Achieving performance portability for a heat conduction solver mini-application on modern multi-core systems

Richard O. Kirk; Gihan R. Mudalige; Istvan Z. Reguly; Steven A. Wright; Matt J. Martineau; Stephen A. Jarvis

doi:10.1109/CLUSTER.2017.122

Achieving performance portability for a heat conduction solver mini-application on modern multi-core systems

Richard O. Kirk, Gihan R. Mudalige, Istvan Z. Reguly, Steven A. Wright, Matt J. Martineau, Stephen A. Jarvis

Engineering and Physical Sciences

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

11 Citations (Scopus)

Abstract

Modernizing production-grade, often legacy applications to take advantage of modern multi-core and many-core architectures can be a difficult and costly undertaking. This is especially true currently, as it is unclear which architectures will dominate future systems. The complexity of these codes can mean that parallelisation for a given architecture requires significant re-engineering. One way to assess the benefit of such an exercise would be to use mini-Applications that are representative of the legacy programs.In this paper, we investigate different implementations of TeaLeaf, a mini-Application from the Mantevo suite that solves the linear heat conduction equation. TeaLeaf has been ported to use many parallel programming models, including OpenMP, CUDA and MPI among others. It has also been re-engineered to use the OPS embedded DSL and template libraries Kokkos and RAJA. We use these different implementations to assess the performance portability of each technique on modern multi-core systems.While manually parallelising the application targeting and optimizing for each platform gives the best performance, this has the obvious disadvantage that it requires the creation of different versions for each and every platform of interest. Frameworks such as OPS, Kokkos and RAJA can produce executables of the program automatically that achieve comparable portability. Based on a recently developed performance portability metric, our results show that OPS and RAJA achieve an application performance portability score of 71% and 77% respectively for this application.

Original language	English
Title of host publication	Proceedings - 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	834-841
Number of pages	8
ISBN (Electronic)	9781538623268
DOIs	https://doi.org/10.1109/CLUSTER.2017.122
Publication status	Published - 22 Sept 2017
Event	2017 IEEE International Conference on Cluster Computing, CLUSTER 2017 - Honolulu, United States Duration: 5 Sept 2017 → 8 Sept 2017

Publication series

Name	Proceedings - IEEE International Conference on Cluster Computing, ICCC
Volume	2017-September
ISSN (Print)	1552-5244

Conference

Conference	2017 IEEE International Conference on Cluster Computing, CLUSTER 2017
Country/Territory	United States
City	Honolulu
Period	5/09/17 → 8/09/17

Bibliographical note

Funding Information:
The OPS project is funded by the UK Engineering and Physical Sciences Research Council projects EP/K038494/1, EP/K038486/1, EP/K038451/1 and EP/K038567/1 on “Future-proof massively-parallel execution of multi-block applications” project. This research was also funded by the Hungarian Human Resources Development Operational Programme (EFOP-3.6.2-16-2017-00013) and by the János Bólyai Research Scholarship of the Hungarian Academy of Sciences.

Funding Information:
This work was supported by the UK Atomic Weapons Establishment under grant CDK0724 (AWE Technical Outreach Programme). Prof. Stephen Jarvis is an AWE William Penney Fellow. This work would not have been possible without the assistance of a number of members of the Applied Computer Science group at AWE, to whom we would like to express our gratitude.

Publisher Copyright:
© 2017 IEEE.

Keywords

Kokkos
Mini-Apps
OPS
Performance Portability
RAJA

ASJC Scopus subject areas

Software
Hardware and Architecture
Signal Processing

Access to Document

10.1109/CLUSTER.2017.122

Cite this

Kirk, R. O., Mudalige, G. R., Reguly, I. Z., Wright, S. A., Martineau, M. J., & Jarvis, S. A. (2017). Achieving performance portability for a heat conduction solver mini-application on modern multi-core systems. In Proceedings - 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017 (pp. 834-841). Article 8049026 (Proceedings - IEEE International Conference on Cluster Computing, ICCC; Vol. 2017-September). Institute of Electrical and Electronics Engineers (IEEE). https://doi.org/10.1109/CLUSTER.2017.122

Kirk, Richard O. ; Mudalige, Gihan R. ; Reguly, Istvan Z. et al. / Achieving performance portability for a heat conduction solver mini-application on modern multi-core systems. Proceedings - 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017. Institute of Electrical and Electronics Engineers (IEEE), 2017. pp. 834-841 (Proceedings - IEEE International Conference on Cluster Computing, ICCC).

