A GPU-based Method for Weakly Compressible Fluids

In this paper we present an efficient GPU pipeline to perform high quality fluid simulation using CUDA. Our method extends standard weakly compressible smoothed particles hydrodynamics to realistically handle complex boundaries. We design a GPU pipeline of the proposed approach and map the entire components of the pipeline to CUDA kernels and data structures, fully exploiting the massive computational power of state-of-the-art GPUs. We describe the GPU pipeline and parallel techniques that can improve performance on latest GPUs. Furthermore, we make use of efficient parallel sorting algorithms for the time-consuming neighbor search process. On a high-end NVIDIA Geforce GTX 780, we observe up to 33x speedup as compared to a single-threaded CPU-based implementation.