Welcome to the new version of CaltechAUTHORS. Login is currently restricted to library staff. If you notice any issues, please email coda@library.caltech.edu
Published June 11, 2014 | Published
Journal Article Open

Event processing time prediction at the CMS experiment of the Large Hadron Collider

Abstract

The physics event reconstruction is one of the biggest challenges for the computing of the LHC experiments. Among the different tasks that computing systems of the CMS experiment performs, the reconstruction takes most of the available CPU resources. The reconstruction time of single collisions varies according to event complexity. Measurements were done in order to determine this correlation quantitatively, creating means to predict it based on the data-taking conditions of the input samples. Currently the data processing system splits tasks in groups with the same number of collisions and does not account for variations in the processing time. These variations can be large and can lead to a considerable increase in the time it takes for CMS workflows to finish. The goal of this study was to use estimates on processing time to more efficiently split the workflow into jobs. By considering the CPU time needed for each job the spread of the job-length distribution in a workflow is reduced.

Additional Information

© 2013. Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. Published under licence by IOP Publishing Ltd. CMS Tier 0 Team, CMS Workload Management Development Team, WLCG DashBoard Team, UERJ Department of High Energy Physics, US-CMS group at the California Institute of Technology, US-CMS group at the Fermi National Accelerator Laboratory.

Attached Files

Published - 1742-6596_513_3_032023.pdf

Files

1742-6596_513_3_032023.pdf
Files (7.0 MB)
Name Size Download all
md5:076f14aea3164b7d8917ea3d0a279619
7.0 MB Preview Download

Additional details

Created:
August 20, 2023
Modified:
October 18, 2023