Achieving portability and performance through OpenACC

J. A. Herdman; W. P. Gaudin; O. Perks; D. A. Beckingsale; A. C. Mallinson; S. A. Jarvis

doi:10.1109/WACCPD.2014.10

Achieving portability and performance through OpenACC

J. A. Herdman, W. P. Gaudin, O. Perks, D. A. Beckingsale, A. C. Mallinson, S. A. Jarvis

Engineering and Physical Sciences

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

11 Citations (Scopus)

Abstract

OpenACC is a directive-based programming model designed to allow easy access to emerging advanced architecture systems for existing production codes based on Fortran, C and C++. It also provides an approach to coding contemporary technologies without the need to learn complex vendor-specific languages, or understand the hardware at the deepest level. Portability and performance are the key features of this programming model, which are essential to productivity in real scientific applications.OpenACC support is provided by a number of vendors and is defined by an open standard. However the standard is relatively new, and the implementations are relatively immature. This paper experimentally evaluates the currently available compilers by assessing two approaches to the OpenACC programming model: the 'parallel' and 'kernels' constructs. The implementation of both of these construct is compared, for each vendor, showing performance differences of up to 84%. Additionally, we observe performance differences of up to 13% between the best vendor implementations. OpenACC features which appear to cause performance issues in certain compilers are identified and linked to differing default vector length clauses between vendors. These studies are carried out over a range of hardware including GPU, APU, Xeon and Xeon Phi based architectures. Finally, OpenACC performance, and productivity, are compared against the alternative native programming approaches on each targeted platform, including CUDA, OpenCL, OpenMP 4.0 and Intel Offload, in addition to MPI and OpenMP.

Original language	English
Title of host publication	Proceedings of WACCPD 2014
Subtitle of host publication	1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	19-26
Number of pages	8
ISBN (Electronic)	9781479970230
DOIs	https://doi.org/10.1109/WACCPD.2014.10
Publication status	Published - 7 Apr 2015
Event	1st Workshop on Accelerator Programming Using Directives, WACCPD 2014 - Held in Conjunction with the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2014 - New Orleans, United States Duration: 17 Nov 2014 → …

Publication series

Name	Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis

Conference

Conference	1st Workshop on Accelerator Programming Using Directives, WACCPD 2014 - Held in Conjunction with the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2014
Country/Territory	United States
City	New Orleans
Period	17/11/14 → …

Bibliographical note

Publisher Copyright:
© 2014 IEEE.

ASJC Scopus subject areas

Software

Access to Document

10.1109/WACCPD.2014.10

Cite this

Herdman, J. A., Gaudin, W. P., Perks, O., Beckingsale, D. A., Mallinson, A. C., & Jarvis, S. A. (2015). Achieving portability and performance through OpenACC. In Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis (pp. 19-26). Article 7081674 (Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis). Institute of Electrical and Electronics Engineers (IEEE). https://doi.org/10.1109/WACCPD.2014.10

Herdman, J. A. ; Gaudin, W. P. ; Perks, O. et al. / Achieving portability and performance through OpenACC. Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers (IEEE), 2015. pp. 19-26 (Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis).

@inproceedings{0890937625074f07b671107287634c18,

title = "Achieving portability and performance through OpenACC",

abstract = "OpenACC is a directive-based programming model designed to allow easy access to emerging advanced architecture systems for existing production codes based on Fortran, C and C++. It also provides an approach to coding contemporary technologies without the need to learn complex vendor-specific languages, or understand the hardware at the deepest level. Portability and performance are the key features of this programming model, which are essential to productivity in real scientific applications.OpenACC support is provided by a number of vendors and is defined by an open standard. However the standard is relatively new, and the implementations are relatively immature. This paper experimentally evaluates the currently available compilers by assessing two approaches to the OpenACC programming model: the 'parallel' and 'kernels' constructs. The implementation of both of these construct is compared, for each vendor, showing performance differences of up to 84%. Additionally, we observe performance differences of up to 13% between the best vendor implementations. OpenACC features which appear to cause performance issues in certain compilers are identified and linked to differing default vector length clauses between vendors. These studies are carried out over a range of hardware including GPU, APU, Xeon and Xeon Phi based architectures. Finally, OpenACC performance, and productivity, are compared against the alternative native programming approaches on each targeted platform, including CUDA, OpenCL, OpenMP 4.0 and Intel Offload, in addition to MPI and OpenMP.",

author = "Herdman, {J. A.} and Gaudin, {W. P.} and O. Perks and Beckingsale, {D. A.} and Mallinson, {A. C.} and Jarvis, {S. A.}",

note = "Publisher Copyright: {\textcopyright} 2014 IEEE.; 1st Workshop on Accelerator Programming Using Directives, WACCPD 2014 - Held in Conjunction with the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2014 ; Conference date: 17-11-2014",

