Bringing Nvidia technology to AMD hardware.

Well it levels the playing field a little bit. Neither AMD/Intel or Nvidia can can really make a tempting offer to game developers. I think AMD and Nvidia physics ready hardware market penetration must add up to somewhere around 20%? Not a nice number to have. We gotta wait till physX hardware hits the mainsteam. We need Radeon 9600's and Geforce 6600GT's that are really physX capible before this goes anywhere. And anyways, neither Havok or PhysX is what I would call a physics engine. There both anti-bumping-together engines. I want to see a real physics engine, with a fluid environment and everything. Tension forces, compression forces, maximum pressures, heat, kenetic sound and electrical energy... Something where every effect, or at least the ballistic ones, are taken into acount. Intel and AMD on the other hand.... hard to argue what processors support havok and which ones do not, so 99% of PC's out there support it, but of course, its a whole lot slower/less powerful than PhysX. Source impressed me, it did, but five years later and we don't have anything thats a whole lot more impressive.

They should try to encorporate Unreal Engine 3 into the hardware (make it universal), like XBOX 360's native physics processor, that is built right in there (I'm not saying UE3 is built in, but the 360 seamlessly handles Havok and UE3 and any other engine at the hardware level). And can run pretty much any game with crazy physics and graphics, yet is way under powered to the average gaming PC. It can because Microsoft made sure that whoever makes games for the XBOX, that they make games that are 100% compatible with the hardware, and they rely on crappy drivers to make it work. The 2 companies (AMD and Nvidia) should come to an agreement and make hardware that uses open physics processing, instead trying force one or the other on each other.

I think that PhysX is overrated, sure it can do pretty particle effects and cloth physics, but Havok is where it's at for interactive physics.

That said, GeForce 6600's and Radeon 9600's will never support programmable hardware physics. Also, among Steam gamers, there are more 8800 series owners than GeForce 6600 and Radeon 9600 series owners combined:

http://www.steampowered.com/status/survey.html

I like looking at the drawing more then this article on physX.

@Sunny
Well that's just great. I suck at writing but I'm great at doodling crappy pictures in MS Paint. So much for being a respected member of the press... :(

I'm just kidding. I spent like 5 times as much time on the picture than on the story. I even focus tested my first pic, gathered feedback then improved it.

Naw man nothing bad towards you just not really interested in the topic.

The drawing just sums up what you mean in the article.

Awesome picture.

Anyway, I think it's a good thing. Whether Havok or PhysX or not-yet-invented is better is something I don't know. But PhysX is here and I'd like to see it supported across the gaming market so that it can start being used. That means better game effects for us sooner, and more importantly avoids a standards war (see Blu-Ray and HD-DVD) that takes years to decide. If something obviously superior comes along later, we can always upgrade. Transitions (e.g. IDE to SATA) are easier than format wars.

I know I used hardware examples to a software problem, but I think that the principle holds. Maybe I'm wrong. Just PhysX seems pretty stinkin' good and it's here and it works, so let's get the supported games coming!

Great picture .

I believe that they should simply support all more widespread physics engines on the cards... but that's just my idea and I don't really know how possible it is .

I think whoever makes the 'standard' physics API (if such a thing ever does happen), it should be from a third party. See, if they use PhysX that puts ATi at a disadvantage in the market; not good for anyone as it puts nVidia (the rich execs up top in nVidia, not the common nVidia fanboy; you get ripped off big time) in a dominant position which means they can choke the market; ultimately that raises prices for consumers on all graphics cards that can handle GPGPU physics well.

That would mean that less people would be able to afford those cards, less developers would be likely to take up the API and the graphics market as a whole could potentially stagnate like it did during G80's market superiority.

What I would really want to know is I wonder if the PhysX from nvidia works with all the titles that worked with the physX by ageia.I have a few titles that took advantage of that physX card so I wonder if the new nvidia card will just take its place. Or will I still have to keep that card? Thats just what I am wondering about regarding physX.

Apparently it does, but the real question is whether or not the GPU doing physics as well as rendering graphics really brings any performance boost over traditional CPU physics - or whether it actually hurts performance because of poor load balancing.

@lightnix My question as well. However, if we are talking about physics processed with stream shaders it will be interesting to see how ATI fares with nVidia code. I would imagine that they would be about on par. The code being more efficient for the nVidia GPU, but the raw shader power of the ATI making up the difference in efficiency. Also, I wonder what Intel has up their sleve as they do own Havok. Perhaps the next Intel CPU architechture will be set up for physics processing. And don't forget about Intels foray into the dedicated graphics market. They have the ability to bring us the most power efficient GPUs ever seen by simply combining the fine grain power management of the Core 2 architechture with a 45nm high-k fabrication process. A good graphics engine and we could have mobile GPUs performing on par with their desktop counterparts. And with Intels Atom processor we may see the birth of the first ultra portable gaming PC (think Asus eePC with an equivalent of an 8600GT and a Core2 Duo E4300).

Awesome times ahead, can't wait to see what the future brings...