@hiph
because Physx is such an important factor when buying gfx cards..... truth of the matter is there are times when you cant turn the industry by yourself, how many physx tittles are there out there? maybe 10%, developers who adopt physx risk narrowing the available audience for their titles, if nVidia do not subsidizes these developers then there would be no physx titles period, at this rate physx will just disappear into obscurity, and all that money that nVidia used to buy and polish the tech would be for nothing anyways
if AMD were to push their own standard then no one wins, the consumer would just wait it out and developers will not divide their budget just to support two standards, that would be the death of hardware accelerated physics engine there and then
im not saying nVidia should just hand over there stuff to AMD, what im saying is nVidia can simply realign their technology with an open standard like OpenCL, yes they will not be able to capitalized on their expenditure on acquiring the tech, maybe then they can actually make an competitive physx card instead of cobbling an ancient implementation to their gfx cards, if the hype is to be believed their cards should excel at this, there should not be that huge a difference between CUDA and OpenCL execution
i would say that more then anything, nVidia would not adopt an open standard because they fear competition, be a crying shame if an AMD card would run CUDA better then a nVIdia card