A detailed performance characterization of Columbia using aeronautics benchmarks and applications

Michael Aftosmis, Marsha Berger, Rupak Biswas, M. Jahed Djomehri, Robert Hood, Haoqiang Jin, Cetin Kiris

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

Abstract

Columbia is a 10,240-processor siipercluster consisting of 20 Altix nodes with 512 processors each, and currently ranked as one of the fastest computers in the world. In this paper, we investigate its suitability as a capability computing platform for aeronautics applications. We present the performance characteristics of Columbia obtained on up to eight computing nodes interconnected via the InfiniBand and/or NUMAlink4 communication fabrics. To perform the assessment, we used a subset of the NAS Parallel Benchmarks, including the multi-zone versions, and three computational fluid dynamics applications of interest to NASA. Our results show that the system holds promise for multinode application scaling to at least 4096 processors.

Original languageEnglish (US)
Title of host publicationCollection of Technical Papers - 44th AIAA Aerospace Sciences Meeting
Pages1084-1100
Number of pages17
Volume2
StatePublished - 2006
Event44th AIAA Aerospace Sciences Meeting 2006 - Reno, NV, United States
Duration: Jan 9 2006Jan 12 2006

Other

Other44th AIAA Aerospace Sciences Meeting 2006
CountryUnited States
CityReno, NV
Period1/9/061/12/06

Fingerprint

Columbia (Orbiter)
computational fluid dynamics
aeronautics
Aviation
central processing units
communication
set theory
NASA
Computational fluid dynamics
platforms
scaling
Communication
fabric
world

Keywords

  • Computational fluid dynamics
  • Multi-block overset grids
  • Multi-level parallelism
  • NAS parallel benchmarks
  • SGI Altix

ASJC Scopus subject areas

  • Space and Planetary Science
  • Aerospace Engineering

Cite this

Aftosmis, M., Berger, M., Biswas, R., Djomehri, M. J., Hood, R., Jin, H., & Kiris, C. (2006). A detailed performance characterization of Columbia using aeronautics benchmarks and applications. In Collection of Technical Papers - 44th AIAA Aerospace Sciences Meeting (Vol. 2, pp. 1084-1100)

A detailed performance characterization of Columbia using aeronautics benchmarks and applications. / Aftosmis, Michael; Berger, Marsha; Biswas, Rupak; Djomehri, M. Jahed; Hood, Robert; Jin, Haoqiang; Kiris, Cetin.

Collection of Technical Papers - 44th AIAA Aerospace Sciences Meeting. Vol. 2 2006. p. 1084-1100.

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

Aftosmis, M, Berger, M, Biswas, R, Djomehri, MJ, Hood, R, Jin, H & Kiris, C 2006, A detailed performance characterization of Columbia using aeronautics benchmarks and applications. in Collection of Technical Papers - 44th AIAA Aerospace Sciences Meeting. vol. 2, pp. 1084-1100, 44th AIAA Aerospace Sciences Meeting 2006, Reno, NV, United States, 1/9/06.
Aftosmis M, Berger M, Biswas R, Djomehri MJ, Hood R, Jin H et al. A detailed performance characterization of Columbia using aeronautics benchmarks and applications. In Collection of Technical Papers - 44th AIAA Aerospace Sciences Meeting. Vol. 2. 2006. p. 1084-1100
Aftosmis, Michael ; Berger, Marsha ; Biswas, Rupak ; Djomehri, M. Jahed ; Hood, Robert ; Jin, Haoqiang ; Kiris, Cetin. / A detailed performance characterization of Columbia using aeronautics benchmarks and applications. Collection of Technical Papers - 44th AIAA Aerospace Sciences Meeting. Vol. 2 2006. pp. 1084-1100
@inproceedings{fbe4625b6eed4c0a93d44bf20bb8b81a,
title = "A detailed performance characterization of Columbia using aeronautics benchmarks and applications",
abstract = "Columbia is a 10,240-processor siipercluster consisting of 20 Altix nodes with 512 processors each, and currently ranked as one of the fastest computers in the world. In this paper, we investigate its suitability as a capability computing platform for aeronautics applications. We present the performance characteristics of Columbia obtained on up to eight computing nodes interconnected via the InfiniBand and/or NUMAlink4 communication fabrics. To perform the assessment, we used a subset of the NAS Parallel Benchmarks, including the multi-zone versions, and three computational fluid dynamics applications of interest to NASA. Our results show that the system holds promise for multinode application scaling to at least 4096 processors.",
keywords = "Computational fluid dynamics, Multi-block overset grids, Multi-level parallelism, NAS parallel benchmarks, SGI Altix",
author = "Michael Aftosmis and Marsha Berger and Rupak Biswas and Djomehri, {M. Jahed} and Robert Hood and Haoqiang Jin and Cetin Kiris",
year = "2006",
language = "English (US)",
isbn = "1563478072",
volume = "2",
pages = "1084--1100",
booktitle = "Collection of Technical Papers - 44th AIAA Aerospace Sciences Meeting",

}

TY - GEN

T1 - A detailed performance characterization of Columbia using aeronautics benchmarks and applications

AU - Aftosmis, Michael

AU - Berger, Marsha

AU - Biswas, Rupak

AU - Djomehri, M. Jahed

AU - Hood, Robert

AU - Jin, Haoqiang

AU - Kiris, Cetin

PY - 2006

Y1 - 2006

N2 - Columbia is a 10,240-processor siipercluster consisting of 20 Altix nodes with 512 processors each, and currently ranked as one of the fastest computers in the world. In this paper, we investigate its suitability as a capability computing platform for aeronautics applications. We present the performance characteristics of Columbia obtained on up to eight computing nodes interconnected via the InfiniBand and/or NUMAlink4 communication fabrics. To perform the assessment, we used a subset of the NAS Parallel Benchmarks, including the multi-zone versions, and three computational fluid dynamics applications of interest to NASA. Our results show that the system holds promise for multinode application scaling to at least 4096 processors.

AB - Columbia is a 10,240-processor siipercluster consisting of 20 Altix nodes with 512 processors each, and currently ranked as one of the fastest computers in the world. In this paper, we investigate its suitability as a capability computing platform for aeronautics applications. We present the performance characteristics of Columbia obtained on up to eight computing nodes interconnected via the InfiniBand and/or NUMAlink4 communication fabrics. To perform the assessment, we used a subset of the NAS Parallel Benchmarks, including the multi-zone versions, and three computational fluid dynamics applications of interest to NASA. Our results show that the system holds promise for multinode application scaling to at least 4096 processors.

KW - Computational fluid dynamics

KW - Multi-block overset grids

KW - Multi-level parallelism

KW - NAS parallel benchmarks

KW - SGI Altix

UR - http://www.scopus.com/inward/record.url?scp=34250778875&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34250778875&partnerID=8YFLogxK

M3 - Conference contribution

SN - 1563478072

SN - 9781563478079

VL - 2

SP - 1084

EP - 1100

BT - Collection of Technical Papers - 44th AIAA Aerospace Sciences Meeting

ER -