That program looks awesome. Will have to give it a try. Though not sure if my GPU is good enough. Got a 512mb 8800 gts i think .need to check when back home
Your 8800GTS is a G80 part and should have 128 cores. For any highly parallel, streaming data intensive application it's going to wipe the floor with your CPU. Exceptions would only include tasks where threads take different directions on conditional statements. CPUs are good at this kind of work but GPUs can be a lot less forgiving.
-edit-
Speaking of which, I recently upgraded my GPGPU playground to CUDA 4 / OpenCL. Now I just need to make some time to enjoy it...
