Some people have a bit of trouble wrapping their head around the basic concept of GPGPU programming. Now I could drone on and on trying to explain it, but if a picture is worth a thousand words then a video must be worth a couple of million. Instead of wasting a few days reading a couple of million words why don't you just watch this 94 second video. (Talk about an optimization...)

See how much faster your work gets done when you perform it in parallel.