerprint.

Cite this