Scaling of a Fast Fourier Transform and a pseudo-spectral fluid solver up to 196608 cores

Anando G. Chatterjee; Mahendra K. Verma; Abhishek Kumar; Ravi Samtaney; Bilel Hadri; Rooh Khurram

doi:10.1016/j.jpdc.2017.10.014

Scaling of a Fast Fourier Transform and a pseudo-spectral fluid solver up to 196608 cores

Anando G. Chatterjee, Mahendra K. Verma, Abhishek Kumar^*, Ravi Samtaney, Bilel Hadri, Rooh Khurram

^*Corresponding author for this work

Physical Sciences and Engineering

Research output: Contribution to journal › Article › peer-review

39 Scopus citations

Abstract

In this paper we present scaling results of a FFT library, FFTK, and a pseudospectral code, Tarang, on grid resolutions up to 8192³ grid using 65536 cores of Blue Gene/P and 196608 cores of Cray XC40 supercomputers. We observe that communication dominates computation, more so on the Cray XC40. The computation time scales as T_comp∼p⁻¹, and the communication time as T_comm∼n^−γ₂ with γ₂ ranging from 0.7 to 0.9 for Blue Gene/P, and from 0.43 to 0.73 for Cray XC40. FFTK, and the fluid and convection solvers of Tarang exhibit weak as well as strong scaling nearly up to 196608 cores of Cray XC40. We perform a comparative study of the performance on the Blue Gene/P and Cray XC40 clusters.

Original language	English (US)
Pages (from-to)	77-91
Number of pages	15
Journal	Journal of Parallel and Distributed Computing
Volume	113
DOIs	https://doi.org/10.1016/j.jpdc.2017.10.014
State	Published - Mar 2018

Bibliographical note

Funding Information:
We thank Shashwat Bhattacharya for his valuable help on plotting. Our numerical simulations were performed at Cray XC40 Shaheen II at KAUST supercomputing laboratory (KSL), Saudi Arabia. We thank KAUST scientists for the kind support while performing our simulations. This work was supported by research grants SERB/F/3279 from Science and Engineering Research Board India , research grant K1052 and baseline research grant BAS/1/1349-01-01 from KAUST , and R&D/20130307 from IIT Kanpur ; and by KAUST baseline research funds of Ravi Samtaney.

Publisher Copyright:
© 2017 Elsevier Inc.

Keywords

Extreme-resolution turbulence simulation
Fast Fourier transform
Pseudospectral method
Turbulence simulation

ASJC Scopus subject areas

Software
Theoretical Computer Science
Hardware and Architecture
Computer Networks and Communications
Artificial Intelligence

Access to Document

10.1016/j.jpdc.2017.10.014

Cite this

@article{9191048885f240bda4e2210b7e54db6f,

title = "Scaling of a Fast Fourier Transform and a pseudo-spectral fluid solver up to 196608 cores",

abstract = "In this paper we present scaling results of a FFT library, FFTK, and a pseudospectral code, Tarang, on grid resolutions up to 81923 grid using 65536 cores of Blue Gene/P and 196608 cores of Cray XC40 supercomputers. We observe that communication dominates computation, more so on the Cray XC40. The computation time scales as Tcomp∼p−1, and the communication time as Tcomm∼n−γ2 with γ2 ranging from 0.7 to 0.9 for Blue Gene/P, and from 0.43 to 0.73 for Cray XC40. FFTK, and the fluid and convection solvers of Tarang exhibit weak as well as strong scaling nearly up to 196608 cores of Cray XC40. We perform a comparative study of the performance on the Blue Gene/P and Cray XC40 clusters.",

keywords = "Extreme-resolution turbulence simulation, Fast Fourier transform, Pseudospectral method, Turbulence simulation",

author = "Chatterjee, {Anando G.} and Verma, {Mahendra K.} and Abhishek Kumar and Ravi Samtaney and Bilel Hadri and Rooh Khurram",

