PhysX card compatability

General discussion about the development of the open source Blender

Moderators: jesterKing, stiv

Post Reply
sephiroth111
Posts: 0
Joined: Thu Jan 01, 1970 1:00 am

PhysX card compatability

Post by sephiroth111 » Mon Apr 03, 2006 3:03 am

Do you think it would be possible to intergrate PhysX chipset support into blender's game engine? If the SDK works the way i think the SDK works, it should be easier to make realistic looking physics.

information on the physX card can be found at their website, located here: http://www.ageia.com/

its obviously very new technology so i dont expect it to be intergrated into any new computers for awhile, so its ok to leave it for a bit, but i'm interested to see if blender can use this functionality to improve performance and make use of its cloth and texturing abilities.

kakapo
Posts: 0
Joined: Sat Sep 04, 2004 2:32 pm

Post by kakapo » Mon Apr 03, 2006 12:56 pm

physx is closed technology and not compatible with blender's gpl license.

i also think they will have a very hard time selling their ppu add-on cards. look at what nvidia, havok and ati are doing. high end gpus actually are more powerful vector processors than ageia's ppu.

http://www.pcper.com/article.php?aid=228&type=expert

sephiroth111
Posts: 0
Joined: Thu Jan 01, 1970 1:00 am

Post by sephiroth111 » Tue Apr 04, 2006 1:54 am

ah. well, I suppose the newer GPUs would be easier to work with, since they are just add on chips. but honestly having a seperate chip for physics alone and then on top of that adding the GPU's effects. i can only imagine what that'd be like. though they have a good idea with the Havok FX, don't you think that it would drain too much from the card for it to be good?

oh, anyone got an idea on why I joined in 1970? I wasn't even alive yet, let alone blender...

z3r0_d
Posts: 289
Joined: Wed Oct 16, 2002 2:38 am
Contact:

Post by z3r0_d » Tue Apr 04, 2006 2:59 am

sephiroth111 wrote:don't you think that it would drain too much from the card for it to be good?
with more complicated shaders you become fill rate bound, not transform bound.

however, newer cards will be running both on the same units... but "fill rate bound" would then mean "memory bandwith bound"...


I wouldn't expect using a graphics card for physics to be helpful to simulate things that require large amounts of constraint solving. particle systems where each particle doesn't depend on the last are well suited for graphics cards [and their drivers which buffer commands perhaps for several frames]... but cloth simulation, fluid simulation, or rigid bodies that can interact [stack] would be very difficult to accelerate with such a lag between when making a request and getting the results.

pildanovak
Posts: 18
Joined: Fri Oct 25, 2002 9:32 am
Contact:

Post by pildanovak » Tue Apr 04, 2006 6:36 pm

i think i've seen on some forum that some parts of bullet are written already with these chips in mind...?

Post Reply