Any sort of computation of a collection of data where elements interact with other elements with high locality is extremely efficient on a GPU. In fact, the GPU can be from 10 to 100 times faster than a Pentium 4 in such situations, says Kirk. Scientific computations such as linear algebra, Fast Fourier Transforms, and partial differential equations can also benefit greatly from GPUs. One research group is looking to break the Linpack benchmark world record using a cluster of 256 PCs with GeForce FXs! Another scenario that surprised Kirk was when a development team recently tried to use a GeForce FX CPU to perform database key sorting. When coded to take advantage of the parallelism and pipelining in the GPU, parallel key sorts and searches can occur much faster than using general CPUs, on the order of 4-5x faster than the Pentium 4, according to Kirk.
Just thought you guys might be interested. Keep in mind ExtremeTech is owned by Ziff-Davis