Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale

M. Matheny; S. Herbein; N. Podhorszki; S. Klasky; M. Taufer

doi:10.1109/PADSW.2014.7097855

Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale

M. Matheny, S. Herbein, N. Podhorszki, S. Klasky, M. Taufer

Workflow Systems

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

7 Scopus citations

Abstract

On petascale systems, the selection of optimal values for I/O parameters without taking into account the I/O size and pattern can cause the I/O time to dominate the simulation time, compromising the application's scalability. In this paper, we adopt and adapt an engineering method called surrogate-based modeling to efficiently search for the optimal I/O parameter values and accurately predict the associated I/O times at the extreme scale. Our approach allows us to address both the search and prediction in a short time, even when the application's I/O is large and exhibits irregular patterns.

Original language	English
Title of host publication	2014 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014 - Proceedings
Publisher	IEEE Computer Society
Pages	568-575
Number of pages	8
ISBN (Electronic)	9781479976157
DOIs	https://doi.org/10.1109/PADSW.2014.7097855
State	Published - 2014
Event	20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014 - Hsinchu, Taiwan, Province of China Duration: Dec 16 2014 → Dec 19 2014

Publication series

Name	Proceedings of the International Conference on Parallel and Distributed Systems - ICPADS
Volume	2015-April
ISSN (Print)	1521-9097

Conference

Conference	20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014
Country/Territory	Taiwan, Province of China
City	Hsinchu
Period	12/16/14 → 12/19/14

Funding

Funders	Funder number
National Stroke Foundation	CCF#1318445

Keywords

I/O modeling and tuning
Irregular I/O pattern
Peta- and exascale computing
QMCPack
Scientific applications

Access to Document

10.1109/PADSW.2014.7097855

Cite this

Matheny, M., Herbein, S., Podhorszki, N., Klasky, S., & Taufer, M. (2014). Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale. In 2014 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014 - Proceedings (pp. 568-575). Article 7097855 (Proceedings of the International Conference on Parallel and Distributed Systems - ICPADS; Vol. 2015-April). IEEE Computer Society. https://doi.org/10.1109/PADSW.2014.7097855

Matheny, M. ; Herbein, S. ; Podhorszki, N. et al. / Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale. 2014 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014 - Proceedings. IEEE Computer Society, 2014. pp. 568-575 (Proceedings of the International Conference on Parallel and Distributed Systems - ICPADS).

@inproceedings{794d6d9d8b7b44708d36b0ce68a56f8c,

title = "Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale",

abstract = "On petascale systems, the selection of optimal values for I/O parameters without taking into account the I/O size and pattern can cause the I/O time to dominate the simulation time, compromising the application's scalability. In this paper, we adopt and adapt an engineering method called surrogate-based modeling to efficiently search for the optimal I/O parameter values and accurately predict the associated I/O times at the extreme scale. Our approach allows us to address both the search and prediction in a short time, even when the application's I/O is large and exhibits irregular patterns.",

keywords = "I/O modeling and tuning, Irregular I/O pattern, Peta- and exascale computing, QMCPack, Scientific applications",

author = "M. Matheny and S. Herbein and N. Podhorszki and S. Klasky and M. Taufer",

note = "Publisher Copyright: {\textcopyright} 2014 IEEE.; 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014 ; Conference date: 16-12-2014 Through 19-12-2014",

year = "2014",

doi = "10.1109/PADSW.2014.7097855",

language = "English",

series = "Proceedings of the International Conference on Parallel and Distributed Systems - ICPADS",

publisher = "IEEE Computer Society",

pages = "568--575",

booktitle = "2014 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014 - Proceedings",

}

Matheny, M, Herbein, S, Podhorszki, N , Klasky, S & Taufer, M 2014, Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale. in 2014 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014 - Proceedings., 7097855, Proceedings of the International Conference on Parallel and Distributed Systems - ICPADS, vol. 2015-April, IEEE Computer Society, pp. 568-575, 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014, Hsinchu, Taiwan, Province of China, 12/16/14. https://doi.org/10.1109/PADSW.2014.7097855

Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale. / Matheny, M.; Herbein, S.; Podhorszki, N. et al.
2014 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014 - Proceedings. IEEE Computer Society, 2014. p. 568-575 7097855 (Proceedings of the International Conference on Parallel and Distributed Systems - ICPADS; Vol. 2015-April).

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

TY - GEN

T1 - Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale

AU - Matheny, M.

AU - Herbein, S.

AU - Podhorszki, N.

AU - Klasky, S.

AU - Taufer, M.

PY - 2014

Y1 - 2014

N2 - On petascale systems, the selection of optimal values for I/O parameters without taking into account the I/O size and pattern can cause the I/O time to dominate the simulation time, compromising the application's scalability. In this paper, we adopt and adapt an engineering method called surrogate-based modeling to efficiently search for the optimal I/O parameter values and accurately predict the associated I/O times at the extreme scale. Our approach allows us to address both the search and prediction in a short time, even when the application's I/O is large and exhibits irregular patterns.

AB - On petascale systems, the selection of optimal values for I/O parameters without taking into account the I/O size and pattern can cause the I/O time to dominate the simulation time, compromising the application's scalability. In this paper, we adopt and adapt an engineering method called surrogate-based modeling to efficiently search for the optimal I/O parameter values and accurately predict the associated I/O times at the extreme scale. Our approach allows us to address both the search and prediction in a short time, even when the application's I/O is large and exhibits irregular patterns.

KW - I/O modeling and tuning

KW - Irregular I/O pattern

KW - Peta- and exascale computing

KW - QMCPack

KW - Scientific applications

UR - http://www.scopus.com/inward/record.url?scp=84988305193&partnerID=8YFLogxK

U2 - 10.1109/PADSW.2014.7097855

DO - 10.1109/PADSW.2014.7097855

M3 - Conference contribution

AN - SCOPUS:84988305193

T3 - Proceedings of the International Conference on Parallel and Distributed Systems - ICPADS

SP - 568

EP - 575

BT - 2014 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014 - Proceedings

PB - IEEE Computer Society

T2 - 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014

Y2 - 16 December 2014 through 19 December 2014

ER -

Matheny M, Herbein S, Podhorszki N , Klasky S, Taufer M. Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale. In 2014 20th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2014 - Proceedings. IEEE Computer Society. 2014. p. 568-575. 7097855. (Proceedings of the International Conference on Parallel and Distributed Systems - ICPADS). doi: 10.1109/PADSW.2014.7097855

Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this