I have been playing with a new idea from Prof. Filipone Salvatore group in particle simulation. It seems to work fine, we will use it with my student. Just for fun I have created a demo to show my student the way of this method. It uses global memory, aspects like using texture memory as they say in the paper later on in our project, our main goal is not to make it as fast as possible. Of course making it as fast as possible is also a sub-goal.
Executable (checked for viruses with NOD32)
The cube rotates with UP/DOWN/RIGHT/LEFT buttons, nothing fancy, after all it was just created in less than one day. :-)
It works on cards with SM >= 2 (Fermi up). Of course the specific atomic is faster with SM >= 3 but I want people to enjoy the demo with older cards.