DOI QR코드

DOI QR Code

Limits on the efficiency of event-based algorithms for Monte Carlo neutron transport

  • Romano, Paul K. (Argonne National Laboratory, Mathematics and Computer Science Division) ;
  • Siegel, Andrew R. (Argonne National Laboratory, Mathematics and Computer Science Division)
  • Received : 2017.05.31
  • Accepted : 2017.06.22
  • Published : 2017.09.25

Abstract

The traditional form of parallelism in Monte Carlo particle transport simulations, wherein each individual particle history is considered a unit of work, does not lend itself well to data-level parallelism. Event-based algorithms, which were originally used for simulations on vector processors, may offer a path toward better utilizing data-level parallelism in modern computer architectures. In this study, a simple model is developed for estimating the efficiency of the event-based particle transport algorithm under two sets of assumptions. Data collected from simulations of four reactor problems using OpenMC was then used in conjunction with the models to calculate the speedup due to vectorization as a function of the size of the particle bank and the vector width. When each event type is assumed to have constant execution time, the achievable speedup is directly related to the particle bank size. We observed that the bank size generally needs to be at least 20 times greater than vector size to achieve vector efficiency greater than 90%. When the execution times for events are allowed to vary, the vector speedup is also limited by differences in the execution time for events being carried out in a single event-iteration.

Keywords

References

  1. F.B. Brown, W.R. Martin, Monte Carlo methods for radiation transport analysis on vector computers, Prog. Nucl. Energy 14 (3) (1984) 269-299. https://doi.org/10.1016/0149-1970(84)90024-6
  2. W.R. Martin, F.B. Brown, Status of vectorized Monte Carlo for particle transport analysis, Int. J. Supercomput. Appl. 1 (2) (1987) 11-32. https://doi.org/10.1177/109434208700100203
  3. F.B. Brown, Vectorization of three-dimensional general-geometry Monte Carlo, Trans. Am. Nucl. Soc. 53 (1986) 283-285.
  4. X. Du, T. Liu, W. Ji, X.G. Xu, F.B. Brown, Evaluation of vectorized Monte Carlo algorithms on GPUs for a neutron eigenvalue problem, in: M&C, Sun Valley, Idaho, May 5-9 2013.
  5. T. Liu, X. Du, W. Ji, X.G. Xu, F.B. Brown, A comparative study of history-based versus vectorized Monte Carlo methods in the GPU/CUDA environment for a simple neutron eigenvalue problem, in: SNA + MC, Paris, France, 2014.
  6. P.S. Brantley, S.A. Dawson, M.S. Mckinley, M.J. O'Brien, D.E. Stevens, B.R. Beck, I.Eugene D. Brooks, Advanced computing architecture challenges for the Mercury Monte Carlo particle transport project, in: Joint Int. Conf. on Mathematics and Computation, Supercomputing in Nuclear Applications, and the Monte Carlo Method, Nashville, Tennessee, Apr. 19-23 2015.
  7. R.M. Bergmann, J.L. Vujic, Algorithmic choices in WARP - a framework for continuous energy Monte Carlo neutron transport in general 3D geometries on GPUs, Ann. Nucl. Energy 77 (2015) 176-193. https://doi.org/10.1016/j.anucene.2014.10.039
  8. S.P. Hamilton, T.M. Evans, S.R. Slattery, GPU acceleration of history-based multigroup Monte Carlo, Trans. Am. Nucl. Soc. 115 (2016) 527-530.
  9. J. Apostolakis, R. Brun, F. Carminati, A. Gheata, S. Wenzel, A concurrent vectorbased steering framework for particle transport, J. Phys. Conf. Ser. 523 (2014) 012004. https://doi.org/10.1088/1742-6596/523/1/012004
  10. J. Apostolakis, M. Bandieramonte, G. Bitzes, R. Brun, P. Canal, F. Carminati, J.C. De Fine Licht, L. Duhem, V.D. Elvira, A. Gheata, S.Y. Jun, G. Lima, M. Novak, R. Sehgal, O. Shadura, S. Wenzel, Adaptive track scheduling to optimize concurrency and vectorization in Geant V, J. Phys. Conf. Ser. 608 (2015) 012003. https://doi.org/10.1088/1742-6596/608/1/012003
  11. G. Amadio, J. Apostolakis, M. Bandieramonte, C. Bianchini, G. Bitzes, R. Brun, P. Canal, F. Carminati, J. De Fine Licht, L. Duhem, D. Elvira, A. Gheata, S.Y. Jun, G. Lima, M. Novak, M. Presbyterian, O. Shadura, R. Seghal, S. Wenzel, First experience of vectorizing electromagnetic physics models for detector simulation, J. Phys. Conf. Ser. 664 (2015) 092013. https://doi.org/10.1088/1742-6596/664/9/092013
  12. P.K. Romano, N.E. Horelik, B.R. Herman, A.G. Nelson, B. Forget, OpenMC: a state-of-the-art Monte Carlo code for research and development, Ann. Nucl. Energy 82 (2015) 90-97. https://doi.org/10.1016/j.anucene.2014.07.048
  13. N. Horelik, B. Herman, B. Forget, K. Smith, Benchmark for Evaluation and Validation of Reactor Simulations (BEAVRS), in: Int. Conf. Mathematics and Computational Methods Applied to Nuclear Science and Engineering, Sun Valley, Idaho, May 5-9 2013.
  14. J.-Y. Lee, S.R. Choi, S.J. Kim, PGSFR core design and performance characteristics, Trans. Am. Nucl. Soc. 114 (2016) 700-703.
  15. N.E. Stauff, C. Lee, P.K. Romano, T.K. Kim, Verification of mixed stochastic/ deterministic approach for fast and thermal reactor analysis, in: International Congress on Advances in Nuclear Power Plants, Fukui and Kyoto, Japan, Apr. 24-28 2017.
  16. J.R. Tramm, A.R. Siegel, Memory bottlenecks and memory contention in multi-core Monte Carlo transport codes, Ann. Nucl. Energy 82 (2015) 195-202. https://doi.org/10.1016/j.anucene.2014.08.038
  17. P.K. Romano, A.R. Siegel, R.O. Rahaman, Influence of the memory subsystem on Monte Carlo code performance, in: Joint Int. Conf. Mathematics and Computation, Supercomputing in Nuclear Applications, and the Monte Carlo method, Knoxville, Tennessee, Apr. 19-23 2015.