Evaluating performance portability of OpenACC

Amit Sabne; Putt Sakdhnagool; Seyong Lee; Jeffrey S. Vetter

doi:10.1007/978-3-319-17473-0_4

Evaluating performance portability of OpenACC

Amit Sabne, Putt Sakdhnagool, Seyong Lee, Jeffrey S. Vetter

Architectures and Performance

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

21 Scopus citations

Abstract

Accelerator-based heterogeneous computing is gaining momentum in High Performance Computing arena. However, the increased complexity of the accelerator architectures demands more generic, highlevel programming models. OpenACC is one such attempt to tackle the problem. While the abstraction endowed by OpenACC offers productivity, it raises questions on its portability. This paper evaluates the performance portability obtained by OpenACC on twelve OpenACC programs on NVIDIA CUDA, AMD GCN, and Intel MIC architectures. We study the effects of various compiler optimizations and OpenACC program settings on these architectures to provide insights into the achieved performance portability.

Original language	English
Title of host publication	Languages and Compilers for Parallel Computing - 27th International Workshop, LCPC 2014, Revised Selected Papers
Editors	James Brodman, Peng Tu
Publisher	Springer Verlag
Pages	51-66
Number of pages	16
ISBN (Electronic)	9783319174723
DOIs	https://doi.org/10.1007/978-3-319-17473-0_4
State	Published - 2015
Event	27th International Workshop on Languages and Compilers for Parallel Computing, LCPC 2014 - Hillsboro, United States Duration: Sep 15 2014 → Sep 17 2014

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	8967
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	27th International Workshop on Languages and Compilers for Parallel Computing, LCPC 2014
Country/Territory	United States
City	Hillsboro
Period	09/15/14 → 09/17/14

Funding

The paper has been authored by Oak Ridge National Laboratory, which is managed by UT-Battelle, LLC under Contract #DE-AC05-00OR22725 to the U.S. Government. Accordingly, the U.S. Government retains a non-exclusive, royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for U.S. Government purposes. This research is sponsored by the Office of Advanced Scientific Computing Research in the U.S. Department of Energy. This research is sponsored by the Office of Advanced Scientific Computing Research in the U.S. Department of Energy.

Keywords

High performance computing
OpenACC
Performance portability

Access to Document

10.1007/978-3-319-17473-0_4

Cite this

Sabne, A., Sakdhnagool, P., Lee, S., & Vetter, J. S. (2015). Evaluating performance portability of OpenACC. In J. Brodman, & P. Tu (Eds.), Languages and Compilers for Parallel Computing - 27th International Workshop, LCPC 2014, Revised Selected Papers (pp. 51-66). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8967). Springer Verlag. https://doi.org/10.1007/978-3-319-17473-0_4

Sabne, Amit ; Sakdhnagool, Putt ; Lee, Seyong et al. / Evaluating performance portability of OpenACC. Languages and Compilers for Parallel Computing - 27th International Workshop, LCPC 2014, Revised Selected Papers. editor / James Brodman ; Peng Tu. Springer Verlag, 2015. pp. 51-66 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{3cc2c3e629ce4ba59d6a50e53befb0b1,

title = "Evaluating performance portability of OpenACC",

abstract = "Accelerator-based heterogeneous computing is gaining momentum in High Performance Computing arena. However, the increased complexity of the accelerator architectures demands more generic, highlevel programming models. OpenACC is one such attempt to tackle the problem. While the abstraction endowed by OpenACC offers productivity, it raises questions on its portability. This paper evaluates the performance portability obtained by OpenACC on twelve OpenACC programs on NVIDIA CUDA, AMD GCN, and Intel MIC architectures. We study the effects of various compiler optimizations and OpenACC program settings on these architectures to provide insights into the achieved performance portability.",

keywords = "High performance computing, OpenACC, Performance portability",

author = "Amit Sabne and Putt Sakdhnagool and Seyong Lee and Vetter, {Jeffrey S.}",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2015.; 27th International Workshop on Languages and Compilers for Parallel Computing, LCPC 2014 ; Conference date: 15-09-2014 Through 17-09-2014",

year = "2015",

doi = "10.1007/978-3-319-17473-0_4",

language = "English",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "51--66",

editor = "James Brodman and Peng Tu",

booktitle = "Languages and Compilers for Parallel Computing - 27th International Workshop, LCPC 2014, Revised Selected Papers",

}

Sabne, A, Sakdhnagool, P, Lee, S & Vetter, JS 2015, Evaluating performance portability of OpenACC. in J Brodman & P Tu (eds), Languages and Compilers for Parallel Computing - 27th International Workshop, LCPC 2014, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8967, Springer Verlag, pp. 51-66, 27th International Workshop on Languages and Compilers for Parallel Computing, LCPC 2014, Hillsboro, United States, 09/15/14. https://doi.org/10.1007/978-3-319-17473-0_4

Evaluating performance portability of OpenACC. / Sabne, Amit; Sakdhnagool, Putt; Lee, Seyong et al.
Languages and Compilers for Parallel Computing - 27th International Workshop, LCPC 2014, Revised Selected Papers. ed. / James Brodman; Peng Tu. Springer Verlag, 2015. p. 51-66 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8967).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Evaluating performance portability of OpenACC

AU - Sabne, Amit

AU - Sakdhnagool, Putt

AU - Lee, Seyong

AU - Vetter, Jeffrey S.

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2015.

PY - 2015

Y1 - 2015

N2 - Accelerator-based heterogeneous computing is gaining momentum in High Performance Computing arena. However, the increased complexity of the accelerator architectures demands more generic, highlevel programming models. OpenACC is one such attempt to tackle the problem. While the abstraction endowed by OpenACC offers productivity, it raises questions on its portability. This paper evaluates the performance portability obtained by OpenACC on twelve OpenACC programs on NVIDIA CUDA, AMD GCN, and Intel MIC architectures. We study the effects of various compiler optimizations and OpenACC program settings on these architectures to provide insights into the achieved performance portability.

AB - Accelerator-based heterogeneous computing is gaining momentum in High Performance Computing arena. However, the increased complexity of the accelerator architectures demands more generic, highlevel programming models. OpenACC is one such attempt to tackle the problem. While the abstraction endowed by OpenACC offers productivity, it raises questions on its portability. This paper evaluates the performance portability obtained by OpenACC on twelve OpenACC programs on NVIDIA CUDA, AMD GCN, and Intel MIC architectures. We study the effects of various compiler optimizations and OpenACC program settings on these architectures to provide insights into the achieved performance portability.

KW - High performance computing

KW - OpenACC

KW - Performance portability

UR - http://www.scopus.com/inward/record.url?scp=84937459040&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-17473-0_4

DO - 10.1007/978-3-319-17473-0_4

M3 - Conference contribution

AN - SCOPUS:84937459040

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 51

EP - 66

BT - Languages and Compilers for Parallel Computing - 27th International Workshop, LCPC 2014, Revised Selected Papers

A2 - Brodman, James

A2 - Tu, Peng

PB - Springer Verlag

T2 - 27th International Workshop on Languages and Compilers for Parallel Computing, LCPC 2014

Y2 - 15 September 2014 through 17 September 2014

ER -

Sabne A, Sakdhnagool P, Lee S , Vetter JS. Evaluating performance portability of OpenACC. In Brodman J, Tu P, editors, Languages and Compilers for Parallel Computing - 27th International Workshop, LCPC 2014, Revised Selected Papers. Springer Verlag. 2015. p. 51-66. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-17473-0_4

Evaluating performance portability of OpenACC

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this