Developing a Future-Proof CFD Code

Daniel Curran; Simon N McIntosh-Smith; Christian B Allen; D.A. Beckingsale

Developing a Future-Proof CFD Code

Daniel Curran, Simon N McIntosh-Smith, Christian B Allen, D.A. Beckingsale

Research output: Contribution to conference › Conference Paper › peer-review

45 Downloads (Pure)

Abstract

The future-proof development of a compressible finite-volume RANS code is considered. An existing domain-decomposed Fortran 90/MPI production application has been redeveloped in C including a generic data communication harness, to allow exploitation of diverse hardware platforms, and future-proof the code for new highly- parallel hardware and languages. This includes development of a kernel stencil DSL and source-to-source translation for the auto generation of OpenMP, MPI, OpenCL, and CUDA code versions, and heterogeneous combinations thereof.

Original language	English
Publication status	In preparation - 20 May 2015
Event	ParCFD - Montreal, Canada Duration: 17 May 2015 → 20 May 2015

Conference

Conference	ParCFD
Country/Territory	Canada
City	Montreal
Period	17/05/15 → 20/05/15

Keywords

CFD
Future-proof
OpenCL
GPU

Access to Document

http://www2.cfdlab.mcgill.ca

Handle.net

Persistent link

1 Finished

Future-proof massively-parallel execution of multi-block applications
McIntosh-Smith, S. N.
1/10/13 → 31/03/17
Project: Research

HPC (High Performance Computing) Facility
Polly E Eccleston (Other), Simon H Atack (Other), D A G Williams (Manager), Sadaf R Alam (Manager) & Steven A Chapman (Manager)
Facility/equipment: Facility

Cite this

@conference{7d2ffbbc3e4a4ef098b26d263db420e2,

title = "Developing a Future-Proof CFD Code",

abstract = "The future-proof development of a compressible finite-volume RANS code is considered. An existing domain-decomposed Fortran 90/MPI production application has been redeveloped in C including a generic data communication harness, to allow exploitation of diverse hardware platforms, and future-proof the code for new highly- parallel hardware and languages. This includes development of a kernel stencil DSL and source-to-source translation for the auto generation of OpenMP, MPI, OpenCL, and CUDA code versions, and heterogeneous combinations thereof.",

keywords = "CFD, Future-proof, OpenCL, GPU",

author = "Daniel Curran and McIntosh-Smith, {Simon N} and Allen, {Christian B} and D.A. Beckingsale",

year = "2015",

month = may,

day = "20",

language = "English",

note = "ParCFD ; Conference date: 17-05-2015 Through 20-05-2015",

}

TY - CONF

T1 - Developing a Future-Proof CFD Code

AU - Curran, Daniel

AU - McIntosh-Smith, Simon N

AU - Allen, Christian B

AU - Beckingsale, D.A.

PY - 2015/5/20

Y1 - 2015/5/20

N2 - The future-proof development of a compressible finite-volume RANS code is considered. An existing domain-decomposed Fortran 90/MPI production application has been redeveloped in C including a generic data communication harness, to allow exploitation of diverse hardware platforms, and future-proof the code for new highly- parallel hardware and languages. This includes development of a kernel stencil DSL and source-to-source translation for the auto generation of OpenMP, MPI, OpenCL, and CUDA code versions, and heterogeneous combinations thereof.

AB - The future-proof development of a compressible finite-volume RANS code is considered. An existing domain-decomposed Fortran 90/MPI production application has been redeveloped in C including a generic data communication harness, to allow exploitation of diverse hardware platforms, and future-proof the code for new highly- parallel hardware and languages. This includes development of a kernel stencil DSL and source-to-source translation for the auto generation of OpenMP, MPI, OpenCL, and CUDA code versions, and heterogeneous combinations thereof.

KW - CFD

KW - Future-proof

KW - OpenCL

KW - GPU

M3 - Conference Paper

T2 - ParCFD

Y2 - 17 May 2015 through 20 May 2015

ER -

Developing a Future-Proof CFD Code

Abstract

Conference

Keywords

Access to Document

Handle.net

Fingerprint

Projects

Future-proof massively-parallel execution of multi-block applications

Equipment

HPC (High Performance Computing) Facility

Cite this