I think that any way to possibly get as real of a game as possible is worth the money, especially if it even frees up CPU time and power. Though I doubt BF2 will support it, as the game looks close to completion and they would have to probably hard-code the physics into the game code for it to read properly and actually handle the game's physics as opposed to the CPU.
Solo I totally agree with you. DirectX improved games tremendously and standerdized everything, if this could be integrated into DirectX (WGF by then hehe) I can see a very bright future for it. Problem is that AGEIA would have to open up their technology and seeing it integrated (into WGF) means other companies can come along and build compatible chips that might cause AGEIA to lose before the real battle has started... Don't forget that only AGEIA's proprietary physics engine NovodeX will support the PPU at first and as far as I know it isn't used in a lot of engines (or at least not in major titles), unlike Havok... Now if everything does come of the ground (wich surely I hope it will): At first this will 'only' be for visual improvement, game developers can't risk using it for gameplay purposes because you always need to make sure people without a PPU can do the exact same things (imagine blowing up something that causes a chainreaction of events, leading up to a twist in the story but you just can't pull that of without a PPU...). But still even if we only get visual improvements for the next 2 years, boy, if you read that pdf and even if they can only make 1/10th of that true...*drools*: clothing, hair, real fluids, dynamic particle system that makes moving fogbanks and fog that reacts to a player walking inside of it... *jumps around with joy :D * Here are some demo-pics: http://www.pcper.com/article.php?aid=140&type=expert Don't forget these guys aren't into graphics, they're just showing the physics part, imagine that water but much more realistic... AND they say that it aint even running at full power yet (bugs) That landslope had about 4000 objects in it, they claim they can reach 32000... But it will take time, sadly.
(first: because this'll be a big one again I'm posting it in a separate post instead of editing) Hmm having had a day to think about it I think it's very possible this might see adoption in multiplayer games as soon as it gets adopted in single player. Why ? What about the extra info that needs to be send over the net that's probably too much ? Actually no. Seeing that the world would be dynamic, ie. the way a box reacts to a bullet is different every time, the calculations are still based on known variables like bullet speed, angle of attack, materials, windspeed, ... so you 'only' need to send those over to a player, the result would be the same on everyones machine if you start with the same input right ? So if you can always calculate the position of the objects that move in a predictable manner (no players aren't predictable ) you only need to know their starting position (and coincidently that's the same on everyones computer as it's known when the map loads ). The only information you would need are of the things that do change randomly in the world: like players, vehicles, bullets, whatever is not controlled in a cyclic (scripted stuff basically) or computer controlled manner. And whenever you receive the information of the creation of a bullet (shot fired) you know it's going to hit that crate, where, what the box is made of,... and your pc can calculate the exact reaction of the box *TADA* :D Only problem is Battlefield engines don't work that way... As far as I know all the collision detection is done on the server, not the client. That's the reason why people most of the time complain about the 'crappy netcode'. It actually isn't the netcode, but just the server that either can't keep up with all the calculations needing to be done, or insufficient bandwith to send all the information. Though I could see that they could keep player/vehicle detection and monitoring on the server (good against cheating), and the other objects on the clients :lookaround: but I don't know enough about the low-level workings to know if that'd be safe or doable at all (and this whole thing here too for that matter hehe) Maybe Rad can answer this better than anyone else (if I'm not mistaken he was a dev at Dynamix !?!) Anyway, only time will tell I guess and that sucks
You are right Mac, but it requires everyone on the server to have a PPU. It would be like the grass settings in BFV in a way. Of course these days everyone has a 3D accelerator of some sort, but in their infancy it was something you get extra - lots of folks bought Voodoos in order to play GLQuake @ twice the res but the same framerate + some nice filtering, everyone else could continue to play on the same servers with the same experience.
in order for PPU's to make an impact, we want to see BF with buildings that can be ran through by tanks and demolish the walls appropriately, have planes come to pieces like they should and infantry get appropriate injuries (think 50cals removing limbs and so on). But theres no way you could do that in software for non PPU's for some years to come :(
Don't get me wrong - i do want to see the PPU really really exceed! Its gonna have PS3 level physics processing, but the battle for its existance on the market place will be in the single player market not MP.
I hate computer-technology-evolving...it goes too fast...:(