GPU/CUDA-Ready Parallel Monte Carlo Codes for Reactor Analysis and Other Applications

摘要

Monte Carlo simulation is widely used in reactorrnanalysis and medical physics for its high accuracy. Therntime-consuming process of achieving a desired precisionrncan be significantly facilitated by parallel computingrnmethodologies. The exascale High PerformancernComputing (HPC) architecture, which is expected tornarrive within this decade has the potential to furtherrnaccelerate the Monte Carlo calculations. However, thernever-increasing parallelism is facing at least twornchallenges. One is the fact that the power consumptionrnincreases with the Floating Point Operations Per Secondrn(FLOPS).