http://www.nvidia.com/object/tesla_computing_solutions.html

newcfd

Running electromagnetic simulations using NVIDIA';s compute hardware and CUDA accelerate processing times by factors of 25 or more

newcfd

the GPU is architected for compute-intensive,
highly parallel computation–exactly what is required for graphics rendering.
Therefore on a GPU, more transistors are devoted to data processing than are to
data caching and flow control
The GPU is especially well-suited to address problems that can be expressed as
data-parallel computations with high arithmetic intensity–in other words when the
same program is executed on many data elements in parallel with a high ratio of
arithmetic to memory operations.