GPU Accelerating Statistical Model Checking for Extended Timed Automata

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), ISSN: 1611-3349, Vol: 15261 LNCS, Page: 267-292
2025
  • 0
    Citations
  • 0
    Usage
  • 0
    Captures
  • 0
    Mentions
  • 0
    Social Media
Metric Options:   Counts1 Year3 Year

Book Chapter Description

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.

Bibliographic Details

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

Springer Science and Business Media LLC

Mathematics; Computer Science

Provide Feedback

Have ideas for a new metric? Would you like to see something else here?Let us know