A Fast Eikonal Equation Solver for Parallel Systems

This paper presents a novel solver for the eikonal equation that is designed to run efficiently on massively parallel systems. The proposed method manages a list of active nodes and iteratively updates the solutions on those grid points until they converge. The management of the list does not entail the extra burden of ordered data structures. The proposed method has suboptimal worst-case performance, but in practice, on real and synthetic datasets, it performs fewer computations per node than optimal alternatives. Furthermore, the proposed method uses local, synchronous updates and therefore has better cache coherency, is simple to implement, and scales efficiently on parallel architectures, e.g., multiprocessor systems or GPUs (graphics processing units). The paper describes the method, the implementation on the GPU, and performance analysis that compares against the state-of-the-art eikonal solvers.