Jobs runtime forecast for JSCC RAS supercomputers using machine learning methods

DOI10.1134/S1995080220120343MaRDI QIDQ2225846zbMATH OpenOpenAlexFDO

Authors G. I. Savin, B. M. Shabanov, D. S. Nikolaev, A. V. Baranov, P. N. Telegin

Publication date 11 February 2021

Published in Lobachevskii Journal of Mathematics (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1134/s1995080220120343

zbMATH Keywords

machine learningsupercomputerrandom foresthigh-performance computingjob management systemjob runtime predictionsupercomputer job schedulingSUPPZ

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05) Performance evaluation, queueing, and scheduling in the context of computer systems (68M20)

Recommendations

• Performance problems of using system-predicted runtimes for parallel job scheduling
• Predicting application run times with historical information
• Application of neural-network methods for prediction of the job execution characteristics in multiprocessor computing systems
• Using moldability to improve the performance of supercomputer jobs
• Group based job scheduling to increase the high-performance computing efficiency

Cites work

• scientific article; zbMATH DE number 2118472 (Why is no real title available?)
• Data Mining with Decision Trees
• Greedy function approximation: A gradient boosting machine.
• Joint supercomputer center of the Russian Academy of Sciences: present and future
• Random forests

Cited in

(2)

• Group based job scheduling to increase the high-performance computing efficiency
• Surprise event detection of the supercomputer execution queues

Describes a project that uses

Uses Software

• Slurm

This page was built for publication: Jobs runtime forecast for JSCC RAS supercomputers using machine learning methods

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2225846)