Parallel performance results for the OpenMOC neutron transport code on multicore platforms

Abstract

The shift towards multi-core architectures has ushered in a new era of shared memory parallelism for scientific applications. This transition has introduced challenges for the nuclear engineering community as it seeks to design high-fidelity full-core reactor physics simulation tools. This paper describes the parallel transport sweep algorithm in the OpenMOC method of characteristics (MOC) neutron transport code for multi-core platforms using OpenMP. Strong and weak scaling studies are performed for both Intel Xeon and IBM Blue Gene/Q multi-core processors. The results demonstrate 100% parallel efficiency for 12 threads on 12 cores on Intel Xeon platforms, and over 90% parallel efficiency with 64 threads on 16 cores on the IBM Blue Gene/Q. These results illustrate the potential for hardware acceleration for MOC neutron transport on modern multi-core and future many-core architectures. In addition, this work highlights the pitfalls of programming for multi-core architectures, with a focal point on false sharing.

Hybrid parallel reduction algorithms for the multi-level CMFD acceleration in the neutron transport code PANDAS-MOC

“…With the considerable increase in the computational power, parallel computing has been widely considered in the high-fidelity neutronics analysis codes, such as DeCART (Joo et al, 2004), MPACT (Larsen et al, 2019), nTRACER (Choi et al, 2018), OpenMOC (Boyd et al, 2016), STREAM (Choi et al, 2021), ARCHER (Zhu et al, 2022) etc., in order to improve the performance of solving the neutron transport equations in the nuclear reactors. Similarly, several parallel models of the PANDAS-MOC have been developed based on the nature of distributed and shared memory architectures, including the Pure MPI parallel model (PMPI), Segment OpenMP threading hybrid model (SGP), and Whole-code OpenMP threading hybrid model (WCP).…”

Parallel schedules for MOC sweep in the neutron transport code PANDAS-MOC

“…Additionally, the parallel transport sweep algorithm of OpenMOC has two versions. One uses the omp atomic clause and the other does not use the omp atomic clause (Boyd et al, 2016). Nevertheless, the algorithm without an atomic clause is finished by introducing extra thread private arrays to store the intermediate scalar fluxes, and then, the FSR scalar fluxes are reduced across threads, which, however, costs significant amount of memory in this procedure.…”