Acceleration of turbomachinery steady simulations on GPU

Mohamed Aissa; Lasse  Müller; Tom Verstraete; Cornelis Vuik

doi:10.1007/978-3-319-58943-5_65

Acceleration of turbomachinery steady simulations on GPU

Mohamed Aissa^*, Lasse Müller, Tom Verstraete, Cornelis Vuik

^*Corresponding author for this work

Numerical Analysis

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

Abstract

Steady state simulations in Computational Fluid Dynamics (CFD), which rely on implicit time integration, are not experiencing great accelerations on GPUs. Moreover, most of the reported acceleration effort concerns solving the linear system of equations while neglecting the acceleration potential of running the entire simulation on the GPU. In this paper, we present the software implementation of an implicit RANS CFD solver, which is fully running on GPU. We use the GMRES linear solver of the Paralution package combined with the incomplete LU factorization for the preconditioning. We propose also a control mechanism -on-demand factorization - capable of reducing the number of times an incomplete LU factorization is performed. The on-demand factorization accelerates the linear solver without altering the flow convergence. The GPU implementation achieved a speedups of 9.2x compared to a single-core CPU and 3.5x compared to a 4-cores CPU for 3-D flow predictions in turbine applications.

Original language	English
Title of host publication	Euro-Par 2016: Parallel Processing Workshops - Euro-Par 2016 International Workshops, Revised Selected Papers
Editors	F. Desprez, P. Dutot, C. Kaklamanis, L Marchal, K Molitorisz, L. ricci, V. Scarano, M.A. Vega-Rodríguez, A.L. Varbanescu, S. Hunold, S.L. Scott, S. Lankes, J. Weidendorfer
Place of Publication	Cham
Publisher	Springer
Pages	814-825
Number of pages	12
ISBN (Electronic)	978-3-319-58943-5
ISBN (Print)	978-3-319-58942-8
DOIs	https://doi.org/10.1007/978-3-319-58943-5_65
Publication status	Published - 2017
Event	Euro-Par 2016 22nd International Conference on Parallel and Distributed Computing - Grenoble, France Duration: 24 Aug 2016 → 26 Aug 2016

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10104
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	Euro-Par 2016 22nd International Conference on Parallel and Distributed Computing
Country/Territory	France
City	Grenoble
Period	24/08/16 → 26/08/16

Keywords

GMRES
GPU
ILU
Krylov subspace
Linear systems
Steady CFD

Access to Document

10.1007/978-3-319-58943-5_65

Cite this

Aissa, M., Müller, L., Verstraete, T., & Vuik, C. (2017). Acceleration of turbomachinery steady simulations on GPU. In F. Desprez, P. Dutot, C. Kaklamanis, L. Marchal, K. Molitorisz, L. ricci, V. Scarano, M. A. Vega-Rodríguez, A. L. Varbanescu, S. Hunold, S. L. Scott, S. Lankes, & J. Weidendorfer (Eds.), Euro-Par 2016: Parallel Processing Workshops - Euro-Par 2016 International Workshops, Revised Selected Papers (pp. 814-825). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10104 ). Springer. https://doi.org/10.1007/978-3-319-58943-5_65

Aissa, Mohamed ; Müller, Lasse ; Verstraete, Tom et al. / Acceleration of turbomachinery steady simulations on GPU. Euro-Par 2016: Parallel Processing Workshops - Euro-Par 2016 International Workshops, Revised Selected Papers. editor / F. Desprez ; P. Dutot ; C. Kaklamanis ; L Marchal ; K Molitorisz ; L. ricci ; V. Scarano ; M.A. Vega-Rodríguez ; A.L. Varbanescu ; S. Hunold ; S.L. Scott ; S. Lankes ; J. Weidendorfer. Cham : Springer, 2017. pp. 814-825 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{3fff409a50c048babf058fd8be3c43b0,

title = "Acceleration of turbomachinery steady simulations on GPU",

abstract = "Steady state simulations in Computational Fluid Dynamics (CFD), which rely on implicit time integration, are not experiencing great accelerations on GPUs. Moreover, most of the reported acceleration effort concerns solving the linear system of equations while neglecting the acceleration potential of running the entire simulation on the GPU. In this paper, we present the software implementation of an implicit RANS CFD solver, which is fully running on GPU. We use the GMRES linear solver of the Paralution package combined with the incomplete LU factorization for the preconditioning. We propose also a control mechanism -on-demand factorization - capable of reducing the number of times an incomplete LU factorization is performed. The on-demand factorization accelerates the linear solver without altering the flow convergence. The GPU implementation achieved a speedups of 9.2x compared to a single-core CPU and 3.5x compared to a 4-cores CPU for 3-D flow predictions in turbine applications.",

keywords = "GMRES, GPU, ILU, Krylov subspace, Linear systems, Steady CFD",

author = "Mohamed Aissa and Lasse M{\"u}ller and Tom Verstraete and Cornelis Vuik",

year = "2017",

doi = "10.1007/978-3-319-58943-5_65",

language = "English",

isbn = "978-3-319-58942-8",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "814--825",

editor = "F. Desprez and P. Dutot and C. Kaklamanis and L Marchal and K Molitorisz and L. ricci and V. Scarano and M.A. Vega-Rodr{\'i}guez and A.L. Varbanescu and S. Hunold and S.L. Scott and S. Lankes and J. Weidendorfer",

booktitle = "Euro-Par 2016: Parallel Processing Workshops - Euro-Par 2016 International Workshops, Revised Selected Papers",

note = "Euro-Par 2016 22nd International Conference on Parallel and Distributed Computing ; Conference date: 24-08-2016 Through 26-08-2016",

