"I hope nVidia didnt pay alot for Ageia"

I think it's a knock on PhysX.

Although he should have said "Why do Nvidia even bother with CUDA and PhysX, they'll have gone bust within a couple of years."

He is right about ageia though and their approach.

Game devs want something that works without hassle, physx is not it.

Nah, nvidia saw a marketing opportunity. It already knew from it and ATI's experience with havocfx and folding at home that the gpu's could do more and people were excited so the opportunity to control such a tool and the marketing potential was too much to resist.

Of course when you strip it all back it hasn't come to much bar smoke and mirrors although overall when it is eventually ditched it won't have done them too much harm to flaunt themselves as "pioneers" of gpu physics.

I find PhysX belh really. Its not really going to be able to just bust in and take Havok out of the game. Havok has been doing it a long time and has their name in almost every game for its great realistic physics.

Of course a hardware approach to it using a GPU might be cool to see if it can do it while doing rendering, as long as the games performance does not go down.

Intel has Havok now which means it may be pointing towards CPUs or possibly LRB. We will see later.

I doubt Rage will have it

I agree, they definately need to integrate it better, starting with allowing it on ATI and future Intel cards, right theres theyll have (after Intel and LRB) a 50% better integration into the market
To me, falling short of this, theyre killing off physx

To me, its walking the fine line alot of us do, turning down the eyecandy for better gameplay.
Eyecandy is somewhat immersive, but a overall better written game is still priority No.1
If anyone can put the eyecandy together with actual gameplay, its no brain winner, and so far, that hasnt come close to happening, and it never will, since nVidias approach is only towards 1/2 the available market

I agree that nvidia, ATI and Intel need to agree on how they handle hardware physics. However there really needs to be a bridge between the two, Just like the first days of 3D acceleration.

What really needs to happen IMO is a single API that can call on either the cpu or the gpu to do physics work. This way game developers can feel comfortable coding physics. As physics in game grows, GPU physics will become the norm and CPU solutions will be phased out.