A new level-set analysis and sparse storage format for the SPTRSV in GPUs

Freire, Manuel - Dufrechou, Ernesto - Ezzatti, Pablo

Resumen:

Due to its relevant role in many numerical methods, the solution of sparse triangular linear systems (SpTRSV) in parallel platforms is continuously studied to extract as much performance as possible from the latest hardware architectures. In the case of GPUs, the latest solvers use the synchronization-free paradigm. When the problem involves several system solutions for the same matrix, they often pre-process it through a levelset analysis to improve the equation solution scheduling in the solution phase. In addition, other optimizations address the load balancing issues and irregular memory access of the SpTRSV. In this work, we modify the classical approach to compute the level sets used in the parallel SpTRSV computation, and we show that the new strategy generally reduces the computation time of the solver. Furthermore, we design an internal matrix representation that can significantly accelerate the solution stage at the cost of increasing the memory storage requirements of the algorithm. The experimental evaluation shows that the proposed modifications can improve the performance of a recent levelset and synchronization-free solver by up to 70%, significantly outperforming other state-of-the-art solvers, especially when several linear systems must be solved for each analysis phase.

Detalles Bibliográficos
2024
FCE_3_2022_1_172419 - MODELAR: Modelado del desempeñO de métoDos numÉricos en pLataformas de hArdware heteRogéneas.
Sparse triangular linear systems
GPU
Level-set analysis
Synchronization-free methods
Inglés
Universidad de la República
COLIBRI
https://hdl.handle.net/20.500.12008/53698
Acceso abierto
Licencia Creative Commons Atribución - No Comercial - Sin Derivadas (CC - By-NC-ND 4.0)
_version_ 1875693219578118144
author Freire, Manuel
author2 Dufrechou, Ernesto
Ezzatti, Pablo
author2_role author
author
author_facet Freire, Manuel
Dufrechou, Ernesto
Ezzatti, Pablo
author_role author
bitstream.checksum.fl_str_mv 6429389a7df7277b72b7924fdc7d47a9
a006180e3f5b2ad0b88185d14284c0e0
6e56b49fba1dc4188aad5969ed84060f
d62648cf14c1e37917d392ac87012955
4bf7f5b565ae0c011a795e4983b94ff2
bitstream.checksumAlgorithm.fl_str_mv MD5
MD5
MD5
MD5
MD5
bitstream.url.fl_str_mv http://localhost:8080/xmlui/bitstream/20.500.12008/53698/5/license.txt
http://localhost:8080/xmlui/bitstream/20.500.12008/53698/2/license_url
http://localhost:8080/xmlui/bitstream/20.500.12008/53698/3/license_text
http://localhost:8080/xmlui/bitstream/20.500.12008/53698/4/license_rdf
http://localhost:8080/xmlui/bitstream/20.500.12008/53698/1/FDE24.pdf
collection COLIBRI
dc.contributor.filiacion.none.fl_str_mv Freire Manuel, Universidad de la República (Uruguay). Facultad de Ingeniería.
Dufrechou Ernesto, Universidad de la República (Uruguay). Facultad de Ingeniería.
Ezzatti Pablo, Universidad de la República (Uruguay). Facultad de Ingeniería.
dc.creator.none.fl_str_mv Freire, Manuel
Dufrechou, Ernesto
Ezzatti, Pablo
dc.date.accessioned.none.fl_str_mv 2026-03-04T15:43:37Z
dc.date.available.none.fl_str_mv 2026-03-04T15:43:37Z
dc.date.issued.none.fl_str_mv 2024
dc.description.abstract.none.fl_txt_mv Due to its relevant role in many numerical methods, the solution of sparse triangular linear systems (SpTRSV) in parallel platforms is continuously studied to extract as much performance as possible from the latest hardware architectures. In the case of GPUs, the latest solvers use the synchronization-free paradigm. When the problem involves several system solutions for the same matrix, they often pre-process it through a levelset analysis to improve the equation solution scheduling in the solution phase. In addition, other optimizations address the load balancing issues and irregular memory access of the SpTRSV. In this work, we modify the classical approach to compute the level sets used in the parallel SpTRSV computation, and we show that the new strategy generally reduces the computation time of the solver. Furthermore, we design an internal matrix representation that can significantly accelerate the solution stage at the cost of increasing the memory storage requirements of the algorithm. The experimental evaluation shows that the proposed modifications can improve the performance of a recent levelset and synchronization-free solver by up to 70%, significantly outperforming other state-of-the-art solvers, especially when several linear systems must be solved for each analysis phase.
dc.description.sponsorship.none.fl_txt_mv FCE_3_2022_1_172419 - MODELAR: Modelado del desempeñO de métoDos numÉricos en pLataformas de hArdware heteRogéneas.
dc.format.extent.es.fl_str_mv 11 p.
