Using Compiler Directives for Performance Portability in Scientific Computing: Kernels from Molecular Simulation

Ada Sedova; Andreas F. Tillack; Arnold Tharrington

doi:10.1007/978-3-030-12274-4_2

Using Compiler Directives for Performance Portability in Scientific Computing: Kernels from Molecular Simulation

Ada Sedova, Andreas F. Tillack, Arnold Tharrington

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

1 Scopus citations

Abstract

Achieving performance portability for high-performance computing (HPC) applications in scientific fields has become an increasingly important initiative due to large differences in emerging supercomputer architectures. Here we test some key kernels from molecular dynamics (MD) to determine whether the use of the OpenACC directive-based programming model when applied to these kernels can result in performance within an acceptable range for these types of programs in the HPC setting. We find that for easily parallelizable kernels, performance on the GPU remains within this range. On the CPU, OpenACC-parallelized pairwise distance kernels would not meet the performance standards required, when using AMD Opteron “Interlagos” processors, but with IBM Power 9 processors, performance remains within an acceptable range for small batch sizes. These kernels provide a test for achieving performance portability with compiler directives for problems with memory-intensive components as are often found in scientific applications.

Original language	English
Title of host publication	Accelerator Programming Using Directives - 5th International Workshop, WACCPD 2018, Proceedings
Editors	Sunita Chandrasekaran, Sandra Wienke, Guido Juckeland
Publisher	Springer Verlag
Pages	22-47
Number of pages	26
ISBN (Print)	9783030122737
DOIs	https://doi.org/10.1007/978-3-030-12274-4_2
State	Published - 2019
Event	5th International Workshop on Accelerator Programming Using Directives, WACCPD 2018 - Dallas, United States Duration: Nov 11 2018 → Nov 17 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11381 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	5th International Workshop on Accelerator Programming Using Directives, WACCPD 2018
Country/Territory	United States
City	Dallas
Period	11/11/18 → 11/17/18

Funding

This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Keywords

Compiler directives
Molecular simulation
OpenACC
Pairwise distance
Performance portability

Access to Document

10.1007/978-3-030-12274-4_2

Cite this

Sedova, A., Tillack, A. F., & Tharrington, A. (2019). Using Compiler Directives for Performance Portability in Scientific Computing: Kernels from Molecular Simulation. In S. Chandrasekaran, S. Wienke, & G. Juckeland (Eds.), Accelerator Programming Using Directives - 5th International Workshop, WACCPD 2018, Proceedings (pp. 22-47). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11381 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-12274-4_2

Sedova, Ada ; Tillack, Andreas F. ; Tharrington, Arnold. / Using Compiler Directives for Performance Portability in Scientific Computing : Kernels from Molecular Simulation. Accelerator Programming Using Directives - 5th International Workshop, WACCPD 2018, Proceedings. editor / Sunita Chandrasekaran ; Sandra Wienke ; Guido Juckeland. Springer Verlag, 2019. pp. 22-47 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{e835205a875f4733a53ab259e8b69bf0,

title = "Using Compiler Directives for Performance Portability in Scientific Computing: Kernels from Molecular Simulation",

abstract = "Achieving performance portability for high-performance computing (HPC) applications in scientific fields has become an increasingly important initiative due to large differences in emerging supercomputer architectures. Here we test some key kernels from molecular dynamics (MD) to determine whether the use of the OpenACC directive-based programming model when applied to these kernels can result in performance within an acceptable range for these types of programs in the HPC setting. We find that for easily parallelizable kernels, performance on the GPU remains within this range. On the CPU, OpenACC-parallelized pairwise distance kernels would not meet the performance standards required, when using AMD Opteron “Interlagos” processors, but with IBM Power 9 processors, performance remains within an acceptable range for small batch sizes. These kernels provide a test for achieving performance portability with compiler directives for problems with memory-intensive components as are often found in scientific applications.",

keywords = "Compiler directives, Molecular simulation, OpenACC, Pairwise distance, Performance portability",

author = "Ada Sedova and Tillack, {Andreas F.} and Arnold Tharrington",

note = "Publisher Copyright: {\textcopyright} 2019, Springer Nature Switzerland AG.; 5th International Workshop on Accelerator Programming Using Directives, WACCPD 2018 ; Conference date: 11-11-2018 Through 17-11-2018",

year = "2019",

doi = "10.1007/978-3-030-12274-4_2",

language = "English",

isbn = "9783030122737",

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

publisher = "Springer Verlag",

pages = "22--47",

editor = "Sunita Chandrasekaran and Sandra Wienke and Guido Juckeland",

booktitle = "Accelerator Programming Using Directives - 5th International Workshop, WACCPD 2018, Proceedings",

