That's just it, Gérald: the speed of a Dforce Gpu simulation running on a GTX 1080 was comparable to VWD running on a core i7 4790. Of course, my system is bloated and not running at a 100%, but I would say that would go double for the CPU over the GPU. At first blush, I wasn't that impressed with Dforce and the only thing that really peaked my interest was the way the simulated object retained its rigging and morphing capabilities, but I think VWD had made its own advances in that direction with the intersection function? I am still waiting for that tutorial (not rushing you, just highlighting that, once again, you've captured my attention).