Quantifying differences between openMP and MPI using a large-scale application suite

Brian Armstrong, Seon Wook Kim, Rudolf Eigenmann

Research output: Chapter in Book/Report/Conference proceedingConference contribution

6 Citations (Scopus)

Abstract

In this paper we provide quantitative information about the performance differences between the OpenMP and the MPI version of a large-scale application benchmark suite, SPECseis. We have gathered extensive performance data using hardware counters on a 4-processor Sun Enterprise system. For the presentation of this information we use a Speedup Component Model, which is able to precisely show the impact of various overheads on the program speedup. We have found that overall, the performance figures of both program versions match closely. However, our analysis also shows interesting differences in individual program phases and in overhead categories incurred. Our work gives initial answers to a largely unanswered research question: what are the sources of inefficiencies of OpenMP programs relative to other programming paradigms on large, realistic applications. Our results indicate that the OpenMP and MPI models are basically performance-equivalent on shared-memory architectures. However, we also found interesting differences in behavioral details, such as the number of instructions executed, and the incurred memory latencies and processor stalls.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages482-493
Number of pages12
Volume1940
ISBN (Print)9783540411284
Publication statusPublished - 2000
Externally publishedYes
Event3rd International Symposium on High Performance Computing, ISHPC 2000 - Tokyo, Japan
Duration: 2000 Oct 162000 Oct 18

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1940
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other3rd International Symposium on High Performance Computing, ISHPC 2000
CountryJapan
CityTokyo
Period00/10/1600/10/18

Fingerprint

OpenMP
Memory architecture
Information use
Sun
Speedup
Hardware
Data storage equipment
Component Model
Shared Memory
Latency
Industry
Figure
Programming
Paradigm
Benchmark
Model

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Armstrong, B., Kim, S. W., & Eigenmann, R. (2000). Quantifying differences between openMP and MPI using a large-scale application suite. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 1940, pp. 482-493). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1940). Springer Verlag.

Quantifying differences between openMP and MPI using a large-scale application suite. / Armstrong, Brian; Kim, Seon Wook; Eigenmann, Rudolf.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 1940 Springer Verlag, 2000. p. 482-493 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1940).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Armstrong, B, Kim, SW & Eigenmann, R 2000, Quantifying differences between openMP and MPI using a large-scale application suite. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 1940, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1940, Springer Verlag, pp. 482-493, 3rd International Symposium on High Performance Computing, ISHPC 2000, Tokyo, Japan, 00/10/16.
Armstrong B, Kim SW, Eigenmann R. Quantifying differences between openMP and MPI using a large-scale application suite. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 1940. Springer Verlag. 2000. p. 482-493. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
Armstrong, Brian ; Kim, Seon Wook ; Eigenmann, Rudolf. / Quantifying differences between openMP and MPI using a large-scale application suite. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 1940 Springer Verlag, 2000. pp. 482-493 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{ae0c8882ebc14e818db97cdb34b8a9c2,
title = "Quantifying differences between openMP and MPI using a large-scale application suite",
abstract = "In this paper we provide quantitative information about the performance differences between the OpenMP and the MPI version of a large-scale application benchmark suite, SPECseis. We have gathered extensive performance data using hardware counters on a 4-processor Sun Enterprise system. For the presentation of this information we use a Speedup Component Model, which is able to precisely show the impact of various overheads on the program speedup. We have found that overall, the performance figures of both program versions match closely. However, our analysis also shows interesting differences in individual program phases and in overhead categories incurred. Our work gives initial answers to a largely unanswered research question: what are the sources of inefficiencies of OpenMP programs relative to other programming paradigms on large, realistic applications. Our results indicate that the OpenMP and MPI models are basically performance-equivalent on shared-memory architectures. However, we also found interesting differences in behavioral details, such as the number of instructions executed, and the incurred memory latencies and processor stalls.",
author = "Brian Armstrong and Kim, {Seon Wook} and Rudolf Eigenmann",
year = "2000",
language = "English",
isbn = "9783540411284",
volume = "1940",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "482--493",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Quantifying differences between openMP and MPI using a large-scale application suite

AU - Armstrong, Brian

AU - Kim, Seon Wook

AU - Eigenmann, Rudolf

PY - 2000

Y1 - 2000

N2 - In this paper we provide quantitative information about the performance differences between the OpenMP and the MPI version of a large-scale application benchmark suite, SPECseis. We have gathered extensive performance data using hardware counters on a 4-processor Sun Enterprise system. For the presentation of this information we use a Speedup Component Model, which is able to precisely show the impact of various overheads on the program speedup. We have found that overall, the performance figures of both program versions match closely. However, our analysis also shows interesting differences in individual program phases and in overhead categories incurred. Our work gives initial answers to a largely unanswered research question: what are the sources of inefficiencies of OpenMP programs relative to other programming paradigms on large, realistic applications. Our results indicate that the OpenMP and MPI models are basically performance-equivalent on shared-memory architectures. However, we also found interesting differences in behavioral details, such as the number of instructions executed, and the incurred memory latencies and processor stalls.

AB - In this paper we provide quantitative information about the performance differences between the OpenMP and the MPI version of a large-scale application benchmark suite, SPECseis. We have gathered extensive performance data using hardware counters on a 4-processor Sun Enterprise system. For the presentation of this information we use a Speedup Component Model, which is able to precisely show the impact of various overheads on the program speedup. We have found that overall, the performance figures of both program versions match closely. However, our analysis also shows interesting differences in individual program phases and in overhead categories incurred. Our work gives initial answers to a largely unanswered research question: what are the sources of inefficiencies of OpenMP programs relative to other programming paradigms on large, realistic applications. Our results indicate that the OpenMP and MPI models are basically performance-equivalent on shared-memory architectures. However, we also found interesting differences in behavioral details, such as the number of instructions executed, and the incurred memory latencies and processor stalls.

UR - http://www.scopus.com/inward/record.url?scp=77957065596&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77957065596&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:77957065596

SN - 9783540411284

VL - 1940

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

SP - 482

EP - 493

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

PB - Springer Verlag

ER -