Nvidia GTX280/260 and 9800GTX/+ 177.39 Beta Driver with PhysX-Support

[DilTech]

The graphics work in the traditional CPU and physics tests is very light, specifically to keep the application from being unintentionally GPU-bound. So the GPU has lots of spare cycles.

Umm... They're talking about with UT3, now that they figured out how to get it to work properly with GPU physics, and it definitely runs faster than it did with CPU physics.

So there goes the nay-sayers who thought it would absolutely kill frame-rates currently.

p.s. Those still bringing up havok, PhysX has one huge advantage over it. You see, it costs money to license the havok engine(thus, why we haven't seen it used by anyone recently that wasn't on the source engine, and there's only maybe one title coming that does use it), while the novadex api(what PhysX runs off of) is absolutely free.

You guys can keep saying this "if 100% of the market" spiel, but that really isn't how things work. 100% of the market would mean games are playable on old williamettes with TNT2 Ultra's. We know UBISoft has no problem leaving people out, with splinter cell: chaos theory they didn't even add a SM2 path for the x800's for almost 6 months after release while the 6800's had the SM3 path! Besides, using the novadex api doesn't mean leaving anyone out, as it still works on cpu's just fine, and is very well multi-threaded to boot. Especially considering if havok ends up working on ati cards but not nvidia cards, that would be leaving out a much larger section of the market, which these same people claim developers wouldn't do.

I'll end with this...Novadex(physx) is built in to UE3. Think about that, they got put into the most widely used engine so far "this generation" when the PPU was in less than 2% of "the market". To say it won't be used because it presently can only be done on NVidia cards, PPU's, and cpu's is merely being naive, because that now IS 100% of the market. Especially since it's free to use for game developers, which saves them a lot of money in the end.