Investigating the performance of two programming models for clusters of SMP PCs

Cappello, Franck; Richard, O.; Etiemble, Daniel

doi:10.1109/hpca.2000.824364

Cited by 14 publications

(11 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…We have taken care to limit the interpretation to the clearly apparent phenomena. A previous study has been done and published [3] for a cluster of 64 2-way Pentium-II nodes. On this platform, we had access to the hardware performance counters and we could provide a deeper analysis.…”

Section: Methodsmentioning

confidence: 99%

MPI versus MPI+OpenMP on the IBM SP for the NAS Benchmarks

Cappello

Etiemble

2000

ACM/IEEE SC 2000 Conference (SC'00)

Self Cite

158

101

View full text Add to dashboard Cite

show abstract

Section: Methodsmentioning

confidence: 99%

MPI versus MPI+OpenMP on the IBM SP for the NAS Benchmarks

Cappello

Etiemble

2000

ACM/IEEE SC 2000 Conference (SC'00)

Self Cite

158

101

View full text Add to dashboard Cite

show abstract

“…The performance of the routines depends on the nature of the problem, the programming model chosen by the user, the parallel code itself, Listing 1. 6 The prototype of the GSL Dynamic Programming routine for the SRAP and the adequacy between programming model and parallel platform. The PELICAN library has been extended with a web service interface that provides a user friendly access to the routines via a web browser.…”

Section: Discussionmentioning

confidence: 99%

“…6 the Node speed-up [6]. Given a fixed number of nodes, this parameter is the ratio between the time spent using two processors on each node and using only one Fig.…”

Section: Hybrid Architecturesmentioning

confidence: 99%

Toward the parallelization of GSL

et al. 2008

View full text Add to dashboard Cite

Registro de acceso restringido Este recurso no está disponible en acceso abierto por política de la editorial. No obstante, se puede acceder al texto completo desde la Universitat Jaume I o si el usuario cuenta con suscripción. Registre d'accés restringit Aquest recurs no està disponible en accés obert per política de l'editorial. No obstant això, es pot accedir al text complet des de la Universitat Jaume I o si l'usuari compta amb subscripció. Restricted access item This item isn't open access because of publisher's policy. The full--text version is only available from Jaume I University or if the user has a running suscription to the publisher's contents.

show abstract

“…Cappello, Olivier, and Etiemble [17][18][19] were among the first to present the hybrid programming model of using MPI in conjunction with a threading model such as OpenMP. They demonstrated that it is sometimes possible to increase efficiency on some code by using a mixture of shared memory and message passing models.…”

Section: Introductionmentioning

confidence: 99%

Scalability of Incompressible Flow Computations on Multi-GPU Clusters Using Dual-Level and Tri-Level Parallelism

Jacobsen

Şenocak

2011

49th AIAA Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace Exposition

View full text Add to dashboard Cite

High performance computing using graphics processing units (GPUs) is gaining popularity in the scientific computing field, with many large compute clusters being augmented with multiple GPUs in each node. We investigate hybrid tri-level (MPI-OpenMP-CUDA) parallel implementations to explore the efficiency and scalability of incompressible flow computations on GPU clusters up to 128 GPUS. This work details some of the unique issues faced when merging fine-grain parallelism on the GPU using CUDA with coarse-grain parallelism using OpenMP for intra-node and MPI for inter-node communication. Comparisons between the tri-level MPI-OpenMP-CUDA and dual-level MPI-CUDA implementations are shown using computationally large computational fluid dynamics (CFD) simulations. Our results demonstrate that a tri-level parallel implementation does not provide a significant advantage in performance over the dual-level implementation, however further research is needed to justify our conclusion for a cluster with a high GPU per node density or when using software that can utilize OpenMP's fine-grain parallelism more effectively.

show abstract

Investigating the performance of two programming models for clusters of SMP PCs

Cited by 14 publications

References 15 publications

MPI versus MPI+OpenMP on the IBM SP for the NAS Benchmarks

MPI versus MPI+OpenMP on the IBM SP for the NAS Benchmarks

Toward the parallelization of GSL

Scalability of Incompressible Flow Computations on Multi-GPU Clusters Using Dual-Level and Tri-Level Parallelism

Contact Info

Product

Resources

About