Modelling the Performance of the Gaussian Chemistry Code on x86 Architectures

Joseph Antony; Mike J. Frisch; Alistair P. Rendell

doi:10.1007/978-3-540-79409-7_4

Modelling the Performance of the Gaussian Chemistry Code on x86 Architectures

Joseph Antony, Mike J. Frisch, Alistair P. Rendell

Research output: Chapter in Book/Report/Conference proceeding › Entry for encyclopedia/dictionary › peer-review

Abstract

Gaussian is a widely used scientific code with application areas in chemistry, biochemistry and material sciences. To operate efficiently on modern architectures Gaussian employs cache blocking in the generation and processing of the two-electron integrals that are used by many of its electronic structure methods. This study uses hardware performance counters to characterise the cache and memory behavior of the integral generation code used by Gaussian for Hartree-Fock calculations. A simple performance model is proposed that aims to predict overall performance as a function of total instruction and cache miss counts. The model is parameterised for three different x86 processors — the Intel Pentium M, the P4 and the AMD Opteron. Results suggest that the model is capable of predicting execution times to an accuracy of between 5 and 15%.

Original language	English
Title of host publication	Modeling, Simulation and Optimization of Complex Processes
Subtitle of host publication	Proceedings of the Third International Conference on High Performance Scientific Computing, March 6–10, 2006, Hanoi, Vietnam
Editors	Hans Georg Bock, Ekaterina Kostina, Hoang Xuan Phu, Rolf Rannacher
Publisher	Springer
Pages	49-58
Number of pages	10
ISBN (Electronic)	978-3-540-79409-7
ISBN (Print)	978-3-540-79408-0
DOIs	https://doi.org/10.1007/978-3-540-79409-7_4
Publication status	Published - 2008
Externally published	Yes

Keywords

Gaussian
Chemistry
Hartree-Fock
Performance model

Access to Document

10.1007/978-3-540-79409-7_4Licence: In Copyright

https://rdcu.be/dA9rLLicence: In Copyright

Cite this

Antony, J., Frisch, M. J., & Rendell, A. P. (2008). Modelling the Performance of the Gaussian Chemistry Code on x86 Architectures. In H. G. Bock, E. Kostina, H. X. Phu, & R. Rannacher (Eds.), Modeling, Simulation and Optimization of Complex Processes: Proceedings of the Third International Conference on High Performance Scientific Computing, March 6–10, 2006, Hanoi, Vietnam (pp. 49-58). Springer . https://doi.org/10.1007/978-3-540-79409-7_4

Antony, Joseph ; Frisch, Mike J. ; Rendell, Alistair P. / Modelling the Performance of the Gaussian Chemistry Code on x86 Architectures. Modeling, Simulation and Optimization of Complex Processes: Proceedings of the Third International Conference on High Performance Scientific Computing, March 6–10, 2006, Hanoi, Vietnam. editor / Hans Georg Bock ; Ekaterina Kostina ; Hoang Xuan Phu ; Rolf Rannacher. Springer , 2008. pp. 49-58

@inbook{456ba6c92b754c2caacfec6c3464e813,

title = "Modelling the Performance of the Gaussian Chemistry Code on x86 Architectures",

abstract = "Gaussian is a widely used scientific code with application areas in chemistry, biochemistry and material sciences. To operate efficiently on modern architectures Gaussian employs cache blocking in the generation and processing of the two-electron integrals that are used by many of its electronic structure methods. This study uses hardware performance counters to characterise the cache and memory behavior of the integral generation code used by Gaussian for Hartree-Fock calculations. A simple performance model is proposed that aims to predict overall performance as a function of total instruction and cache miss counts. The model is parameterised for three different x86 processors — the Intel Pentium M, the P4 and the AMD Opteron. Results suggest that the model is capable of predicting execution times to an accuracy of between 5 and 15%.",

keywords = "Gaussian, Chemistry, Hartree-Fock, Performance model",

author = "Joseph Antony and Frisch, {Mike J.} and Rendell, {Alistair P.}",

year = "2008",

doi = "10.1007/978-3-540-79409-7_4",

language = "English",

isbn = "978-3-540-79408-0",

pages = "49--58",

editor = "Bock, {Hans Georg} and Ekaterina Kostina and Phu, {Hoang Xuan} and Rolf Rannacher",

