Concurrency and Real Physics / Simulations

[subvertio]

I am looking for a physics simulation library that takes advantage of all available CPU Cores, I've looked into bullet and PhysX (which don't). I'm looking at SolidWorks soon too. I don't really need real-time rendering, I just want answers as close as possible to real world results just using Newtonian physics, possibly with elastomer / soft bodies too.

hardware requirements like CUDA, XEON, XEON Phi or other scalable concurrent architectures are ok too.

...any recommendations?

SubV