Problem is while MS would give you a vendor agnostic setup, they won't be platform agnostic which PhysX already is. Right now PhysX runs on far more systems used for gaming then DirectX. I'm not saying an open solution would be bad in any way at all, just pointing out that MS certainly isn't going to provide us with anything close to an open standard, PhysX is already more open in terms of developer's needs then DirectX.
Personally I would much rather see something like Bullet take off which could be run on any of the available gaming platforms which would be a far better solution then swithching between a bunch of entirely proprietary standards(PhysX, DC, Havok) all with hardware companies behind them having vested interests in making things not work on another platform.
If someone (e.g Havok etc) could code their existing physics engine to work with MS provided vendor agnostic hooks into the GPU then it wouldn't be a problem.
DirectX is MS and PC platform only (-ish) and yet an engine which uses it, e.g. Unreal Engine 3, is multi platform.
You can make a cross platform physics engine which can expose GPU hardware with MS help without it being unable to be a cross-platform solution, so I don't see your argument there.
The main issue is making use of "spare" hardware capability.
There are two main barriers to using physics as an actual more complex gameplay tool (outside the level of interaction we currently have), and it has nothing to do with vendor lock-in.
Already the idea of cross platform stuff has been mentioned, and it's a key sticking point for accelerated physics use in gameplay elements, since even if you can make a PC game which works on ATI and NV hardware on the PC, if you use the hardware to make the physics different, what are you going to do if you want to make it a cross platform game? You just can't do the same thing on a console that you can do on a PC, because the hardware (GPU) won't let you, and the processing power might not/will not even be there, let alone be available for you to use.
There is a limiting factor to adoption of more detailed physics driving gameplay elements and IMO it's not vendor lock in, it's more a console issue and...
The other part is PC support. So you can run all this stuff on NV and ATI hardware. So what? Most of your potential customers don't have hardware fast enough for it to be a comfortable graphics+physics experience, so they can't really make use of it.
The only way you can start to tap unused hardware power would be, IMO, at the expense of graphical power until we are far enough into the future that either the available power is immense, and the lowest denominator is high, or you take a different design philosophy and reconsider how much you are going to do graphics wise.
An 8800GT struggles in Mirror's Edge when you push the graphics, and that's a game which only uses PhysX for fluff, not anything game changing. If you want to try and make a game based around fancy physics and maybe also add graphical fluff with it, you're going to demand even more power, and we're not there yet on PC.
PhysX is not particularly compelling, but that's not the fault of developers or NV per-se, it's just the nature of the beast. No one person can make it compelling, and it's difficult to force the issue, it's almost one of those things that will have its time eventually.
Maybe it's a tenuous comparison, but I would liken it somewhat to tablet PCs. They have been around for years, but they have been mostly expensive, limited and not exactly having a mass market appeal. With development of technology they have become more affordable, more useful, and have more appeal. But that's taken time because the technology has needed to come together, costs drop, other things like Wi-Fi have become mainstream, the internet has taken off. Such things have all helped tablets be the upcoming thing, even though tablets have been around for ages.
Maybe when everything comes together for hardware accelerated physics we will see it hit the mainstream, but it might take time.
Facebook
Twitter