dc.format.mimetype.es.fl_str_mv application/pdf
dc.identifier.citation.es.fl_str_mv Freire, M., Dufrechou, E. y Ezzatti, P. A new level-set analysis and sparse storage format for the SPTRSV in GPUs [Preprint] Publicado en : 2024 IEEE 36th International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD), Hilo, HI, USA, 2024, pp. 59-69, DOI: 10.1109/SBAC-PAD63648.2024.00014.
dc.identifier.uri.none.fl_str_mv https://hdl.handle.net/20.500.12008/53698
dc.language.iso.none.fl_str_mv en
eng
dc.rights.license.none.fl_str_mv Licencia Creative Commons Atribución - No Comercial - Sin Derivadas (CC - By-NC-ND 4.0)
dc.rights.none.fl_str_mv info:eu-repo/semantics/openAccess
dc.source.none.fl_str_mv reponame:COLIBRI
instname:Universidad de la República
instacron:Universidad de la República
dc.subject.es.fl_str_mv Sparse triangular linear systems
GPU
Level-set analysis
Synchronization-free methods
dc.title.none.fl_str_mv A new level-set analysis and sparse storage format for the SPTRSV in GPUs
dc.type.es.fl_str_mv Preprint
dc.type.none.fl_str_mv info:eu-repo/semantics/preprint
dc.type.version.none.fl_str_mv info:eu-repo/semantics/submittedVersion
description Due to its relevant role in many numerical methods, the solution of sparse triangular linear systems (SpTRSV) in parallel platforms is continuously studied to extract as much performance as possible from the latest hardware architectures. In the case of GPUs, the latest solvers use the synchronization-free paradigm. When the problem involves several system solutions for the same matrix, they often pre-process it through a levelset analysis to improve the equation solution scheduling in the solution phase. In addition, other optimizations address the load balancing issues and irregular memory access of the SpTRSV. In this work, we modify the classical approach to compute the level sets used in the parallel SpTRSV computation, and we show that the new strategy generally reduces the computation time of the solver. Furthermore, we design an internal matrix representation that can significantly accelerate the solution stage at the cost of increasing the memory storage requirements of the algorithm. The experimental evaluation shows that the proposed modifications can improve the performance of a recent levelset and synchronization-free solver by up to 70%, significantly outperforming other state-of-the-art solvers, especially when several linear systems must be solved for each analysis phase.
eu_rights_str_mv openAccess
format preprint
id COLIBRI_4f3b9e1349a57c06d6d3574a67ba5567
identifier_str_mv Freire, M., Dufrechou, E. y Ezzatti, P. A new level-set analysis and sparse storage format for the SPTRSV in GPUs [Preprint] Publicado en : 2024 IEEE 36th International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD), Hilo, HI, USA, 2024, pp. 59-69, DOI: 10.1109/SBAC-PAD63648.2024.00014.
instacron_str Universidad de la República
institution Universidad de la República
instname_str Universidad de la República
language eng
language_invalid_str_mv en
network_acronym_str COLIBRI
network_name_str COLIBRI
oai_identifier_str oai:colibri.udelar.edu.uy:20.500.12008/53698
publishDate 2024
reponame_str COLIBRI
repository.mail.fl_str_mv karina.camps@seciu.edu.uy
repository.name.fl_str_mv COLIBRI - Universidad de la República
repository_id_str 4771
rights_invalid_str_mv Licencia Creative Commons Atribución - No Comercial - Sin Derivadas (CC - By-NC-ND 4.0)
spelling Freire Manuel, Universidad de la República (Uruguay). Facultad de Ingeniería.Dufrechou Ernesto, Universidad de la República (Uruguay). Facultad de Ingeniería.Ezzatti Pablo, Universidad de la República (Uruguay). Facultad de Ingeniería.2026-03-04T15:43:37Z2026-03-04T15:43:37Z2024Freire, M., Dufrechou, E. y Ezzatti, P. A new level-set analysis and sparse storage format for the SPTRSV in GPUs [Preprint] Publicado en : 2024 IEEE 36th International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD), Hilo, HI, USA, 2024, pp. 59-69, DOI: 10.1109/SBAC-PAD63648.2024.00014.https://hdl.handle.net/20.500.12008/53698Due to its relevant role in many numerical methods, the solution of sparse triangular linear systems (SpTRSV) in parallel platforms is continuously studied to extract as much performance as possible from the latest hardware architectures. In the case of GPUs, the latest solvers use the synchronization-free paradigm. When the problem involves several system solutions for the same matrix, they often pre-process it through a levelset analysis to improve the equation solution scheduling in the solution phase. In