Charge Mapping is critical to electrostatics computations such as performed in Molecular Dynamics simulations. It reduces the complexity of evaluating long-range Coulombic forces by diffusing discrete particle charges onto a regular grid. Acceleration of this stage using GPUs is non-trivial, with the compute and memory intensive nature of the algorithm limiting performance benefits in naive implementations. In this paper, we explore methods for performing efficient charge mapping on GPUs. By utilizing available resources effectively and reducing compute and memory transactions, high throughput can be achieved. Our best case implementation shows > 14 × speed-up over existing GPU codes and > 25 × speed-up over production CPU codes such as NAMD.