GPU Acceleration of Monte Carlo Tree Search Algorithm for Amazon chess and Its Evaluation Function

The evaluation function is a very important part of the Monte Carlo Tree Search (MCTS) algorithm for Amazon chess. This paper introduces the principle of Monte Carlo Tree Search algorithm, and realizes GPU accelerated calculation of Amazon chess situation's evaluation function, based on CUDA computing platform. Using the CPU code and CUDA code, two experiments were carried out to evaluate the function calculation accuracy and the effect of kernel function call parameters on the GPU acceleration performance. The results show that the evaluation function is feasible and effective for GPU acceleration, and the speedup ratio of GPU to CPU reaches 20.0.