| Issue |
EPJ Web of Conf.
Volume 295, 2024
26th International Conference on Computing in High Energy and Nuclear Physics (CHEP 2023)
|
|
|---|---|---|
| Article Number | 11007 | |
| Number of page(s) | 8 | |
| Section | Heterogeneous Computing and Accelerators | |
| DOI | https://doi.org/10.1051/epjconf/202429511007 | |
| Published online | 06 May 2024 | |
https://doi.org/10.1051/epjconf/202429511007
ARMing HEP for the future Energy Efficiency of WLCG sites (ARM vs. x86)
School of Physics and Astronomy, University of Glasgow Kelvin Building, University Avenue, Glasgow, G12 8QQ, United Kingdom
* e-mail: emanuele.simili@glasgow.ac.uk
Published online: 6 May 2024
We present a case for ARM chips as an alternative to standard x86 at WLCG sites to help reduce power consumption. New measurements are presented on the performance and energy consumption of two machines (one ARM and one x86), that were otherwise similar in specification and cost. The comparison was extended to a dual socket x86 node, representative of our site. These new results include the energy-efficiency and speed of singleand multithreaded jobs; the effect of hyper-threading; and an initial look at clock throttling as a way of shaping power-load. We observe significantly lower power consumption and often slightly better performance on the ARM machine and, noting the increased availability of ARM software builds from all LHC experiments and beyond, we plan to install a 2k-core ARM cluster at our WLCG Tier2 site at Glasgow in the summer of 2023. This will enable testing, physicsvalidation, and eventually an ARM production environment that will inform and influence other WLCG sites in the UK and worldwide.
© The Authors, published by EDP Sciences, 2024
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.