Hi! I'm new here, so sorry if this is a naive question.
Apparently bullet 3 is currently being developed, with an emphasis on running on gpu. Just curious, is the goal for this gpu version to become the standard bullet version replacing cpu-run bullet for most situations (including games), or is this more targeted for "niche" applications like 3d rendering?
As for my actual question:
Most demos of gpu-run physics engines shows simulations of thousands of simple, unconstrained objects since this is easily parallelizable. Would running bullet on gpu also offer a nice advantage for simulation of relatively few (about 30-40) objects, with several joints/constraints at more physics frames per seconds? (to stimulate higher forces/faster speeds accurately)
I doubt this kind of scenario can be parallelized well.
Thanks for taking the time to read this.