Published 2022 | Version v1
Book

Toward portable GPU acceleration of the OpenMC Monte Carlo particle transport code

  • 1. Argonne National Laboratory, 9700 S Cass Ave., Lemont, IL, 60439 (United States)
  • 2. Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA, 02139 (United States)
  • 3. University of New South Wales, Sydney, NSW 2502 (Australia)

Description

OpenMC is an open source Monte Carlo particle transport code originally designed for CPU-based systems. In this work, we use the OpenMP target off-load programming model to port OpenMC to a variety of GPU architectures from NVIDIA and AMD. We detail our porting strategy and provide preliminary performance comparisons. The GPU version of OpenMC run on an NVIDIA A100 GPU achieves performance equivalent to approximately 200 Xeon CPU cores for a depleted fuel reactor benchmark problem. (authors)

Availability note (English)

Available (CD Rom) from the American Nuclear Society, 555 North Kensington Avenue, La Grange Park, Illinois 60526 (US)
Part of:
Proceedings of the international conference on physics of reactors - Physor 2022

Additional details

Publishing Information

Publisher
ANS - American Nuclear Society
Imprint Place
La Grange Park (United States)
ISBN
978-0-89448-787-3
Imprint Title
Proceedings of the international conference on physics of reactors - PHYSOR 2022
Imprint Pagination
3701 p.
Journal Page Range
p. 2760-2769

Conference

Title
International conference on physics of reactors
Acronym
PHYSOR 2022
Dates
15-20 May 2022
Place
Pittsburg (United States)

INIS

Country of Publication
United States
Country of Input or Organization
France
INIS RN
54042135
Subject category
S99: GENERAL AND MISCELLANEOUS;
Resource subtype / Literary indicator
Conference
Descriptors DEI
COMPUTER ARCHITECTURE; COMPUTERIZED SIMULATION; MONTE CARLO METHOD; O CODES; PROGRAMMING; RADIATION TRANSPORT
Descriptors DEC
CALCULATION METHODS; COMPUTER CODES; SIMULATION

Optional Information

Notes
12 refs.