}

Aissa, M, Müller, L, Verstraete, T & Vuik, C 2017, Acceleration of turbomachinery steady simulations on GPU. in F Desprez, P Dutot, C Kaklamanis, L Marchal, K Molitorisz, L ricci, V Scarano, MA Vega-Rodríguez, AL Varbanescu, S Hunold, SL Scott, S Lankes & J Weidendorfer (eds), Euro-Par 2016: Parallel Processing Workshops - Euro-Par 2016 International Workshops, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10104 , Springer, Cham, pp. 814-825, Euro-Par 2016 22nd International Conference on Parallel and Distributed Computing, Grenoble, France, 24/08/16. https://doi.org/10.1007/978-3-319-58943-5_65

Acceleration of turbomachinery steady simulations on GPU. / Aissa, Mohamed; Müller, Lasse ; Verstraete, Tom et al.
Euro-Par 2016: Parallel Processing Workshops - Euro-Par 2016 International Workshops, Revised Selected Papers. ed. / F. Desprez; P. Dutot; C. Kaklamanis; L Marchal; K Molitorisz; L. ricci; V. Scarano; M.A. Vega-Rodríguez; A.L. Varbanescu; S. Hunold; S.L. Scott; S. Lankes; J. Weidendorfer. Cham: Springer, 2017. p. 814-825 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10104 ).

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Acceleration of turbomachinery steady simulations on GPU

AU - Aissa, Mohamed

AU - Müller, Lasse

AU - Verstraete, Tom

AU - Vuik, Cornelis

PY - 2017

Y1 - 2017

N2 - Steady state simulations in Computational Fluid Dynamics (CFD), which rely on implicit time integration, are not experiencing great accelerations on GPUs. Moreover, most of the reported acceleration effort concerns solving the linear system of equations while neglecting the acceleration potential of running the entire simulation on the GPU. In this paper, we present the software implementation of an implicit RANS CFD solver, which is fully running on GPU. We use the GMRES linear solver of the Paralution package combined with the incomplete LU factorization for the preconditioning. We propose also a control mechanism -on-demand factorization - capable of reducing the number of times an incomplete LU factorization is performed. The on-demand factorization accelerates the linear solver without altering the flow convergence. The GPU implementation achieved a speedups of 9.2x compared to a single-core CPU and 3.5x compared to a 4-cores CPU for 3-D flow predictions in turbine applications.

AB - Steady state simulations in Computational Fluid Dynamics (CFD), which rely on implicit time integration, are not experiencing great accelerations on GPUs. Moreover, most of the reported acceleration effort concerns solving the linear system of equations while neglecting the acceleration potential of running the entire simulation on the GPU. In this paper, we present the software implementation of an implicit RANS CFD solver, which is fully running on GPU. We use the GMRES linear solver of the Paralution package combined with the incomplete LU factorization for the preconditioning. We propose also a control mechanism -on-demand factorization - capable of reducing the number of times an incomplete LU factorization is performed. The on-demand factorization accelerates the linear solver without altering the flow convergence. The GPU implementation achieved a speedups of 9.2x compared to a single-core CPU and 3.5x compared to a 4-cores CPU for 3-D flow predictions in turbine applications.

KW - GMRES

KW - GPU

KW - ILU

KW - Krylov subspace

KW - Linear systems

KW - Steady CFD

UR - http://www.scopus.com/inward/record.url?scp=85020447839&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-58943-5_65

DO - 10.1007/978-3-319-58943-5_65

M3 - Conference contribution

AN - SCOPUS:85020447839

SN - 978-3-319-58942-8

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

SP - 814

EP - 825

BT - Euro-Par 2016: Parallel Processing Workshops - Euro-Par 2016 International Workshops, Revised Selected Papers

A2 - Desprez, F.

A2 - Dutot, P.

A2 - Kaklamanis, C.

A2 - Marchal, L

A2 - Molitorisz, K

A2 - ricci, L.

A2 - Scarano, V.

A2 - Vega-Rodríguez, M.A.

A2 - Varbanescu, A.L.

A2 - Hunold, S.

A2 - Scott, S.L.

A2 - Lankes, S.

A2 - Weidendorfer, J.

PB - Springer

CY - Cham

T2 - Euro-Par 2016 22nd International Conference on Parallel and Distributed Computing

Y2 - 24 August 2016 through 26 August 2016

ER -

Aissa M, Müller L, Verstraete T, Vuik C. Acceleration of turbomachinery steady simulations on GPU. In Desprez F, Dutot P, Kaklamanis C, Marchal L, Molitorisz K, ricci L, Scarano V, Vega-Rodríguez MA, Varbanescu AL, Hunold S, Scott SL, Lankes S, Weidendorfer J, editors, Euro-Par 2016: Parallel Processing Workshops - Euro-Par 2016 International Workshops, Revised Selected Papers. Cham: Springer. 2017. p. 814-825. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-58943-5_65

Acceleration of turbomachinery steady simulations on GPU

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this