Limits on the efficiency of event-based algorithms for Monte Carlo neutron transport
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...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2017-09-01
|
| Series: | Nuclear Engineering and Technology |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1738573317302966 |
| _version_ | 1828784544327663616 |
|---|---|
| author | Paul K. Romano Andrew R. Siegel |
| author_facet | Paul K. Romano Andrew R. Siegel |
| author_sort | Paul K. Romano |
| collection | DOAJ |
| description | 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. |
| first_indexed | 2024-12-11T23:33:23Z |
| format | Article |
| id | doaj.art-b668fe9ff2f342cca99b671df6e9d30c |
| institution | Directory Open Access Journal |
| issn | 1738-5733 |
| language | English |
| last_indexed | 2024-12-11T23:33:23Z |
| publishDate | 2017-09-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Nuclear Engineering and Technology |
| spelling | doaj.art-b668fe9ff2f342cca99b671df6e9d30c2022-12-22T00:45:57ZengElsevierNuclear Engineering and Technology1738-57332017-09-014961165117110.1016/j.net.2017.06.006Limits on the efficiency of event-based algorithms for Monte Carlo neutron transportPaul K. RomanoAndrew R. SiegelThe 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.http://www.sciencedirect.com/science/article/pii/S1738573317302966Monte CarloNeutron transportParallelismVectorization |
| spellingShingle | Paul K. Romano Andrew R. Siegel Limits on the efficiency of event-based algorithms for Monte Carlo neutron transport Nuclear Engineering and Technology Monte Carlo Neutron transport Parallelism Vectorization |
| title | Limits on the efficiency of event-based algorithms for Monte Carlo neutron transport |
| title_full | Limits on the efficiency of event-based algorithms for Monte Carlo neutron transport |
| title_fullStr | Limits on the efficiency of event-based algorithms for Monte Carlo neutron transport |
| title_full_unstemmed | Limits on the efficiency of event-based algorithms for Monte Carlo neutron transport |
| title_short | Limits on the efficiency of event-based algorithms for Monte Carlo neutron transport |
| title_sort | limits on the efficiency of event based algorithms for monte carlo neutron transport |
| topic | Monte Carlo Neutron transport Parallelism Vectorization |
| url | http://www.sciencedirect.com/science/article/pii/S1738573317302966 |
| work_keys_str_mv | AT paulkromano limitsontheefficiencyofeventbasedalgorithmsformontecarloneutrontransport AT andrewrsiegel limitsontheefficiencyofeventbasedalgorithmsformontecarloneutrontransport |