year = "2015",

month = apr,

day = "7",

doi = "10.1109/WACCPD.2014.10",

language = "English",

series = "Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

pages = "19--26",

booktitle = "Proceedings of WACCPD 2014",

}

Herdman, JA, Gaudin, WP, Perks, O, Beckingsale, DA, Mallinson, AC & Jarvis, SA 2015, Achieving portability and performance through OpenACC. in Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis., 7081674, Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis, Institute of Electrical and Electronics Engineers (IEEE), pp. 19-26, 1st Workshop on Accelerator Programming Using Directives, WACCPD 2014 - Held in Conjunction with the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2014, New Orleans, United States, 17/11/14. https://doi.org/10.1109/WACCPD.2014.10

Achieving portability and performance through OpenACC. / Herdman, J. A.; Gaudin, W. P.; Perks, O. et al.
Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers (IEEE), 2015. p. 19-26 7081674 (Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis).

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

TY - GEN

T1 - Achieving portability and performance through OpenACC

AU - Herdman, J. A.

AU - Gaudin, W. P.

AU - Perks, O.

AU - Beckingsale, D. A.

AU - Mallinson, A. C.

AU - Jarvis, S. A.

PY - 2015/4/7

Y1 - 2015/4/7

N2 - OpenACC is a directive-based programming model designed to allow easy access to emerging advanced architecture systems for existing production codes based on Fortran, C and C++. It also provides an approach to coding contemporary technologies without the need to learn complex vendor-specific languages, or understand the hardware at the deepest level. Portability and performance are the key features of this programming model, which are essential to productivity in real scientific applications.OpenACC support is provided by a number of vendors and is defined by an open standard. However the standard is relatively new, and the implementations are relatively immature. This paper experimentally evaluates the currently available compilers by assessing two approaches to the OpenACC programming model: the 'parallel' and 'kernels' constructs. The implementation of both of these construct is compared, for each vendor, showing performance differences of up to 84%. Additionally, we observe performance differences of up to 13% between the best vendor implementations. OpenACC features which appear to cause performance issues in certain compilers are identified and linked to differing default vector length clauses between vendors. These studies are carried out over a range of hardware including GPU, APU, Xeon and Xeon Phi based architectures. Finally, OpenACC performance, and productivity, are compared against the alternative native programming approaches on each targeted platform, including CUDA, OpenCL, OpenMP 4.0 and Intel Offload, in addition to MPI and OpenMP.

AB - OpenACC is a directive-based programming model designed to allow easy access to emerging advanced architecture systems for existing production codes based on Fortran, C and C++. It also provides an approach to coding contemporary technologies without the need to learn complex vendor-specific languages, or understand the hardware at the deepest level. Portability and performance are the key features of this programming model, which are essential to productivity in real scientific applications.OpenACC support is provided by a number of vendors and is defined by an open standard. However the standard is relatively new, and the implementations are relatively immature. This paper experimentally evaluates the currently available compilers by assessing two approaches to the OpenACC programming model: the 'parallel' and 'kernels' constructs. The implementation of both of these construct is compared, for each vendor, showing performance differences of up to 84%. Additionally, we observe performance differences of up to 13% between the best vendor implementations. OpenACC features which appear to cause performance issues in certain compilers are identified and linked to differing default vector length clauses between vendors. These studies are carried out over a range of hardware including GPU, APU, Xeon and Xeon Phi based architectures. Finally, OpenACC performance, and productivity, are compared against the alternative native programming approaches on each targeted platform, including CUDA, OpenCL, OpenMP 4.0 and Intel Offload, in addition to MPI and OpenMP.

UR - http://www.scopus.com/inward/record.url?scp=84930447583&partnerID=8YFLogxK

U2 - 10.1109/WACCPD.2014.10

DO - 10.1109/WACCPD.2014.10

M3 - Conference contribution

AN - SCOPUS:84930447583

T3 - Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis

SP - 19

EP - 26

BT - Proceedings of WACCPD 2014

PB - Institute of Electrical and Electronics Engineers (IEEE)

T2 - 1st Workshop on Accelerator Programming Using Directives, WACCPD 2014 - Held in Conjunction with the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2014

Y2 - 17 November 2014

ER -

Herdman JA, Gaudin WP, Perks O, Beckingsale DA, Mallinson AC, Jarvis SA. Achieving portability and performance through OpenACC. In Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers (IEEE). 2015. p. 19-26. 7081674. (Proceedings of WACCPD 2014: 1st Workshop on Accelerator Programming Using Directives - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis). doi: 10.1109/WACCPD.2014.10

Achieving portability and performance through OpenACC

Abstract

Publication series

Conference

Bibliographical note

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this