Detail publikace

A GPU solver for symmetric positive-definite matrices vs. traditional codes

BOHÁČEK, J. KARIMI-SIBAKI, E. KHARICHA, A. LUDWIG, A. WU, M. HOLZMANN, T.

Anglický název

A GPU solver for symmetric positive-definite matrices vs. traditional codes

Typ

článek v časopise ve Web of Science, Jimp

Jazyk

en

Originální abstrakt

In Heat Transfer and Fluid Flow Laboratory in Brno, the inverse heat conduction problem (IHCP) has been extensively used to reconstruct thermal boundary conditions at hot surfaces of solid materials cooled by spraying nozzles. More than three decades of experience and cooperation with industries has proven our experimental/numerical technique to be reliable and very accurate. However, a typical calculation requires relatively long calculation time. The transient heat diffusion in a multi-material sample is the most computationally costly ingredient of the algorithm. In the present paper, the potential for speeding up our calculations is manifested by firstly benchmarking it against traditional CFD codes such as OpenFOAM (FDIC) and ANSYS Fluent (AMG). Secondly, we also unveil a unique comparison between the performance of three inhouse GPU codes each written by a different PhD student/postdoc. Chronologically listed, one student pushed his luck with a fully explicit scheme, while the other two, including us, bet on implicit methods namely the line-by-line method in OpenCL and the conjugate gradient method with the deflated truncated Neumann series preconditioner in CUDA C. (C) 2019 The Authors. Published by Elsevier Ltd.

Anglický abstrakt

In Heat Transfer and Fluid Flow Laboratory in Brno, the inverse heat conduction problem (IHCP) has been extensively used to reconstruct thermal boundary conditions at hot surfaces of solid materials cooled by spraying nozzles. More than three decades of experience and cooperation with industries has proven our experimental/numerical technique to be reliable and very accurate. However, a typical calculation requires relatively long calculation time. The transient heat diffusion in a multi-material sample is the most computationally costly ingredient of the algorithm. In the present paper, the potential for speeding up our calculations is manifested by firstly benchmarking it against traditional CFD codes such as OpenFOAM (FDIC) and ANSYS Fluent (AMG). Secondly, we also unveil a unique comparison between the performance of three inhouse GPU codes each written by a different PhD student/postdoc. Chronologically listed, one student pushed his luck with a fully explicit scheme, while the other two, including us, bet on implicit methods namely the line-by-line method in OpenCL and the conjugate gradient method with the deflated truncated Neumann series preconditioner in CUDA C. (C) 2019 The Authors. Published by Elsevier Ltd.

Klíčová slova anglicky

Linear solver; Inverse task; Heat transfer; GPU; CUDA; OpenFOAM

Vydáno

07.03.2019

Nakladatel

PERGAMON-ELSEVIER SCIENCE LTD

Místo

OXFORD

ISSN

0898-1221

Ročník

78

Číslo

9

Strany od–do

2933–2943

Počet stran

11

BIBTEX


@article{BUT163681,
  author="Jan {Boháček} and Ebrahim {Karimi-Sibaki},
  title="A GPU solver for symmetric positive-definite matrices vs. traditional codes",
  year="2019",
  volume="78",
  number="9",
  month="March",
  pages="2933--2943",
  publisher="PERGAMON-ELSEVIER SCIENCE LTD",
  address="OXFORD",
  issn="0898-1221"
}