}

Sedova, A, Tillack, AF & Tharrington, A 2019, Using Compiler Directives for Performance Portability in Scientific Computing: Kernels from Molecular Simulation. in S Chandrasekaran, S Wienke & G Juckeland (eds), Accelerator Programming Using Directives - 5th International Workshop, WACCPD 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11381 LNCS, Springer Verlag, pp. 22-47, 5th International Workshop on Accelerator Programming Using Directives, WACCPD 2018, Dallas, United States, 11/11/18. https://doi.org/10.1007/978-3-030-12274-4_2

Using Compiler Directives for Performance Portability in Scientific Computing: Kernels from Molecular Simulation. / Sedova, Ada; Tillack, Andreas F.; Tharrington, Arnold.
Accelerator Programming Using Directives - 5th International Workshop, WACCPD 2018, Proceedings. ed. / Sunita Chandrasekaran; Sandra Wienke; Guido Juckeland. Springer Verlag, 2019. p. 22-47 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11381 LNCS).

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

TY - GEN

T1 - Using Compiler Directives for Performance Portability in Scientific Computing

T2 - 5th International Workshop on Accelerator Programming Using Directives, WACCPD 2018

AU - Sedova, Ada

AU - Tillack, Andreas F.

AU - Tharrington, Arnold

PY - 2019

Y1 - 2019

N2 - Achieving performance portability for high-performance computing (HPC) applications in scientific fields has become an increasingly important initiative due to large differences in emerging supercomputer architectures. Here we test some key kernels from molecular dynamics (MD) to determine whether the use of the OpenACC directive-based programming model when applied to these kernels can result in performance within an acceptable range for these types of programs in the HPC setting. We find that for easily parallelizable kernels, performance on the GPU remains within this range. On the CPU, OpenACC-parallelized pairwise distance kernels would not meet the performance standards required, when using AMD Opteron “Interlagos” processors, but with IBM Power 9 processors, performance remains within an acceptable range for small batch sizes. These kernels provide a test for achieving performance portability with compiler directives for problems with memory-intensive components as are often found in scientific applications.

AB - Achieving performance portability for high-performance computing (HPC) applications in scientific fields has become an increasingly important initiative due to large differences in emerging supercomputer architectures. Here we test some key kernels from molecular dynamics (MD) to determine whether the use of the OpenACC directive-based programming model when applied to these kernels can result in performance within an acceptable range for these types of programs in the HPC setting. We find that for easily parallelizable kernels, performance on the GPU remains within this range. On the CPU, OpenACC-parallelized pairwise distance kernels would not meet the performance standards required, when using AMD Opteron “Interlagos” processors, but with IBM Power 9 processors, performance remains within an acceptable range for small batch sizes. These kernels provide a test for achieving performance portability with compiler directives for problems with memory-intensive components as are often found in scientific applications.

KW - Compiler directives

KW - Molecular simulation

KW - OpenACC

KW - Pairwise distance

KW - Performance portability

UR - http://www.scopus.com/inward/record.url?scp=85066153629&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-12274-4_2

DO - 10.1007/978-3-030-12274-4_2

M3 - Conference contribution

AN - SCOPUS:85066153629

SN - 9783030122737

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

SP - 22

EP - 47

BT - Accelerator Programming Using Directives - 5th International Workshop, WACCPD 2018, Proceedings

A2 - Chandrasekaran, Sunita

A2 - Wienke, Sandra

A2 - Juckeland, Guido

PB - Springer Verlag

Y2 - 11 November 2018 through 17 November 2018

ER -

Sedova A, Tillack AF, Tharrington A. Using Compiler Directives for Performance Portability in Scientific Computing: Kernels from Molecular Simulation. In Chandrasekaran S, Wienke S, Juckeland G, editors, Accelerator Programming Using Directives - 5th International Workshop, WACCPD 2018, Proceedings. Springer Verlag. 2019. p. 22-47. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-12274-4_2

Using Compiler Directives for Performance Portability in Scientific Computing: Kernels from Molecular Simulation

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this