@inproceedings{cb4e53d21de745e5a55b829d90f9513b,

title = "Achieving performance portability for a heat conduction solver mini-application on modern multi-core systems",

abstract = "Modernizing production-grade, often legacy applications to take advantage of modern multi-core and many-core architectures can be a difficult and costly undertaking. This is especially true currently, as it is unclear which architectures will dominate future systems. The complexity of these codes can mean that parallelisation for a given architecture requires significant re-engineering. One way to assess the benefit of such an exercise would be to use mini-Applications that are representative of the legacy programs.In this paper, we investigate different implementations of TeaLeaf, a mini-Application from the Mantevo suite that solves the linear heat conduction equation. TeaLeaf has been ported to use many parallel programming models, including OpenMP, CUDA and MPI among others. It has also been re-engineered to use the OPS embedded DSL and template libraries Kokkos and RAJA. We use these different implementations to assess the performance portability of each technique on modern multi-core systems.While manually parallelising the application targeting and optimizing for each platform gives the best performance, this has the obvious disadvantage that it requires the creation of different versions for each and every platform of interest. Frameworks such as OPS, Kokkos and RAJA can produce executables of the program automatically that achieve comparable portability. Based on a recently developed performance portability metric, our results show that OPS and RAJA achieve an application performance portability score of 71% and 77% respectively for this application.",

keywords = "Kokkos, Mini-Apps, OPS, Performance Portability, RAJA",

author = "Kirk, {Richard O.} and Mudalige, {Gihan R.} and Reguly, {Istvan Z.} and Wright, {Steven A.} and Martineau, {Matt J.} and Jarvis, {Stephen A.}",

note = "Funding Information: The OPS project is funded by the UK Engineering and Physical Sciences Research Council projects EP/K038494/1, EP/K038486/1, EP/K038451/1 and EP/K038567/1 on “Future-proof massively-parallel execution of multi-block applications” project. This research was also funded by the Hungarian Human Resources Development Operational Programme (EFOP-3.6.2-16-2017-00013) and by the J{\'a}nos B{\'o}lyai Research Scholarship of the Hungarian Academy of Sciences. Funding Information: This work was supported by the UK Atomic Weapons Establishment under grant CDK0724 (AWE Technical Outreach Programme). Prof. Stephen Jarvis is an AWE William Penney Fellow. This work would not have been possible without the assistance of a number of members of the Applied Computer Science group at AWE, to whom we would like to express our gratitude. Publisher Copyright: {\textcopyright} 2017 IEEE.; 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017 ; Conference date: 05-09-2017 Through 08-09-2017",

year = "2017",

month = sep,

day = "22",

doi = "10.1109/CLUSTER.2017.122",

language = "English",

series = "Proceedings - IEEE International Conference on Cluster Computing, ICCC",

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

pages = "834--841",

booktitle = "Proceedings - 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017",

}

Kirk, RO, Mudalige, GR, Reguly, IZ, Wright, SA, Martineau, MJ & Jarvis, SA 2017, Achieving performance portability for a heat conduction solver mini-application on modern multi-core systems. in Proceedings - 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017., 8049026, Proceedings - IEEE International Conference on Cluster Computing, ICCC, vol. 2017-September, Institute of Electrical and Electronics Engineers (IEEE), pp. 834-841, 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017, Honolulu, United States, 5/09/17. https://doi.org/10.1109/CLUSTER.2017.122

Achieving performance portability for a heat conduction solver mini-application on modern multi-core systems. / Kirk, Richard O.; Mudalige, Gihan R.; Reguly, Istvan Z. et al.
Proceedings - 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017. Institute of Electrical and Electronics Engineers (IEEE), 2017. p. 834-841 8049026 (Proceedings - IEEE International Conference on Cluster Computing, ICCC; Vol. 2017-September).

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

TY - GEN

T1 - Achieving performance portability for a heat conduction solver mini-application on modern multi-core systems

AU - Kirk, Richard O.

AU - Mudalige, Gihan R.

AU - Reguly, Istvan Z.

AU - Wright, Steven A.

AU - Martineau, Matt J.

AU - Jarvis, Stephen A.

