Application Failures and Machine Computational Efficiency

DOI10.1137/1.9781611979022.3MaRDI QIDQ7284734arXivFDO

Authors Carlo Graziani, Bethany Lusch, O. E. Bronson Messer

Publication date 5 August 2026

Published in arXiv

Tldr We introduce a framework to measure uptime efficiency for heterogeneous systems under non-negligible application failure rates, accounting for checkpointing and restart costs, and demonstrate its application on real-world data from the Frontier supercomputer.

Full work available at URL https://doi.org/10.1137/1.9781611979022.3

arXiv classification

cs.DC

Cites work

• A first order approximation to the optimum checkpoint interval
• A Primer on Statistical Distributions

This page was built for publication: Application Failures and Machine Computational Efficiency

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7284734)