An empirical performance evaluation of scalable scientific applications

Jeffrey S. Vetter; Andy Yoo

doi:10.1109/SC.2002.10036

An empirical performance evaluation of scalable scientific applications

Jeffrey S. Vetter, Andy Yoo

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

48 Scopus citations

Abstract

We investigate the scalability, architectural requirements, and performance characteristics of eight scalable scientific applications. Our analysis is driven by empirical measurements using statistical and tracing instrumentation for both communication and computation. Based on these measurements, we refine our analysis into precise explanations of the factors that influence performance and scalability for each application; we distill these factors into common traits and overall recommendations for both users and designers of scalable platforms. Our experiments demonstrate that some traits, such as improvements in the scaling and performance of MPI's collective operations, will benefit most applications. We also find specific characteristics of some applications that limit performance. For example, one application's intensive use of a 64-bit, floating-point divide instruction, which has high latency and is not pipelined on the POWER3, limits the performance of the application's primary computation.

Original language	English
Title of host publication	Proceedings of the IEEE/ACM SC 2002 Conference, SC 2002
Publisher	Association for Computing Machinery
ISBN (Electronic)	076951524X
DOIs	https://doi.org/10.1109/SC.2002.10036
State	Published - 2002
Externally published	Yes
Event	2002 IEEE/ACM Conference on Supercomputing, SC 2002 - Baltimore, United States Duration: Nov 16 2002 → Nov 22 2002

Publication series

Name	Proceedings of the International Conference on Supercomputing
Volume	2002-November

Conference

Conference	2002 IEEE/ACM Conference on Supercomputing, SC 2002
Country/Territory	United States
City	Baltimore
Period	11/16/02 → 11/22/02

Funding

We wish to thank the ASCI Purple Benchmark Team and the application developers for making these benchmarks available, and the anonymous reviewers for their detailed comments. This work was performed under the auspices of the U.S. Department of Energy by the University of California, Lawrence Livermore National Laboratory under contract No. W-7405-Eng-48. This paper is available as LLNL Technical Report UCRL-JC-148061.

Access to Document

10.1109/SC.2002.10036

Cite this

@inproceedings{5efcc16970164b37863772cb9edfaaee,

title = "An empirical performance evaluation of scalable scientific applications",

abstract = "We investigate the scalability, architectural requirements, and performance characteristics of eight scalable scientific applications. Our analysis is driven by empirical measurements using statistical and tracing instrumentation for both communication and computation. Based on these measurements, we refine our analysis into precise explanations of the factors that influence performance and scalability for each application; we distill these factors into common traits and overall recommendations for both users and designers of scalable platforms. Our experiments demonstrate that some traits, such as improvements in the scaling and performance of MPI's collective operations, will benefit most applications. We also find specific characteristics of some applications that limit performance. For example, one application's intensive use of a 64-bit, floating-point divide instruction, which has high latency and is not pipelined on the POWER3, limits the performance of the application's primary computation.",

author = "Vetter, {Jeffrey S.} and Andy Yoo",

note = "Publisher Copyright: {\textcopyright} 2002 IEEE.; 2002 IEEE/ACM Conference on Supercomputing, SC 2002 ; Conference date: 16-11-2002 Through 22-11-2002",

year = "2002",

doi = "10.1109/SC.2002.10036",

language = "English",

series = "Proceedings of the International Conference on Supercomputing",

publisher = "Association for Computing Machinery",

booktitle = "Proceedings of the IEEE/ACM SC 2002 Conference, SC 2002",

}

Vetter, JS & Yoo, A 2002, An empirical performance evaluation of scalable scientific applications. in Proceedings of the IEEE/ACM SC 2002 Conference, SC 2002. Proceedings of the International Conference on Supercomputing, vol. 2002-November, Association for Computing Machinery, 2002 IEEE/ACM Conference on Supercomputing, SC 2002, Baltimore, United States, 11/16/02. https://doi.org/10.1109/SC.2002.10036

An empirical performance evaluation of scalable scientific applications. / Vetter, Jeffrey S.; Yoo, Andy.
Proceedings of the IEEE/ACM SC 2002 Conference, SC 2002. Association for Computing Machinery, 2002. (Proceedings of the International Conference on Supercomputing; Vol. 2002-November).

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

TY - GEN

T1 - An empirical performance evaluation of scalable scientific applications

AU - Vetter, Jeffrey S.

AU - Yoo, Andy

PY - 2002

Y1 - 2002

N2 - We investigate the scalability, architectural requirements, and performance characteristics of eight scalable scientific applications. Our analysis is driven by empirical measurements using statistical and tracing instrumentation for both communication and computation. Based on these measurements, we refine our analysis into precise explanations of the factors that influence performance and scalability for each application; we distill these factors into common traits and overall recommendations for both users and designers of scalable platforms. Our experiments demonstrate that some traits, such as improvements in the scaling and performance of MPI's collective operations, will benefit most applications. We also find specific characteristics of some applications that limit performance. For example, one application's intensive use of a 64-bit, floating-point divide instruction, which has high latency and is not pipelined on the POWER3, limits the performance of the application's primary computation.

AB - We investigate the scalability, architectural requirements, and performance characteristics of eight scalable scientific applications. Our analysis is driven by empirical measurements using statistical and tracing instrumentation for both communication and computation. Based on these measurements, we refine our analysis into precise explanations of the factors that influence performance and scalability for each application; we distill these factors into common traits and overall recommendations for both users and designers of scalable platforms. Our experiments demonstrate that some traits, such as improvements in the scaling and performance of MPI's collective operations, will benefit most applications. We also find specific characteristics of some applications that limit performance. For example, one application's intensive use of a 64-bit, floating-point divide instruction, which has high latency and is not pipelined on the POWER3, limits the performance of the application's primary computation.

UR - http://www.scopus.com/inward/record.url?scp=85117198273&partnerID=8YFLogxK

U2 - 10.1109/SC.2002.10036

DO - 10.1109/SC.2002.10036

M3 - Conference contribution

AN - SCOPUS:85117198273

T3 - Proceedings of the International Conference on Supercomputing

BT - Proceedings of the IEEE/ACM SC 2002 Conference, SC 2002

PB - Association for Computing Machinery

T2 - 2002 IEEE/ACM Conference on Supercomputing, SC 2002

Y2 - 16 November 2002 through 22 November 2002

ER -

An empirical performance evaluation of scalable scientific applications

Abstract

Publication series

Conference

Funding

Access to Document

Other files and links

Fingerprint

Cite this