Impact of thermal throttling on long-term visual inference in a cpu-based edge device

Many application scenarios of edge visual inference, e.g., robotics or environmental monitoring, eventually require long periods of continuous operation. In such periods, the processor temperature plays a critical role to keep a prescribed frame rate. Particularly, the heavy computational load of co...

ver descrição completa

Detalhes bibliográficos
Autores: Benoit-Cattin, Théo, Velasco Montero, Delia, Fernández Berni, Jorge
Formato: artículo
Estado:Versión publicada
Fecha de publicación:2020
País:España
Recursos:Universidad de Sevilla (US)
Repositorio:idUS. Depósito de Investigación de la Universidad de Sevilla
OAI Identifier:oai:idus.us.es:11441/128911
Acesso em linha:https://hdl.handle.net/11441/128911
https://doi.org/10.3390/electronics9122106
Access Level:acceso abierto
Palavra-chave:Ambient conditions
Convolutional neural networks
Edge vision
Long-term inference
Raspberry Pi
Thermal throttling
id ES_afcb5107f6d78953b499af4e0a1b8b7c
oai_identifier_str oai:idus.us.es:11441/128911
network_acronym_str ES
network_name_str España
repository_id_str
spelling Impact of thermal throttling on long-term visual inference in a cpu-based edge deviceBenoit-Cattin, ThéoVelasco Montero, DeliaFernández Berni, JorgeAmbient conditionsConvolutional neural networksEdge visionLong-term inferenceRaspberry PiThermal throttlingMany application scenarios of edge visual inference, e.g., robotics or environmental monitoring, eventually require long periods of continuous operation. In such periods, the processor temperature plays a critical role to keep a prescribed frame rate. Particularly, the heavy computational load of convolutional neural networks (CNNs) may lead to thermal throttling and hence performance degradation in few seconds. In this paper, we report and analyze the long-term performance of 80 different cases resulting from running five CNN models on four software frameworks and two operating systems without and with active cooling. This comprehensive study was conducted on a low-cost edge platform, namely Raspberry Pi 4B (RPi4B), under stable indoor conditions. The results show that hysteresis-based active cooling prevented thermal throttling in all cases, thereby improving the throughput up to approximately 90% versus no cooling. Interestingly, the range of fan usage during active cooling varied from 33% to 65%. Given the impact of the fan on the power consumption of the system as a whole, these results stress the importance of a suitable selection of CNN model and software components. To assess the performance in outdoor applications, we integrated an external temperature sensor with the RPi4B and conducted a set of experiments with no active cooling in a wide interval of ambient temperature, ranging from 22 ºC to 36 ºC. Variations up to 27.7% were measured with respect to the maximum throughput achieved in that interval. This demonstrates that ambient temperature is a critical parameter in case active cooling cannot be applied.Ministerio de Ciencia e Innovación RTI2018-097088-B-C31European Union 765866US Office of Naval Research N00014-19-1-2156Multidisciplinary Digital Publishing Institute (MDPI)Electrónica y Electromagnetismo2020info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionapplication/pdfapplication/pdfhttps://hdl.handle.net/11441/128911https://doi.org/10.3390/electronics9122106reponame:idUS. Depósito de Investigación de la Universidad de Sevillainstname:Universidad de Sevilla (US)InglésElectronics, 9 (12), 2106.RTI2018-097088-B-C31765866N00014-19-1-2156https://doi.org/10.3390/electronics9122106info:eu-repo/semantics/openAccessoai:idus.us.es:11441/1289112026-06-17T12:51:07Z
dc.title.none.fl_str_mv Impact of thermal throttling on long-term visual inference in a cpu-based edge device
title Impact of thermal throttling on long-term visual inference in a cpu-based edge device
spellingShingle Impact of thermal throttling on long-term visual inference in a cpu-based edge device
Benoit-Cattin, Théo
Ambient conditions
Convolutional neural networks
Edge vision
Long-term inference
Raspberry Pi
Thermal throttling
title_short Impact of thermal throttling on long-term visual inference in a cpu-based edge device
title_full Impact of thermal throttling on long-term visual inference in a cpu-based edge device
title_fullStr Impact of thermal throttling on long-term visual inference in a cpu-based edge device
title_full_unstemmed Impact of thermal throttling on long-term visual inference in a cpu-based edge device
title_sort Impact of thermal throttling on long-term visual inference in a cpu-based edge device
dc.creator.none.fl_str_mv Benoit-Cattin, Théo
Velasco Montero, Delia
Fernández Berni, Jorge
author Benoit-Cattin, Théo
author_facet Benoit-Cattin, Théo
Velasco Montero, Delia
Fernández Berni, Jorge
author_role author
author2 Velasco Montero, Delia
Fernández Berni, Jorge
author2_role author
author
dc.contributor.none.fl_str_mv Electrónica y Electromagnetismo
dc.subject.none.fl_str_mv Ambient conditions
Convolutional neural networks
Edge vision
Long-term inference
Raspberry Pi
Thermal throttling
topic Ambient conditions
Convolutional neural networks
Edge vision
Long-term inference
Raspberry Pi
Thermal throttling
description Many application scenarios of edge visual inference, e.g., robotics or environmental monitoring, eventually require long periods of continuous operation. In such periods, the processor temperature plays a critical role to keep a prescribed frame rate. Particularly, the heavy computational load of convolutional neural networks (CNNs) may lead to thermal throttling and hence performance degradation in few seconds. In this paper, we report and analyze the long-term performance of 80 different cases resulting from running five CNN models on four software frameworks and two operating systems without and with active cooling. This comprehensive study was conducted on a low-cost edge platform, namely Raspberry Pi 4B (RPi4B), under stable indoor conditions. The results show that hysteresis-based active cooling prevented thermal throttling in all cases, thereby improving the throughput up to approximately 90% versus no cooling. Interestingly, the range of fan usage during active cooling varied from 33% to 65%. Given the impact of the fan on the power consumption of the system as a whole, these results stress the importance of a suitable selection of CNN model and software components. To assess the performance in outdoor applications, we integrated an external temperature sensor with the RPi4B and conducted a set of experiments with no active cooling in a wide interval of ambient temperature, ranging from 22 ºC to 36 ºC. Variations up to 27.7% were measured with respect to the maximum throughput achieved in that interval. This demonstrates that ambient temperature is a critical parameter in case active cooling cannot be applied.
publishDate 2020
dc.date.none.fl_str_mv 2020
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
format article
status_str publishedVersion
dc.identifier.none.fl_str_mv https://hdl.handle.net/11441/128911
https://doi.org/10.3390/electronics9122106
url https://hdl.handle.net/11441/128911
https://doi.org/10.3390/electronics9122106
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.relation.none.fl_str_mv Electronics, 9 (12), 2106.
RTI2018-097088-B-C31
765866
N00014-19-1-2156
https://doi.org/10.3390/electronics9122106
dc.rights.none.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
application/pdf
dc.publisher.none.fl_str_mv Multidisciplinary Digital Publishing Institute (MDPI)
publisher.none.fl_str_mv Multidisciplinary Digital Publishing Institute (MDPI)
dc.source.none.fl_str_mv reponame:idUS. Depósito de Investigación de la Universidad de Sevilla
instname:Universidad de Sevilla (US)
instname_str Universidad de Sevilla (US)
reponame_str idUS. Depósito de Investigación de la Universidad de Sevilla
collection idUS. Depósito de Investigación de la Universidad de Sevilla
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869416714780803072
score 15,301629

Registos relacionados