CFD builder: A library builder for computational fluid dynamics

Jagan Jayaraj, Pei Hung Lin, Paul R. Woodward, Pen Chung Yew

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Scopus citations

Abstract

Computational Fluid Dynamics is an important area in scientific computing. The weak scaling of codes is well understood with about two decades of experiences using MPI. As a result, per-node performance has become very crucial to the overall machine performance. However, despite the use of multi-threading, obtaining good performance at each core is still extremely challenging. The challenges are primarily due to memory bandwidth limitations and difficulties in using short SIMD engines effectively. This work is about the techniques and a tool to improve in-core performance. Fundamental to the strategy is a hierarchical data layout made of small cubical structures of the problem states that can fit well in the cache hierarchy. The difficulties in computing the spatial derivatives (also called near neighbor computation in the literature) in a hierarchical data layout are well known, hence, such a data layout has rarely been used in finite difference codes. This work discusses how to program relatively easily for such a hierarchical data layout, the inefficiencies in this programming strategy, and how to overcome the inefficiencies. The key technique to eliminate the overheads is called pipeline-for-reuse. It is followed by a storage optimization called maximal array contraction. Both pipeline-for-reuse and maximal array contraction are highly tedious and error-prone. Therefore, we built a source-to-source translator called CFD Builder to automate the transformations using directives. The directive based approach leverages domain experts' knowledge about the code, and eliminates the need for complex analysis before program transformations. We demonstrated the effectiveness of this approach using three different applications on two different architectures and two different compilers. We see up to 6.92 performance improvement using such an approach. We believe such an approach could enable library and application writers to build efficient CFD libraries.

Original languageEnglish (US)
Title of host publicationProceedings - IEEE 28th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2014
PublisherIEEE Computer Society
Pages1029-1038
Number of pages10
ISBN (Electronic)9780769552088
DOIs
StatePublished - Nov 27 2014
Event28th IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2014 - Phoenix, United States
Duration: May 19 2014May 23 2014

Publication series

NameProceedings of the International Parallel and Distributed Processing Symposium, IPDPS
ISSN (Print)1530-2075
ISSN (Electronic)2332-1237

Other

Other28th IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2014
CountryUnited States
CityPhoenix
Period5/19/145/23/14

Keywords

  • Briquette
  • CFD
  • Hierarchical data layout
  • High performance
  • Source-to-source

Fingerprint Dive into the research topics of 'CFD builder: A library builder for computational fluid dynamics'. Together they form a unique fingerprint.

Cite this