Cargando…
Power Consumption Optimization for LHCb Online and DAQ System
The optimization of energy efficiency and power consumption of the LHCb Online System's computer servers is an important aspect for future datacenter upgrades. Since all the servers are typically operated in maximum performance mode, more energy-efficient configurations could highly impact powe...
Autor principal: | |
---|---|
Lenguaje: | eng |
Publicado: |
2023
|
Materias: | |
Acceso en línea: | http://cds.cern.ch/record/2869770 |
_version_ | 1780978306378629120 |
---|---|
author | Leandro Flores, Matias Ernesto |
author_facet | Leandro Flores, Matias Ernesto |
author_sort | Leandro Flores, Matias Ernesto |
collection | CERN |
description | The optimization of energy efficiency and power consumption of the LHCb Online System's computer servers is an important aspect for future datacenter upgrades. Since all the servers are typically operated in maximum performance mode, more energy-efficient configurations could highly impact power consumption, especially due to the fans working at full duty all the time. There is also consideration of evaluating alternate cooling solutions and new server technologies for their future implementations on different parts of the Online System. In this work, general system power efficiency and a performance metric are established to compare different server configurations, direct liquid cold-plate cooling is tested for its possible viability as a new cooling solution, and a new energy-efficient configuration is created and executed, by controlling the fans' duty based on input mainly from the temperature of the essential PCIe40 boards. The methodology to gather system power efficiency and performance of servers proves successful to compare different server designs, while direct liquid cold-plate cooling is shown to be a potential alternative for a more energy-efficient cooling solution, and the new fan configuration proves to save up a significant amount of power compared to the typical maximum performance configuration. |
id | cern-2869770 |
institution | Organización Europea para la Investigación Nuclear |
language | eng |
publishDate | 2023 |
record_format | invenio |
spelling | cern-28697702023-09-08T19:08:43Zhttp://cds.cern.ch/record/2869770engLeandro Flores, Matias ErnestoPower Consumption Optimization for LHCb Online and DAQ SystemComputing and ComputersEngineeringThe optimization of energy efficiency and power consumption of the LHCb Online System's computer servers is an important aspect for future datacenter upgrades. Since all the servers are typically operated in maximum performance mode, more energy-efficient configurations could highly impact power consumption, especially due to the fans working at full duty all the time. There is also consideration of evaluating alternate cooling solutions and new server technologies for their future implementations on different parts of the Online System. In this work, general system power efficiency and a performance metric are established to compare different server configurations, direct liquid cold-plate cooling is tested for its possible viability as a new cooling solution, and a new energy-efficient configuration is created and executed, by controlling the fans' duty based on input mainly from the temperature of the essential PCIe40 boards. The methodology to gather system power efficiency and performance of servers proves successful to compare different server designs, while direct liquid cold-plate cooling is shown to be a potential alternative for a more energy-efficient cooling solution, and the new fan configuration proves to save up a significant amount of power compared to the typical maximum performance configuration.CERN-STUDENTS-Note-2023-119oai:cds.cern.ch:28697702023-09-08 |
spellingShingle | Computing and Computers Engineering Leandro Flores, Matias Ernesto Power Consumption Optimization for LHCb Online and DAQ System |
title | Power Consumption Optimization for LHCb Online and DAQ System |
title_full | Power Consumption Optimization for LHCb Online and DAQ System |
title_fullStr | Power Consumption Optimization for LHCb Online and DAQ System |
title_full_unstemmed | Power Consumption Optimization for LHCb Online and DAQ System |
title_short | Power Consumption Optimization for LHCb Online and DAQ System |
title_sort | power consumption optimization for lhcb online and daq system |
topic | Computing and Computers Engineering |
url | http://cds.cern.ch/record/2869770 |
work_keys_str_mv | AT leandrofloresmatiasernesto powerconsumptionoptimizationforlhcbonlineanddaqsystem |