Hybrid Parallelization of CFD Applications with Dynamic Thread Balancing

Spiegel, Alexander; Mey, Dieter an; Bischof, Christian

doi:10.1007/11558958_51

Cited by 8 publications

(6 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Tikir et al [29] showed that a migrate-on-next-touch directive can be used to create a transparent data distribution engine based on hardware access counts. Also, Spiegel and an Mey [30] showed how to use the migrate-on-next-touch call to speed up a hybrid CFD solver.…”

Section: The Numa Systemmentioning

confidence: 98%

Dynamic Data Migration for Structured AMR Solvers

Nordén

Löf

Rantakokko

et al. 2007

Int J Parallel Prog

View full text Add to dashboard Cite

On cc-NUMA multi-processors, the non-uniformity of main memory latencies motivates the need for co-location of threads and data. We call this special form of data locality, geographical locality. In this article, we study the performance of a parallel PDE solver with adaptive mesh refinement (AMR). The solver is parallelized using OpenMP and the adaptive mesh refinement makes dynamic load balancing necessary. Due to the dynamically changing memory access pattern caused by the runtime adaption, it is a challenging task to achieve a high degree of geographical locality. The main conclusions of the study are: (1) that geographical locality is very important for the performance of the solver, (2) that the performance can be improved significantly using dynamic page migration of misplaced data, (3) that a migrate-onnext-touch directive works well whereas the first-touch strategy is less advantageous for programs exhibiting a dynamically changing memory access patterns, and (4) that the overhead for such migration is low compared to the total execution time.

show abstract

Section: The Numa Systemmentioning

confidence: 98%

Dynamic Data Migration for Structured AMR Solvers

Nordén

Löf

Rantakokko

et al. 2007

Int J Parallel Prog

View full text Add to dashboard Cite

show abstract

“…Solutions that use the redistribution of computational power and are aimed at applications with two levels of parallelism are the works done by Spiegel et al [13], Duran et al [14] and LeWI [1]. They aim to balance applications with two levels of parallelism by redistributing the computational power of the inner level.…”

Section: Redistribute the Computational Powermentioning

confidence: 99%

Hints to improve automatic load balancing with LeWI for hybrid applications

García

Labarta

Corbalan

2014

Journal of Parallel and Distributed Computing

View full text Add to dashboard Cite

The DLB (Dynamic Load Balancing) library and LeWI (LEnd When Idle) algorithm provide a runtime solution to deal with the load imbalance of parallel applications independently of the source of imbalance. DLB relies on the usage of hybrid programming models and exploits the malleability of the second level of parallelism to redistribute computation power across processes.When executing real applications with LeWI, although application's performance is significantly improved, we have observed in some cases efficiency values between 60% and 70%, far from our theoretical limit.This work is a deep analysis of the sources of efficiency loss correlated with application characteristics, parallelization schemes and programming models. We have based our analysis in fine grain monitoring tools and metrics and validated our conclusions by reproducing them in synthetic experiments. As a result, this work teaches us some lessons that can be seen as hints to programmers to help LeWI make an efficient use of computational resources and obtain the maximum performance.

show abstract

“…The idea of the dynamic thread balancing scheme has previously been used to solve load imbalance of hybrid (MPI + OpenMP) programs [19]. Here the number of threads of the inner teams were adjusted to the size of the corresponding blocks.…”

Section: Thread Balancingmentioning

confidence: 99%

Nested Parallelization with OpenMP

Mey

Sarholz

Terboven

2007

Int J Parallel Prog

Self Cite

View full text Add to dashboard Cite

OpenMP is widely accepted as a de facto standard for shared memory parallel programming in Fortran, C and C++. Nested parallelization has been included in the first OpenMP specification, but it took a few years until the first commercially available compilers supported this optional part of the specification. We employed nested parallelization using OpenMP in three production codes: a C++ code for content-based image retrieval, a C++ code for the computation of critical points in multi-block CFD datasets, and a multi-block Navier-Stokes solver written in Fortran90. In this paper we discuss the opportunities as well as the deficiencies of the nested parallelization support in OpenMP.

show abstract

Hybrid Parallelization of CFD Applications with Dynamic Thread Balancing

Cited by 8 publications

References 4 publications

Dynamic Data Migration for Structured AMR Solvers

Dynamic Data Migration for Structured AMR Solvers

Hints to improve automatic load balancing with LeWI for hybrid applications

Nested Parallelization with OpenMP

Contact Info

Product

Resources

About