We implement a high-order finite-element application, which performs the
numerical simulation of seismic wave propagation resulting for instance
from earthquakes at the scale of a continent or from active seismic
acquisition experiments in the oil industry, on a large GPU-enhanced
cluster. Mesh coloring enables an efficient accumulation of degrees of
freedom in the assembly process over an unstructured mesh. We use
non-blocking MPI and show that computations and communications over the
network and between the CPUs and the GPUs are almost fully overlapped.
The GPU solver scales excellently up to 192 GPUs and achie/-ves
significant speedup over a carefully tuned equivalent CPU code.