addition, other optimizations address the load balancing issues and irregular memory access of the SpTRSV. In this work, we modify the classical approach to compute the level sets used in the parallel SpTRSV computation, and we show that the new strategy generally reduces the computation time of the solver. Furthermore, we design an internal matrix representation that can significantly accelerate the solution stage at the cost of increasing the memory storage requirements of the algorithm. The experimental evaluation shows that the proposed modifications can improve the performance of a recent levelset and synchronization-free solver by up to 70%, significantly outperforming other state-of-the-art solvers, especially when several linear systems must be solved for each analysis phase.Submitted by Machado Jimena (jmachado@fing.edu.uy) on 2026-03-03T17:43:02Z No. of bitstreams: 2 license_rdf: 27293 bytes, checksum: d62648cf14c1e37917d392ac87012955 (MD5) FDE24.pdf: 1707454 bytes, checksum: 4bf7f5b565ae0c011a795e4983b94ff2 (MD5)Approved for entry into archive by Machado Jimena (jmachado@fing.edu.uy) on 2026-03-04T14:14:37Z (GMT) No. of bitstreams: 2 license_rdf: 27293 bytes, checksum: d62648cf14c1e37917d392ac87012955 (MD5) FDE24.pdf: 1707454 bytes, checksum: 4bf7f5b565ae0c011a795e4983b94ff2 (MD5)Made available in DSpace by Luna Fabiana (fabiana.luna@seciu.edu.uy) on 2026-03-04T15:43:37Z (GMT). No. of bitstreams: 2 license_rdf: 27293 bytes, checksum: d62648cf14c1e37917d392ac87012955 (MD5) FDE24.pdf: 1707454 bytes, checksum: 4bf7f5b565ae0c011a795e4983b94ff2 (MD5) Previous issue date: 2024FCE_3_2022_1_172419 - MODELAR: Modelado del desempeñO de métoDos numÉricos en pLataformas de hArdware heteRogéneas.11 p.application/pdfenengLas obras depositadas en el Repositorio se rigen por la Ordenanza de los Derechos de la Propiedad Intelectual de la Universidad de la República.(Res. Nº 91 de C.D.C. de 8/III/1994 – D.O. 7/IV/1994) y por la Ordenanza del Repositorio Abierto de la Universidad de la República (Res. Nº 16 de C.D.C. de 07/10/2014)info:eu-repo/semantics/openAccessLicencia Creative Commons Atribución - No Comercial - Sin Derivadas (CC - By-NC-ND 4.0)Sparse triangular linear systemsGPULevel-set analysisSynchronization-free methodsA new level-set analysis and sparse storage format for the SPTRSV in GPUsPreprintinfo:eu-repo/semantics/preprintinfo:eu-repo/semantics/submittedVersionreponame:COLIBRIinstname:Universidad de la Repúblicainstacron:Universidad de la RepúblicaFreire, ManuelDufrechou, ErnestoEzzatti, PabloLICENSElicense.txtlicense.txttext/plain; charset=utf-84267http://localhost:8080/xmlui/bitstream/20.500.12008/53698/5/license.txt6429389a7df7277b72b7924fdc7d47a9MD55CC-LICENSElicense_urllicense_urltext/plain; charset=utf-850http://localhost:8080/xmlui/bitstream/20.500.12008/53698/2/license_urla006180e3f5b2ad0b88185d14284c0e0MD52license_textlicense_texttext/html; charset=utf-838500http://localhost:8080/xmlui/bitstream/20.500.12008/53698/3/license_text6e56b49fba1dc4188aad5969ed84060fMD53license_rdflicense_rdfapplication/rdf+xml; charset=utf-827293http://localhost:8080/xmlui/bitstream/20.500.12008/53698/4/license_rdfd62648cf14c1e37917d392ac87012955MD54ORIGINALFDE24.pdfFDE24.pdfapplication/pdf1707454http://localhost:8080/xmlui/bitstream/20.500.12008/53698/1/FDE24.pdf4bf7f5b565ae0c011a795e4983b94ff2MD5120.500.12008/536982026-03-04 12:43:37.703oai:colibri.udelar.edu.uy:20.500.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Institucionalhttps://www.colibri.udelar.edu.uyUniversidadhttps://udelar.edu.uy/https://www.colibri.udelar.edu.uy/oai/requestkarina.camps@seciu.edu.uyUruguayopendoar:47712026-03-04T15:43:37COLIBRI - Universidad de la Repúblicafalse
spellingShingle A new level-set analysis and sparse storage format for the SPTRSV in GPUs
Freire, Manuel
Sparse triangular linear systems
GPU
Level-set analysis
Synchronization-free methods
status_str submittedVersion
title A new level-set analysis and sparse storage format for the SPTRSV in GPUs
title_full A new level-set analysis and sparse storage format for the SPTRSV in GPUs
title_fullStr A new level-set analysis and sparse storage format for the SPTRSV in GPUs
title_full_unstemmed A new level-set analysis and sparse storage format for the SPTRSV in GPUs
title_short A new level-set analysis and sparse storage format for the SPTRSV in GPUs
title_sort A new level-set analysis and sparse storage format for the SPTRSV in GPUs
topic Sparse triangular linear systems
GPU
Level-set analysis
Synchronization-free methods
url https://hdl.handle.net/20.500.12008/53698