Physics companies to take games to next level;_ylt=ArT16C9wNFqfMr5x0HVoi4QjtBAF;_ylu=X3oDMTA2Z2szazkxBHNlYwN0bQ

If all you saw of Agiea’s engine is Cellfactor there’s a bit more to see if you click the link on the bottom of the page and they’re slowly adding stuff.

This article states that physics engines and physics accelerators will revolutionize interactivity in games. No more boxes that won’t budge or break or objects not moving. What you see here is the promise physics will be unparrelleld interactivity in games. Agiea’s link is at the bottom and believe me there’s more then just cellfactor there and there’s also a techdemos page you can find on the site.

EDIT: now if only bullet physics had this, but Erwin may need some assistance.

meh, they’re still rigid bodies

[if anything more physics results in a more difficult to debug game, but I hope we see more destructable outdoor environments instead of invulnerable source engine corridors in the future]

Destructable outdoor environments will come with Crysis, you’d literally be able to mow down the whole jungle. Meanwhile Lucas arts is working on Molecular matter, a technology that will allow metal to dent, wood to splinter, crystal to break ect… in a believable way, expect that in Star wars 2007 and Indiana Jones 2007

Yea right! There are limits like ram. No 3rd party card can change that! Also how many games will you sell if you need a good 3d card and a physics card? You won’t.

There are 3 reasons:

  1. There is no killer app. 3d accel cards had quake. But physics?
  2. unlike grfx, current CPUs are good at physics (and dual core will be great) with the exception of fluids (well they are good at those too, but custom hardware can give benifits)
  3. Nvida and ATI will put it in there cards anyway. This helps in 2 ways. I’m not forced to use there physics lib and i don’t need 2 cards on PCIe buses using all avalible bandwidth.

There are a bunch of other reasons as well, but these are the main ones.

I doubt that “phyisics” cards will gain any decent momentum :slight_smile:


I do support Nvidia and ATI putting physics in their cards, as it would be so much easier. I’d like to have a card that does both graphics and heavy duty physics with fluids, but it’d be way too expensive.

But then again, I think i’ve read before in another forum that few thought 3D cards would take off either and here we are with cards.

A lot of ppl in the know thought 3d cards were just a matter of sooner or later. Why? because everyone already had video cards. That was the only way to plug in a monitor. So it was not a “extra” nor optional. Also there were already 3d accel applications in science, this is where opengl and SGI comes into the picture.

Physics cards have none of these traits. They are not going to be in bundeled systems and your computer works fine without one. Afterall, about the only thing that it will make a difference with is fluids… and that can already be done in 3d cards…

Case in point, even the last gen of games sold slowly, even hardcore gamers upgrade less nowdays. Expect the trend to continue.


I think physics cards are a good idea, however I would like to see them put in consoles rather than in PCs.
The only use for them is for games, and nothing else (at the moment), and as such, you are not going to see them being part of a regular PC setup you can buy, so will have to buy it and install it yourself.
However, most games won’t be utilising this technology in the near future, because hardly anyone will have this card installed, and it would be a waste of time and money for the developers to make games with the physics cards in mind, unless the price of them goes down dramatically, which it won’t, unless the companies behind them are willing to lose lots of money for a long time.

However, if they get put inside the next gen consoles after the Xbox360/PS3 era, (the Wii is not next gen), then it would be good for all games.

These consoles already have hardware thats pretty good for physics. This is not like gfx, normal CPU’s are very good at physics and specilazed hardware does not make a big difference. We not talking Quantum optics here, just old fassion netwon laws of motion.


I think that physics processors could be a good idea. Just as some processors have special instructions to increase calculation speed for certain processes, I think that a physics processor could have similar instructions to speed up physics. It’s not that these instructions are new or anything, it’s just that they’re built in to make the process faster. I’m not sure how these new processors work, though, so I could be wrong.

I also don’t think that the possibilities of a physics processor lie just in the area of gaming. You could also use it for calculating complex physics for movies (like the bake game IPO option in Blender). You might hear sometimes about all the computer testing that’s done on cars with the physics and how they would crumple if they crashed, etc. With a dedicated physics processor, people might be able to design their own cars and test their safety. Imagine thousands of indie car manufacturers!

There’ll probably be physics screensavers, of course; and little things in new operating systems to have fancy physics effects (you delete a file and it shatters into dozens of pieces).

I think that physics processors don’t seem to have as much potential as some other new technologies, but I think we’ll also find new possibilities as time goes on.

I read an article where they tested this card on a finnish computer site. They ran the few games and demos that worked with the card with and without “physics acceleration” and their final conclusion was that the card is waste of money. The games ran at almost same speed with the card and without the card(on a highly powered dual core system). Getting 3% increase in efficiency by bying an expensive physics card and wasting a PCI-e slot isn’t worth the money or the effort atm. Future will tell if this thing dies off or lives on…

Personally I think that we’ll something similiar integrated in the high-end gfx cards in the future. Both ATi and NVIDIA have ran physics calculations on their shader units. :slight_smile: The problem is that the power is currently needed for drawing the GFX.

This is what i’m tring to say. There are no “special” instructions to make physics go faster. We are already very fast. All you need for some very specilized physics is a vector unit. Once you are out of x86 land these are common too. Hell we use PC’s to do physics simulations! Accel hardware will be useless for anything but games cus accuracy will be sacraficed for speed. Just like movies are not rendered on 3d hardware but 100% on the CPU.

3D works cus there are very easy ways to parallise the basic task of drawing a polygon. Thats about all they do but the do it fast (thats not the same as well).


Physics is an interesting area. I’ve been working with both Havok and porting some Ageia to PS3 and currently I’m optimizing Bullet physics for PS3. The PS3 Cell SPU’s are faster then the Ageia PhysX PPU card. GPU is faster then PhysX PPU too, but less ‘generic’ so far (but every generation GPU allows more general purpose algorithms).

For Blender, it’s important to have a multiplatform solution.