GREAPMC: development status, capabilities, and future developments

This article focuses on the capabilities, performance, and verification of G PU-optimized REA ctor P hysics M onte C arlo (GREAPMC), a multigroup Monte Carlo code, against multigroup simulation on MCS; an in-house, CPU-based, Monte Carlo code. The simulation results for a three-dimensional fuel assembly indicate a superior performance of GREAPMC over the conventional MC codes optimized for CPU with one GPU card equivalent to approximately 13 CPU boards for the hardware employed in this work. The immense potential of GPUs has led to the development of GPU-aware continuous energy capability which is in progress with the target velocity sampling (TVS) part completed. The essence of using GPU for continuous energy MC is quantified by simulating the effect of the number of histories on the execution time for the TVS methods. The results show a remarkable acceleration compared to the CPU. The findings provide a basis for further development in GREAPMC.