Cargando…

Embedded Neural Networks on FPGAs for Real-Time Computation of the Energy Deposited in the ATLAS Liquid Argon Calorimeter

At the HL-LHC, the number of proton-proton collisions in one bunch-crossing (called pileup) increases significantly, putting more stringent requirements on the LHC detectors electronics and real-time data-processing capabilities. The ATLAS LAr calorimeter measures with an excellent resolution the en...

Descripción completa

Detalles Bibliográficos
Autor principal: Aad, Georges
Lenguaje:eng
Publicado: 2022
Materias:
Acceso en línea:http://cds.cern.ch/record/2834032
_version_ 1780975562683056128
author Aad, Georges
author_facet Aad, Georges
author_sort Aad, Georges
collection CERN
description At the HL-LHC, the number of proton-proton collisions in one bunch-crossing (called pileup) increases significantly, putting more stringent requirements on the LHC detectors electronics and real-time data-processing capabilities. The ATLAS LAr calorimeter measures with an excellent resolution the energy of particles produced in LHC collisions. The energy is computed in real-time using optimal filtering (OF) algorithms running on dedicated data-acquisition electronic boards based on FPGAs. However, with the increased pileup, the performance of these algorithms decreases significantly. Dedicated Neural networks (NNs) are found to outperform the OF algorithms. The architecture, performance, and firmware implementation for these NNs will be presented.
id cern-2834032
institution Organización Europea para la Investigación Nuclear
language eng
publishDate 2022
record_format invenio
spelling cern-28340322023-06-12T07:28:38Zhttp://cds.cern.ch/record/2834032engAad, GeorgesEmbedded Neural Networks on FPGAs for Real-Time Computation of the Energy Deposited in the ATLAS Liquid Argon CalorimeterParticle Physics - ExperimentAt the HL-LHC, the number of proton-proton collisions in one bunch-crossing (called pileup) increases significantly, putting more stringent requirements on the LHC detectors electronics and real-time data-processing capabilities. The ATLAS LAr calorimeter measures with an excellent resolution the energy of particles produced in LHC collisions. The energy is computed in real-time using optimal filtering (OF) algorithms running on dedicated data-acquisition electronic boards based on FPGAs. However, with the increased pileup, the performance of these algorithms decreases significantly. Dedicated Neural networks (NNs) are found to outperform the OF algorithms. The architecture, performance, and firmware implementation for these NNs will be presented.ATL-LARG-SLIDE-2022-496oai:cds.cern.ch:28340322022-09-21
spellingShingle Particle Physics - Experiment
Aad, Georges
Embedded Neural Networks on FPGAs for Real-Time Computation of the Energy Deposited in the ATLAS Liquid Argon Calorimeter
title Embedded Neural Networks on FPGAs for Real-Time Computation of the Energy Deposited in the ATLAS Liquid Argon Calorimeter
title_full Embedded Neural Networks on FPGAs for Real-Time Computation of the Energy Deposited in the ATLAS Liquid Argon Calorimeter
title_fullStr Embedded Neural Networks on FPGAs for Real-Time Computation of the Energy Deposited in the ATLAS Liquid Argon Calorimeter
title_full_unstemmed Embedded Neural Networks on FPGAs for Real-Time Computation of the Energy Deposited in the ATLAS Liquid Argon Calorimeter
title_short Embedded Neural Networks on FPGAs for Real-Time Computation of the Energy Deposited in the ATLAS Liquid Argon Calorimeter
title_sort embedded neural networks on fpgas for real-time computation of the energy deposited in the atlas liquid argon calorimeter
topic Particle Physics - Experiment
url http://cds.cern.ch/record/2834032
work_keys_str_mv AT aadgeorges embeddedneuralnetworksonfpgasforrealtimecomputationoftheenergydepositedintheatlasliquidargoncalorimeter