PhsyX add in card

Joined
Mar 2, 2005
Messages
50
Well, the dual core chips will be shipping soon, XP 64 bit OS is out, multi threading is starting to take off and graphic cards are just getting better so computational power should really be ramping up. Unfortunately, it must not be enough power because the PhysX add in cards will be on shelves by years end. As I understand it, PhysX is a physics processing unit (PPU) that uses NovodeX as its implementation. I know nothing of NovodeX, other than it will be used in Unreal 3.0 engine; never the less, I am wondering why NovodeX can’t be optimized for 64 bit OS, multi thread dual core or whatever.

Will you buy one when they come out, will you take a wait and see attitude, what will it take for you to consider it a must have piece of hardware?

I guess if I need one to fully enjoy the next generation of Unreal games, then it’s a given that I will buy it, but its starting to smell like the old days again… 3D daughter cards, math co-processors, yada yada yada and a ham sandwich. You know if this thing takes off the consoles will get them at some point and it won’t change the target MSRP much but the same part in the form of a add in card will end up costing PC gamers 1/3 to 1/2 the cost of a console unit. Feh!
 
Novodex doesn't require a Physx PPU, it just has the capability of using the the PPU to hardware accelerate physics operations.

Remember that until relatively recently many games could run in "software mode" without strictly needing a 3dcard at all.

Also, Ageia (who own novodex and the physx tech) have said that they are basically going to make the API open or licensable so you don't have to be using the novodex suite to use the physx PPU, i.e, Havok and Karma (and any others) will be able to utilise the PPU aswell (I'm kind of interested to see how that works out...).

In any event, we could be benefiting from PPUs well before we benefit from multi-core cpus or even 64-bit (6months vs 18).
 
With the results that we've seen in the use of dedicated processing to perform specific purposes, it sounds like it'd still be faster than a properly optimized and coded multi-threaded game on a dual-core proc. Just like dual-core isn't going to make us trade in our GPU's for VGA cards.

Personally any dedicated tech that yields impressive results will get my vote - with my wallet. I will be waiting to see the results though before completely drawing the conclusion that it's a must have.
 
Back
Top