high performance computing on graphics processing units: hgpu.org

The reliability analysis of critical systems can be performed using fault tree analysis. One of the common approaches used for fault tree analysis is Monte Carlo simulation. The purpose of this paper is therefore to show an algorithm to speed up Monte Carlo simulation for analyzing fault tree with parallel computing in GPU. To this end, we use time-to-failure tree to model fault tree with Compute Unified Device Architecture (CUDA) which is used to accelerate the execution of loops with many repetitions. We also use this technique to accelerate Monte Carlo simulations. In addition, we visualize fault tree so that the user is able to generate fault tree in detail using our developed software and can execute it. The computational outcomes validate the effectiveness of the suggested approach, as we approached about 310 times speedup in large fault trees.