Mostrar el registro sencillo del ítem

Artículo

dc.creatorVelasco Montero, Deliaes
dc.creatorGoossens, Bartes
dc.creatorFernández Berni, Jorgees
dc.creatorRodríguez Vázquez, Ángel Benitoes
dc.creatorPhilips, Wilfriedes
dc.date.accessioned2024-02-16T14:47:05Z
dc.date.available2024-02-16T14:47:05Z
dc.date.issued2023
dc.identifier.citationVelasco Montero, D., Goossens, B., Fernández Berni, J., Rodríguez Vázquez, Á.B. y Philips, W. (2023). A Pipelining-Based Heterogeneous Scheduling and Energy-Throughput Optimization Scheme for CNNs Leveraging Apache TVM. IEEE Access, 11, 35007-35021. https://doi.org/10.1109/ACCESS.2023.3264828.
dc.identifier.issn2169-3536es
dc.identifier.urihttps://hdl.handle.net/11441/155304
dc.description.abstractExtracting information of interest from continuous video streams is a strongly demanded computer vision task. For the realization of this task at the edge using the current de-facto standard approach, i.e., deep learning, it is critical to optimize key performance metrics such as throughput and energy consumption according to prescribed application requirements. This allows achieving timely decision-making while extending the battery lifetime as much as possible. In this context, we propose a method to boost neural-network performance based on a co-execution strategy that exploits hardware heterogeneity on edge platforms. The enabling tool is Apache TVM, a highly efficient machine-learning compiler compatible with a diversity of hardware back-ends. The proposed approach solves the problem of network partitioning and distributes the workloads to make concurrent use of all the processors available on the board following a pipeline scheme. We conducted experiments on various popular CNNs compiled with TVM on the Jetson TX2 platform. The experimental results based on measurements show a significant improvement in throughput with respect to a single-processor execution, ranging from 14% to 150% over all tested networks. Power-efficient configurations were also identified, accomplishing energy reductions above 10%.es
dc.description.sponsorshipFlemish Government PID2021-128009OB-C31es
dc.description.sponsorshipMinisterio de Ciencia e Innovación TED2021-131835B-I00es
dc.description.sponsorshipOffice of Naval Research N00014-19-1-2156es
dc.formatapplication/pdfes
dc.format.extent15 p.es
dc.language.isoenges
dc.publisherIEEEes
dc.relation.ispartofIEEE Access, 11, 35007-35021.
dc.rightsAttribution-NonCommercial-NoDerivatives 4.0 Internacional*
dc.rights.urihttp://creativecommons.org/licenses/by-nc-nd/4.0/*
dc.subjectApache TVMes
dc.subjectContinuous inferencees
dc.subjectConvolutional neural networkses
dc.subjectEdge visiones
dc.subjectHeterogeneous processinges
dc.subjectJetson TX2es
dc.subjectPerformance optimizationes
dc.titleA Pipelining-Based Heterogeneous Scheduling and Energy-Throughput Optimization Scheme for CNNs Leveraging Apache TVMes
dc.typeinfo:eu-repo/semantics/articlees
dc.type.versioninfo:eu-repo/semantics/publishedVersiones
dc.rights.accessRightsinfo:eu-repo/semantics/openAccesses
dc.contributor.affiliationUniversidad de Sevilla. Departamento de Electrónica y Electromagnetismoes
dc.relation.projectIDPID2021-128009OB-C31es
dc.relation.projectIDTED2021-131835B-I00es
dc.relation.projectIDN00014-19-1-2156es
dc.relation.publisherversionhttps://dx.doi.org/10.1109/ACCESS.2023.3264828es
dc.identifier.doi10.1109/ACCESS.2023.3264828es
dc.journaltitleIEEE Accesses
dc.publication.volumen11es
dc.publication.initialPage35007es
dc.publication.endPage35021es
dc.contributor.funderFlemish Government (Belgium)es
dc.contributor.funderMinisterio de Ciencia e Innovación (MICIN). Españaes
dc.contributor.funderOffice of Naval Research (ONR). United Stateses

FicherosTamañoFormatoVerDescripción
A Pipelining-Based Heterogeneo ...2.406MbIcon   [PDF] Ver/Abrir  

Este registro aparece en las siguientes colecciones

Mostrar el registro sencillo del ítem

Attribution-NonCommercial-NoDerivatives 4.0 Internacional
Excepto si se señala otra cosa, la licencia del ítem se describe como: Attribution-NonCommercial-NoDerivatives 4.0 Internacional