Characterizing the impact of prefetching on scientific application performance

Collin McCurdy; Gabriel Marin; Jeffrey S. Vetter

doi:10.1007/978-3-319-10214-6_6

Characterizing the impact of prefetching on scientific application performance

Collin McCurdy, Gabriel Marin, Jeffrey S. Vetter

Computer Science and Mathematics Div

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

2 Scopus citations

Abstract

In order to better understand the impact of hardware and software data prefetching on scientific application performance, this paper introduces two analysis techniques, one micro-architecture-centric and the other application-centric.We use these techniques to analyze representative full-scale production applications from five important Exascale target areas. We find that despite a great diversity in prefetching effectiveness across and even within applications, there is a strong correlation between regions where prefetching is most needed, due to high levels of memory traffic, and where it is most effective. We also observe that the application-centric analysis can explain many of the differences in prefetching effectiveness observed across the studied applications.

Original language	English
Title of host publication	High Performance Computing Systems
Subtitle of host publication	Performance Modeling, Benchmarking and Simulation - 4th International Workshop, PMBS 2013, Revised Selected Papers
Editors	Stephen A. Jarvis, Steven A. Wright, Simon D. Hammond
Publisher	Springer Verlag
Pages	115-135
Number of pages	21
ISBN (Electronic)	9783319102139
DOIs	https://doi.org/10.1007/978-3-319-10214-6_6
State	Published - 2014
Event	4th International Workshop on Performance Modeling, Benchmarking and Simulation of High-Performance Computing Systems, PMBS 2013 - Denver, United States Duration: Nov 18 2013 → Nov 18 2013

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	8551
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	4th International Workshop on Performance Modeling, Benchmarking and Simulation of High-Performance Computing Systems, PMBS 2013
Country/Territory	United States
City	Denver
Period	11/18/13 → 11/18/13

Bibliographical note

Publisher Copyright:
© Springer International Publishing Switzerland 2014.

Keywords

Data streaming
Performance evaluation
Prefetching

Access to Document

10.1007/978-3-319-10214-6_6

Cite this

McCurdy, C., Marin, G., & Vetter, J. S. (2014). Characterizing the impact of prefetching on scientific application performance. In S. A. Jarvis, S. A. Wright, & S. D. Hammond (Eds.), High Performance Computing Systems: Performance Modeling, Benchmarking and Simulation - 4th International Workshop, PMBS 2013, Revised Selected Papers (pp. 115-135). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8551). Springer Verlag. https://doi.org/10.1007/978-3-319-10214-6_6

McCurdy, Collin ; Marin, Gabriel ; Vetter, Jeffrey S. / Characterizing the impact of prefetching on scientific application performance. High Performance Computing Systems: Performance Modeling, Benchmarking and Simulation - 4th International Workshop, PMBS 2013, Revised Selected Papers. editor / Stephen A. Jarvis ; Steven A. Wright ; Simon D. Hammond. Springer Verlag, 2014. pp. 115-135 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{e6ef9d36733e41c4b7b049326c09ab02,

title = "Characterizing the impact of prefetching on scientific application performance",

abstract = "In order to better understand the impact of hardware and software data prefetching on scientific application performance, this paper introduces two analysis techniques, one micro-architecture-centric and the other application-centric.We use these techniques to analyze representative full-scale production applications from five important Exascale target areas. We find that despite a great diversity in prefetching effectiveness across and even within applications, there is a strong correlation between regions where prefetching is most needed, due to high levels of memory traffic, and where it is most effective. We also observe that the application-centric analysis can explain many of the differences in prefetching effectiveness observed across the studied applications.",

keywords = "Data streaming, Performance evaluation, Prefetching",

author = "Collin McCurdy and Gabriel Marin and Vetter, {Jeffrey S.}",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2014.; 4th International Workshop on Performance Modeling, Benchmarking and Simulation of High-Performance Computing Systems, PMBS 2013 ; Conference date: 18-11-2013 Through 18-11-2013",

year = "2014",

doi = "10.1007/978-3-319-10214-6_6",

language = "English",

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

publisher = "Springer Verlag",

pages = "115--135",

editor = "Jarvis, {Stephen A.} and Wright, {Steven A.} and Hammond, {Simon D.}",

booktitle = "High Performance Computing Systems",

}

McCurdy, C, Marin, G & Vetter, JS 2014, Characterizing the impact of prefetching on scientific application performance. in SA Jarvis, SA Wright & SD Hammond (eds), High Performance Computing Systems: Performance Modeling, Benchmarking and Simulation - 4th International Workshop, PMBS 2013, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8551, Springer Verlag, pp. 115-135, 4th International Workshop on Performance Modeling, Benchmarking and Simulation of High-Performance Computing Systems, PMBS 2013, Denver, United States, 11/18/13. https://doi.org/10.1007/978-3-319-10214-6_6

Characterizing the impact of prefetching on scientific application performance. / McCurdy, Collin; Marin, Gabriel; Vetter, Jeffrey S.
High Performance Computing Systems: Performance Modeling, Benchmarking and Simulation - 4th International Workshop, PMBS 2013, Revised Selected Papers. ed. / Stephen A. Jarvis; Steven A. Wright; Simon D. Hammond. Springer Verlag, 2014. p. 115-135 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8551).

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

TY - GEN

T1 - Characterizing the impact of prefetching on scientific application performance

AU - McCurdy, Collin

AU - Marin, Gabriel

AU - Vetter, Jeffrey S.

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2014.

PY - 2014

Y1 - 2014

N2 - In order to better understand the impact of hardware and software data prefetching on scientific application performance, this paper introduces two analysis techniques, one micro-architecture-centric and the other application-centric.We use these techniques to analyze representative full-scale production applications from five important Exascale target areas. We find that despite a great diversity in prefetching effectiveness across and even within applications, there is a strong correlation between regions where prefetching is most needed, due to high levels of memory traffic, and where it is most effective. We also observe that the application-centric analysis can explain many of the differences in prefetching effectiveness observed across the studied applications.

AB - In order to better understand the impact of hardware and software data prefetching on scientific application performance, this paper introduces two analysis techniques, one micro-architecture-centric and the other application-centric.We use these techniques to analyze representative full-scale production applications from five important Exascale target areas. We find that despite a great diversity in prefetching effectiveness across and even within applications, there is a strong correlation between regions where prefetching is most needed, due to high levels of memory traffic, and where it is most effective. We also observe that the application-centric analysis can explain many of the differences in prefetching effectiveness observed across the studied applications.

KW - Data streaming

KW - Performance evaluation

KW - Prefetching

UR - http://www.scopus.com/inward/record.url?scp=84908680680&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-10214-6_6

DO - 10.1007/978-3-319-10214-6_6

M3 - Conference contribution

AN - SCOPUS:84908680680

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

SP - 115

EP - 135

BT - High Performance Computing Systems

A2 - Jarvis, Stephen A.

A2 - Wright, Steven A.

A2 - Hammond, Simon D.

PB - Springer Verlag

T2 - 4th International Workshop on Performance Modeling, Benchmarking and Simulation of High-Performance Computing Systems, PMBS 2013

Y2 - 18 November 2013 through 18 November 2013

ER -

McCurdy C, Marin G, Vetter JS. Characterizing the impact of prefetching on scientific application performance. In Jarvis SA, Wright SA, Hammond SD, editors, High Performance Computing Systems: Performance Modeling, Benchmarking and Simulation - 4th International Workshop, PMBS 2013, Revised Selected Papers. Springer Verlag. 2014. p. 115-135. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-10214-6_6

Characterizing the impact of prefetching on scientific application performance

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this