Investigating the performance of collective communications on SMP clusters: A case for MPI-Allgather

Feng Long Gu, Nzigou M. Hyacinthe, Guilherme De Melo Baptista Domingues, Takeshi Nanri, Kazuaki Murakami

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

Abstract

Message-passing interface (MPI) is a specification targeting parallel processing on Multiple instruction stream, multiple data stream (MIMD) architectures. Collective operation in MPI specification needs intensive communications. This paper reports an investigation on the performance of collective operations on PC clusters and multi-core Symmetric MultiProcessor (SMP) computers. MPI-Allgather is one of the most relevant MPI collective communications. From the comparison between the different MPI Allgather algorithms (Ring and Pair-wise) on different platforms, we describe how multi-core systems can take the best advantage by exploiting difference between inter-node and intra-node communications.

Original languageEnglish
Title of host publicationComputation in Modern Science and Engineering - Proceedings of the International Conference on Computational Methods in Science and Engineering 2007 (ICCMSE 2007)
Pages52-56
Number of pages5
Edition2
DOIs
Publication statusPublished - Dec 1 2007
EventInternational Conference on Computational Methods in Science and Engineering 2007, ICCMSE 2007 - Corfu, Greece
Duration: Sep 25 2007Sep 30 2007

Publication series

NameAIP Conference Proceedings
Number2
Volume963
ISSN (Print)0094-243X
ISSN (Electronic)1551-7616

Other

OtherInternational Conference on Computational Methods in Science and Engineering 2007, ICCMSE 2007
CountryGreece
CityCorfu
Period9/25/079/30/07

Fingerprint

messages
communication
specifications
MIMD (computers)
platforms
rings

All Science Journal Classification (ASJC) codes

  • Physics and Astronomy(all)

Cite this

Gu, F. L., Hyacinthe, N. M., Domingues, G. D. M. B., Nanri, T., & Murakami, K. (2007). Investigating the performance of collective communications on SMP clusters: A case for MPI-Allgather. In Computation in Modern Science and Engineering - Proceedings of the International Conference on Computational Methods in Science and Engineering 2007 (ICCMSE 2007) (2 ed., pp. 52-56). (AIP Conference Proceedings; Vol. 963, No. 2). https://doi.org/10.1063/1.2836131

Investigating the performance of collective communications on SMP clusters : A case for MPI-Allgather. / Gu, Feng Long; Hyacinthe, Nzigou M.; Domingues, Guilherme De Melo Baptista; Nanri, Takeshi; Murakami, Kazuaki.

Computation in Modern Science and Engineering - Proceedings of the International Conference on Computational Methods in Science and Engineering 2007 (ICCMSE 2007). 2. ed. 2007. p. 52-56 (AIP Conference Proceedings; Vol. 963, No. 2).

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

Gu, FL, Hyacinthe, NM, Domingues, GDMB, Nanri, T & Murakami, K 2007, Investigating the performance of collective communications on SMP clusters: A case for MPI-Allgather. in Computation in Modern Science and Engineering - Proceedings of the International Conference on Computational Methods in Science and Engineering 2007 (ICCMSE 2007). 2 edn, AIP Conference Proceedings, no. 2, vol. 963, pp. 52-56, International Conference on Computational Methods in Science and Engineering 2007, ICCMSE 2007, Corfu, Greece, 9/25/07. https://doi.org/10.1063/1.2836131
Gu FL, Hyacinthe NM, Domingues GDMB, Nanri T, Murakami K. Investigating the performance of collective communications on SMP clusters: A case for MPI-Allgather. In Computation in Modern Science and Engineering - Proceedings of the International Conference on Computational Methods in Science and Engineering 2007 (ICCMSE 2007). 2 ed. 2007. p. 52-56. (AIP Conference Proceedings; 2). https://doi.org/10.1063/1.2836131
Gu, Feng Long ; Hyacinthe, Nzigou M. ; Domingues, Guilherme De Melo Baptista ; Nanri, Takeshi ; Murakami, Kazuaki. / Investigating the performance of collective communications on SMP clusters : A case for MPI-Allgather. Computation in Modern Science and Engineering - Proceedings of the International Conference on Computational Methods in Science and Engineering 2007 (ICCMSE 2007). 2. ed. 2007. pp. 52-56 (AIP Conference Proceedings; 2).
@inproceedings{2551842ba73f4df8a6394d406c6abf6d,
title = "Investigating the performance of collective communications on SMP clusters: A case for MPI-Allgather",
abstract = "Message-passing interface (MPI) is a specification targeting parallel processing on Multiple instruction stream, multiple data stream (MIMD) architectures. Collective operation in MPI specification needs intensive communications. This paper reports an investigation on the performance of collective operations on PC clusters and multi-core Symmetric MultiProcessor (SMP) computers. MPI-Allgather is one of the most relevant MPI collective communications. From the comparison between the different MPI Allgather algorithms (Ring and Pair-wise) on different platforms, we describe how multi-core systems can take the best advantage by exploiting difference between inter-node and intra-node communications.",
author = "Gu, {Feng Long} and Hyacinthe, {Nzigou M.} and Domingues, {Guilherme De Melo Baptista} and Takeshi Nanri and Kazuaki Murakami",
year = "2007",
month = "12",
day = "1",
doi = "10.1063/1.2836131",
language = "English",
isbn = "9780735404786",
series = "AIP Conference Proceedings",
number = "2",
pages = "52--56",
booktitle = "Computation in Modern Science and Engineering - Proceedings of the International Conference on Computational Methods in Science and Engineering 2007 (ICCMSE 2007)",
edition = "2",

}

TY - GEN

T1 - Investigating the performance of collective communications on SMP clusters

T2 - A case for MPI-Allgather

AU - Gu, Feng Long

AU - Hyacinthe, Nzigou M.

AU - Domingues, Guilherme De Melo Baptista

AU - Nanri, Takeshi

AU - Murakami, Kazuaki

PY - 2007/12/1

Y1 - 2007/12/1

N2 - Message-passing interface (MPI) is a specification targeting parallel processing on Multiple instruction stream, multiple data stream (MIMD) architectures. Collective operation in MPI specification needs intensive communications. This paper reports an investigation on the performance of collective operations on PC clusters and multi-core Symmetric MultiProcessor (SMP) computers. MPI-Allgather is one of the most relevant MPI collective communications. From the comparison between the different MPI Allgather algorithms (Ring and Pair-wise) on different platforms, we describe how multi-core systems can take the best advantage by exploiting difference between inter-node and intra-node communications.

AB - Message-passing interface (MPI) is a specification targeting parallel processing on Multiple instruction stream, multiple data stream (MIMD) architectures. Collective operation in MPI specification needs intensive communications. This paper reports an investigation on the performance of collective operations on PC clusters and multi-core Symmetric MultiProcessor (SMP) computers. MPI-Allgather is one of the most relevant MPI collective communications. From the comparison between the different MPI Allgather algorithms (Ring and Pair-wise) on different platforms, we describe how multi-core systems can take the best advantage by exploiting difference between inter-node and intra-node communications.

UR - http://www.scopus.com/inward/record.url?scp=71449113509&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=71449113509&partnerID=8YFLogxK

U2 - 10.1063/1.2836131

DO - 10.1063/1.2836131

M3 - Conference contribution

AN - SCOPUS:71449113509

SN - 9780735404786

T3 - AIP Conference Proceedings

SP - 52

EP - 56

BT - Computation in Modern Science and Engineering - Proceedings of the International Conference on Computational Methods in Science and Engineering 2007 (ICCMSE 2007)

ER -