Electrostatic discharge simulation using a GPU-accelerated DGTD solver targeting modern graphics processors

The presence of graphics processors (GPUs) in supercomputers constantly increased in the last decade. FDTD and DGTD (Time Domain Discontinuous Galerkin) are traditionally employed on GPUs for their scalability, however the limitations of past hardware required particular care in the implementation in order to obtain good performance. In this work, we discuss an implementation of DGTD for the Maxwell’s equations on modern GPUs and we assess its performance on the simulation of an electrostatic discharge.