GPU-based Particle-in-Cell Simulation

Accelerating Plasma Simulations using novel computing hardware is a promising way towards a cost-efficient decrease of the runtime of realistic plasma simulations. We present a new particle-in-cell algorithm for simulating laser plasma acceleration of particles which has been developed to run on a NVIDIA GPU system. Using a sliding-window technique we are able to run large-scale 2D simulations of laser wakefield acceleration. We present the algorithm in detail and show that it is easy to extend to 3D geometries. We will conclude by discussing both obstacles and promises encountered when porting the algorithm to a standard CPU-based Linux cluster in which each node is equipped with one or more GPUs.