Optimisation of Patch Distribution Strategies for AMR Applications

Beckingsale, David; Perks, Oliver; Gaudin, Wayne; Herdman, J. A.; Jarvis, Stephen A.

doi:10.1007/978-3-642-36781-6_15

Cited by 3 publications

(2 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another body of research which is similar to this work, develops performance models of MPI-based wavefront [1] and Adaptive Mesh Refinement (AMR) codes [18]. Despite the similarities between these models and our own, significant work would be required to prepare them for use with geometric multigrid applications.…”

Section: Related Workmentioning

confidence: 95%

“…Despite the similarities between these models and our own, significant work would be required to prepare them for use with geometric multigrid applications. Furthermore, we use performance models to assess the suitability of different partitioning algorithms/libraries at varying scale, rather than to examine different hardware and software configurations [1] or to optimising AMR patch distribution [18]. Giles et al have published several papers on the design and performance of OPlus and its successor OP2 [19]- [21].…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Predictive Evaluation of Partitioning Algorithms through Runtime Modelling

Bunt

Wright

Jarvis

et al. 2016

2016 IEEE 23rd International Conference on High Performance Computing (HiPC)

Self Cite

View full text Add to dashboard Cite

Abstract-Performance modelling unstructured mesh codes is a challenging process, due to the difficulty of capturing their memory access patterns, and their communication patterns at varying scale. In this paper we first develop extensions to an existing runtime performance model, aimed at overcoming the former, which we validate on up to 1,024 cores of a Haswellbased cluster, using both a geometric partitioning algorithm and ParMETIS to partition the input deck, with a maximum absolute runtime error of 12.63% and 11.55% respectively. To overcome the latter, we develop an application representative of the mesh partitioning process internal to an unstructured mesh code. This application is able to generate partitioning data that is usable with the performance model to produce predicted application runtimes within 7.31% of those produced using empirically collected data. We then demonstrate the use of the performance model by undertaking a predictive comparison among several partitioning algorithms on up to 30,000 cores. Additionally, we correctly predict the ineffectiveness of the geometric partitioning algorithm at 512 and 1024 cores.

show abstract

Section: Related Workmentioning

confidence: 95%