Newton on GPU (CUDA, OpenCL)

[thedmd]

I mean a kind of stress test of various parts of the engine. That may be useful to track how the engine is improved, how various optimizations influence the CPU/GPU load. And most of all to prove Newton is doing things better than alternative solutions. Numbers have force of convincing the people much better than arguments. They do not require people to think, to make the judgment.

Reason for stress test is to see where are the limits of current implementation. OpenCL give the opportunity to get significant performance boost. If this is the case, people should see this. This take Newton closer to be widely used (portable) physics engine.

In my personal opinion, your work there, Newton, deserve much more attention.

[jiandingzhe]

What is the current status of massive-parallel-accelerated Newton?
I'm making a game with maybe thousands of rigid bodies, so I desire that a lot.
I will wait for that, because I planned to finish my game before I'm retired . There are so many years until that...