note = "Funding Information: We thank Shashwat Bhattacharya for his valuable help on plotting. Our numerical simulations were performed at Cray XC40 Shaheen II at KAUST supercomputing laboratory (KSL), Saudi Arabia. We thank KAUST scientists for the kind support while performing our simulations. This work was supported by research grants SERB/F/3279 from Science and Engineering Research Board India , research grant K1052 and baseline research grant BAS/1/1349-01-01 from KAUST , and R&D/20130307 from IIT Kanpur ; and by KAUST baseline research funds of Ravi Samtaney. Publisher Copyright: {\textcopyright} 2017 Elsevier Inc.",

year = "2018",

month = mar,

doi = "10.1016/j.jpdc.2017.10.014",

language = "English (US)",

volume = "113",

pages = "77--91",

journal = "Journal of Parallel and Distributed Computing",

issn = "0743-7315",

publisher = "Elsevier",

}

TY - JOUR

T1 - Scaling of a Fast Fourier Transform and a pseudo-spectral fluid solver up to 196608 cores

AU - Chatterjee, Anando G.

AU - Verma, Mahendra K.

AU - Kumar, Abhishek

AU - Samtaney, Ravi

AU - Hadri, Bilel

AU - Khurram, Rooh

N1 - Funding Information: We thank Shashwat Bhattacharya for his valuable help on plotting. Our numerical simulations were performed at Cray XC40 Shaheen II at KAUST supercomputing laboratory (KSL), Saudi Arabia. We thank KAUST scientists for the kind support while performing our simulations. This work was supported by research grants SERB/F/3279 from Science and Engineering Research Board India , research grant K1052 and baseline research grant BAS/1/1349-01-01 from KAUST , and R&D/20130307 from IIT Kanpur ; and by KAUST baseline research funds of Ravi Samtaney. Publisher Copyright: © 2017 Elsevier Inc.

PY - 2018/3

Y1 - 2018/3

N2 - In this paper we present scaling results of a FFT library, FFTK, and a pseudospectral code, Tarang, on grid resolutions up to 81923 grid using 65536 cores of Blue Gene/P and 196608 cores of Cray XC40 supercomputers. We observe that communication dominates computation, more so on the Cray XC40. The computation time scales as Tcomp∼p−1, and the communication time as Tcomm∼n−γ2 with γ2 ranging from 0.7 to 0.9 for Blue Gene/P, and from 0.43 to 0.73 for Cray XC40. FFTK, and the fluid and convection solvers of Tarang exhibit weak as well as strong scaling nearly up to 196608 cores of Cray XC40. We perform a comparative study of the performance on the Blue Gene/P and Cray XC40 clusters.

AB - In this paper we present scaling results of a FFT library, FFTK, and a pseudospectral code, Tarang, on grid resolutions up to 81923 grid using 65536 cores of Blue Gene/P and 196608 cores of Cray XC40 supercomputers. We observe that communication dominates computation, more so on the Cray XC40. The computation time scales as Tcomp∼p−1, and the communication time as Tcomm∼n−γ2 with γ2 ranging from 0.7 to 0.9 for Blue Gene/P, and from 0.43 to 0.73 for Cray XC40. FFTK, and the fluid and convection solvers of Tarang exhibit weak as well as strong scaling nearly up to 196608 cores of Cray XC40. We perform a comparative study of the performance on the Blue Gene/P and Cray XC40 clusters.

KW - Extreme-resolution turbulence simulation

KW - Fast Fourier transform

KW - Pseudospectral method

KW - Turbulence simulation

UR - http://www.scopus.com/inward/record.url?scp=85034626510&partnerID=8YFLogxK

U2 - 10.1016/j.jpdc.2017.10.014

DO - 10.1016/j.jpdc.2017.10.014

M3 - Article

AN - SCOPUS:85034626510

SN - 0743-7315

VL - 113

SP - 77

EP - 91

JO - Journal of Parallel and Distributed Computing

JF - Journal of Parallel and Distributed Computing

ER -

Scaling of a Fast Fourier Transform and a pseudo-spectral fluid solver up to 196608 cores

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this