NVIDIA has stated they've no interest in making discrete PPU's at this time. There will still be drivers available for those who did buy the PhysX PPU, but for the moment, NV is more intent on integrating the ability to process PhysX API calculations into video cards, much like they'd initially planned to do with HavoK before intel purchased them (HavoK, not NV).
don't the 8800 series already use this it says on my 8800gt box :?IQT786
It was planned that geForce 8 cards should be able to be used for physics acceleration. Problem is, HavoK backed out, because their physics implementation simply could not cover the sheer breadth and scope that AGEIA's PhysX API was able to. but by the time HavoK backed out, NV was already committing the design to silicon.
maybe the nvidia wants to take a advantage form atia again. with nvidia physx and intel havok the ati will have a "good" physic solution to.Sistem_42
It would be in NVIDIA's best interest to either license or collaborate with ATi to be able to share PhysX GPU acceleration, because it then gives them both a competitive advantage to intel. Right now, intel's next-gen CPU and NVIDIA GPU could then accelerate both HavoK and PhysX, and if ATi/AMD doesn't join up, it won't have either. But if DAAMiT joins with NVIDIA, suddenly there are a lot more people that will be able to have hardware acceleration for PhysX, and make intel's decision to purchase HavoK look all the more stupid. And such a move would also help a lot to build the friendly alliance that AMD and NVIDIA shared before they (AMD) purchased ATi.
yes that makes sense, but still there is the nvidia ment to be played sure. you know what i mean right.
in one hand if they make evrythig right they could triumph over both the ati and the intel, but it is a lot of risk and thy wont take it. still it remains to be seen if nvidia means to join with ati or they will join but the nvidia ment to be played way (holding most the apples for themself).
Meh, I thought it was a cool idea to have a seperate card for physics. Spreading the load, keeping it off the cpu, seems like a good plan. Since the GPU gets bogged down by a lot of the current games, I don't know if adding even more to its load will be helpful? /shrug
Meh, I thought it was a cool idea to have a seperate card for physics. Spreading the load, keeping it off the cpu, seems like a good plan. Since the GPU gets bogged down by a lot of the current games, I don't know if adding even more to its load will be helpful? /shrugMunkyman587
True, but i would imagine that NVIDIA will use this primarily to sell more multi-GPU configurations. That way, you could buy one gpu for graphics, and then do SLI with another for physics (and maybe a little more graphics if there is some overhead left)
Log in to comment