An evaluation of MPI and OpenMP paradigms in finite-difference explicit methods for PDEs on shared-memory multi- and manycore systems

Registro completo de metadados
MetadadosDescriçãoIdioma
Autor(es): dc.creatorCabral, Frederico L.-
Autor(es): dc.creatorOliveira, Sanderson L. Gonzaga de-
Autor(es): dc.creatorOsthoff, Carla-
Autor(es): dc.creatorCosta, Gabriel P.-
Autor(es): dc.creatorBrandão, Diego N.-
Autor(es): dc.creatorKischinhevsky, Mauricio-
Data de aceite: dc.date.accessioned2026-02-09T11:53:03Z-
Data de disponibilização: dc.date.available2026-02-09T11:53:03Z-
Data de envio: dc.date.issued2021-09-13-
Data de envio: dc.date.issued2021-09-13-
Data de envio: dc.date.issued2020-10-24-
Fonte completa do material: dc.identifierhttps://repositorio.ufla.br/handle/1/48110-
Fonte completa do material: dc.identifierhttps://doi.org/10.1002/cpe.5642-
Fonte: dc.identifier.urihttp://educapes.capes.gov.br/handle/capes/1149567-
Descrição: dc.descriptionThis paper focuses on parallel implementations of three two-dimensional explicit numerical methods on Intel® Xeon® Scalable Processor and the coprocessor Knights Landing. In this study, the performance of a hybrid parallel programming with message passing interface (MPI) and Open Multi-Processing (OpenMP) and a pure MPI implementation used with two thread binding policies is compared with an improved OpenMP-based implementation in three explicit finite-difference methods for solving partial differential equations on shared-memory multicore and manycore systems. Specifically, the improved OpenMP-based version is a strategy that synchronizes adjacent threads and eliminates the implicit barriers of a naïve OpenMP-based implementation. The experiments show that the most suitable approach depends on several characteristics related to the nonuniform memory access (NUMA) effect and load balancing, such as the size of the MPI domain and the number of synchronization points used in the parallel implementation. In algorithms that use four and five synchronization points, hybrid MPI/OpenMP approaches yielded better speedups than the other versions did in runs performed on both systems. The pure MPI-based strategy, however, achieved better results than the other proposed approaches did in the method that employs only one synchronization point.-
Idioma: dc.languageen-
Publicador: dc.publisherWiley-
Direitos: dc.rightsrestrictAccess-
???dc.source???: dc.sourceConcurrency and Computation: Practice and Experience-
Palavras-chave: dc.subjectHigh-performance computing-
Palavras-chave: dc.subjectMulticore architectures-
Palavras-chave: dc.subjectParallelism-
Palavras-chave: dc.subjectParallel processing-
Palavras-chave: dc.subjectComputação de alto desempenho-
Palavras-chave: dc.subjectArquiteturas multicore-
Palavras-chave: dc.subjectParalelismo-
Palavras-chave: dc.subjectProcessamento paralelo-
Título: dc.titleAn evaluation of MPI and OpenMP paradigms in finite-difference explicit methods for PDEs on shared-memory multi- and manycore systems-
Tipo de arquivo: dc.typeArtigo-
Aparece nas coleções:Repositório Institucional da Universidade Federal de Lavras (RIUFLA)

Não existem arquivos associados a este item.