N1 - Funding Information: The OPS project is funded by the UK Engineering and Physical Sciences Research Council projects EP/K038494/1, EP/K038486/1, EP/K038451/1 and EP/K038567/1 on “Future-proof massively-parallel execution of multi-block applications” project. This research was also funded by the Hungarian Human Resources Development Operational Programme (EFOP-3.6.2-16-2017-00013) and by the János Bólyai Research Scholarship of the Hungarian Academy of Sciences. Funding Information: This work was supported by the UK Atomic Weapons Establishment under grant CDK0724 (AWE Technical Outreach Programme). Prof. Stephen Jarvis is an AWE William Penney Fellow. This work would not have been possible without the assistance of a number of members of the Applied Computer Science group at AWE, to whom we would like to express our gratitude. Publisher Copyright: © 2017 IEEE.

PY - 2017/9/22

Y1 - 2017/9/22

N2 - Modernizing production-grade, often legacy applications to take advantage of modern multi-core and many-core architectures can be a difficult and costly undertaking. This is especially true currently, as it is unclear which architectures will dominate future systems. The complexity of these codes can mean that parallelisation for a given architecture requires significant re-engineering. One way to assess the benefit of such an exercise would be to use mini-Applications that are representative of the legacy programs.In this paper, we investigate different implementations of TeaLeaf, a mini-Application from the Mantevo suite that solves the linear heat conduction equation. TeaLeaf has been ported to use many parallel programming models, including OpenMP, CUDA and MPI among others. It has also been re-engineered to use the OPS embedded DSL and template libraries Kokkos and RAJA. We use these different implementations to assess the performance portability of each technique on modern multi-core systems.While manually parallelising the application targeting and optimizing for each platform gives the best performance, this has the obvious disadvantage that it requires the creation of different versions for each and every platform of interest. Frameworks such as OPS, Kokkos and RAJA can produce executables of the program automatically that achieve comparable portability. Based on a recently developed performance portability metric, our results show that OPS and RAJA achieve an application performance portability score of 71% and 77% respectively for this application.

AB - Modernizing production-grade, often legacy applications to take advantage of modern multi-core and many-core architectures can be a difficult and costly undertaking. This is especially true currently, as it is unclear which architectures will dominate future systems. The complexity of these codes can mean that parallelisation for a given architecture requires significant re-engineering. One way to assess the benefit of such an exercise would be to use mini-Applications that are representative of the legacy programs.In this paper, we investigate different implementations of TeaLeaf, a mini-Application from the Mantevo suite that solves the linear heat conduction equation. TeaLeaf has been ported to use many parallel programming models, including OpenMP, CUDA and MPI among others. It has also been re-engineered to use the OPS embedded DSL and template libraries Kokkos and RAJA. We use these different implementations to assess the performance portability of each technique on modern multi-core systems.While manually parallelising the application targeting and optimizing for each platform gives the best performance, this has the obvious disadvantage that it requires the creation of different versions for each and every platform of interest. Frameworks such as OPS, Kokkos and RAJA can produce executables of the program automatically that achieve comparable portability. Based on a recently developed performance portability metric, our results show that OPS and RAJA achieve an application performance portability score of 71% and 77% respectively for this application.

KW - Kokkos

KW - Mini-Apps

KW - OPS

KW - Performance Portability

KW - RAJA

UR - http://www.scopus.com/inward/record.url?scp=85032643232&partnerID=8YFLogxK

U2 - 10.1109/CLUSTER.2017.122

DO - 10.1109/CLUSTER.2017.122

M3 - Conference contribution

AN - SCOPUS:85032643232

T3 - Proceedings - IEEE International Conference on Cluster Computing, ICCC

SP - 834

EP - 841

BT - Proceedings - 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017

PB - Institute of Electrical and Electronics Engineers (IEEE)

T2 - 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017

Y2 - 5 September 2017 through 8 September 2017

ER -

Kirk RO, Mudalige GR, Reguly IZ, Wright SA, Martineau MJ, Jarvis SA. Achieving performance portability for a heat conduction solver mini-application on modern multi-core systems. In Proceedings - 2017 IEEE International Conference on Cluster Computing, CLUSTER 2017. Institute of Electrical and Electronics Engineers (IEEE). 2017. p. 834-841. 8049026. (Proceedings - IEEE International Conference on Cluster Computing, ICCC). doi: 10.1109/CLUSTER.2017.122

Achieving performance portability for a heat conduction solver mini-application on modern multi-core systems

Abstract

Publication series

Conference

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this