Evaluating the soft error sensitivity of a GPU-based SoC for matrix multiplication

León, Germán; Badía, Jose M.; Belloch Rodríguez, José Antonio; Lindoso Muñoz, Almudena; Entrena Arrontes, Luis Alfonso

Publication:
Evaluating the soft error sensitivity of a GPU-based SoC for matrix multiplication

dc.affiliation.dpto	UC3M. Departamento de Tecnología Electrónica	es
dc.affiliation.grupoinv	UC3M. Grupo de Investigación: Diseño Microelectrónico y Aplicaciones (DMA)	es
dc.contributor.author	León, Germán
dc.contributor.author	Badía, Jose M.
dc.contributor.author	Belloch Rodríguez, José Antonio
dc.contributor.author	Lindoso Muñoz, Almudena
dc.contributor.author	Entrena Arrontes, Luis Alfonso
dc.contributor.funder	Ministerio de Economía y Competitividad (España)	es
dc.contributor.funder	Ministerio de Ciencia y Tecnología (España)	es
dc.date.accessioned	2022-09-01T09:44:11Z
dc.date.available	2022-11-01T00:00:07Z
dc.date.issued	2020-11
dc.description	Proceeding of: 31th European Symposium on Reliability of Electron Devices, Failure Physics and Analysis (ESREF 2020), Athens, Greece, 4th to 8 October 2020 (Virtual conference)	en
dc.description.abstract	System-on-Chip (SoC) devices can be composed of low-power multicore processors combined with a small graphics accelerator (or GPU) which offers a trade-off between computational capacity and low-power consumption. In this work we use the LLFI-GPU fault injection tool on one of these devices to compare the sensitivity to soft errors of two different CUDA versions of matrix multiplication benchmark. Specifically, we perform fault injection campaigns on a Jetson TK1 development kit, a board equipped with a SoC including an NVIDIA 'Kepler” Graphics Processing Unit (GPU). We evaluate the effect of modifying the size of the problem and also the thread-block size on the behaviour of the algorithms. Our results show that the block version of the matrix multiplication benchmark that leverages the shared memory of the GPU is not only faster than the element-wise version, but it is also much more resilient to soft errors. We also use the cuda-gdb debugger to analyze the main causes of the crashes in the code due to soft errors. Our experiments show that most of the errors are due to accesses to invalid positions of the different memories of the GPU, which causes that the block version suffers a higher percentage of this kind of errors.	en
dc.description.sponsorship	This work has been supported by the Spanish Government through TIN2017-82972-R and ESP2015-68245-C4-1-P, and by the Valencian Regional Government through PROMETEO/2019/109.	en
dc.format.extent	6	es
dc.identifier.bibliographicCitation	Germán, L., et al. Evaluating the soft error sensitivity of a GPU-based SoC for matrix multiplication. Microelectronics reliability, Vol. 114, 113856 , (Special issue: Proceedings of ESREF 2020, 31th European Symposium on Reliability of Electron Devices, Failure Physics and Analysis). Elsevier, 2020, 6 p.	en
dc.identifier.doi	https://doi.org/10.1016/j.microrel.2020.113856
dc.identifier.issn	0026-2714
dc.identifier.publicationfirstpage	1	es
dc.identifier.publicationlastpage	6	es
dc.identifier.publicationtitle	Microelectronics reliability (Special issue: Proceedings of ESREF 2020, 31th European Symposium on Reliability of Electron Devices, Failure Physics and Analysis)	en
dc.identifier.publicationvolume	114, 113856	es
dc.identifier.uri	https://hdl.handle.net/10016/35621
dc.identifier.uxxi	CC/0000033574
dc.language.iso	eng	en
dc.publisher	Elsevier Ltd.	en
dc.relation.eventdate	2020-10-04	es
dc.relation.eventplace	Atenas, GRECIA (Conferencia virtual)	es
dc.relation.eventtitle	31th European Symposium on Reliability of Electron Devices, Failure Physics and Analysis (ESREF 2020)	en
dc.relation.projectID	Gobierno de España. TIN2017-82972-R	es
dc.relation.projectID	Gobierno de España. ESP2015-68245-C4-1-P	es
dc.rights	© 2020 Elsevier Ltd. All rights reserved.	en
dc.rights	This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.	en
dc.rights	Atribución-NoComercial-SinDerivadas 3.0 España	*
dc.rights.accessRights	open access	en
dc.rights.uri	http://creativecommons.org/licenses/by-nc-nd/3.0/es/	*
dc.subject.eciencia	Electrónica	es
dc.subject.other	GPU	en
dc.subject.other	Soft errors	en
dc.subject.other	Sensitivity	en
dc.subject.other	Fault injection	en
dc.title	Evaluating the soft error sensitivity of a GPU-based SoC for matrix multiplication	en
dc.type	conference paper	*
dc.type.hasVersion	AM	*
dspace.entity.type	Publication