booktitle = "Modeling, Simulation and Optimization of Complex Processes",

publisher = "Springer ",

}

Antony, J, Frisch, MJ & Rendell, AP 2008, Modelling the Performance of the Gaussian Chemistry Code on x86 Architectures. in HG Bock, E Kostina, HX Phu & R Rannacher (eds), Modeling, Simulation and Optimization of Complex Processes: Proceedings of the Third International Conference on High Performance Scientific Computing, March 6–10, 2006, Hanoi, Vietnam. Springer , pp. 49-58. https://doi.org/10.1007/978-3-540-79409-7_4

Modelling the Performance of the Gaussian Chemistry Code on x86 Architectures. / Antony, Joseph; Frisch, Mike J.; Rendell, Alistair P.
Modeling, Simulation and Optimization of Complex Processes: Proceedings of the Third International Conference on High Performance Scientific Computing, March 6–10, 2006, Hanoi, Vietnam. ed. / Hans Georg Bock; Ekaterina Kostina; Hoang Xuan Phu; Rolf Rannacher. Springer , 2008. p. 49-58.

Research output: Chapter in Book/Report/Conference proceeding › Entry for encyclopedia/dictionary › peer-review

TY - CHAP

T1 - Modelling the Performance of the Gaussian Chemistry Code on x86 Architectures

AU - Antony, Joseph

AU - Frisch, Mike J.

AU - Rendell, Alistair P.

PY - 2008

Y1 - 2008

N2 - Gaussian is a widely used scientific code with application areas in chemistry, biochemistry and material sciences. To operate efficiently on modern architectures Gaussian employs cache blocking in the generation and processing of the two-electron integrals that are used by many of its electronic structure methods. This study uses hardware performance counters to characterise the cache and memory behavior of the integral generation code used by Gaussian for Hartree-Fock calculations. A simple performance model is proposed that aims to predict overall performance as a function of total instruction and cache miss counts. The model is parameterised for three different x86 processors — the Intel Pentium M, the P4 and the AMD Opteron. Results suggest that the model is capable of predicting execution times to an accuracy of between 5 and 15%.

AB - Gaussian is a widely used scientific code with application areas in chemistry, biochemistry and material sciences. To operate efficiently on modern architectures Gaussian employs cache blocking in the generation and processing of the two-electron integrals that are used by many of its electronic structure methods. This study uses hardware performance counters to characterise the cache and memory behavior of the integral generation code used by Gaussian for Hartree-Fock calculations. A simple performance model is proposed that aims to predict overall performance as a function of total instruction and cache miss counts. The model is parameterised for three different x86 processors — the Intel Pentium M, the P4 and the AMD Opteron. Results suggest that the model is capable of predicting execution times to an accuracy of between 5 and 15%.

KW - Gaussian

KW - Chemistry

KW - Hartree-Fock

KW - Performance model

UR - http://purl.org/au-research/grants/ARC/LP0347178

U2 - 10.1007/978-3-540-79409-7_4

DO - 10.1007/978-3-540-79409-7_4

M3 - Entry for encyclopedia/dictionary

SN - 978-3-540-79408-0

SP - 49

EP - 58

BT - Modeling, Simulation and Optimization of Complex Processes

A2 - Bock, Hans Georg

A2 - Kostina, Ekaterina

A2 - Phu, Hoang Xuan

A2 - Rannacher, Rolf

PB - Springer

ER -

Antony J, Frisch MJ, Rendell AP. Modelling the Performance of the Gaussian Chemistry Code on x86 Architectures. In Bock HG, Kostina E, Phu HX, Rannacher R, editors, Modeling, Simulation and Optimization of Complex Processes: Proceedings of the Third International Conference on High Performance Scientific Computing, March 6–10, 2006, Hanoi, Vietnam. Springer . 2008. p. 49-58 doi: 10.1007/978-3-540-79409-7_4

Modelling the Performance of the Gaussian Chemistry Code on x86 Architectures

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this