City of Villains To Support PPU

PhysX is built into Unreal Engine 3.0; unreal engine is one of, if not the most licensed out engine

Aegia is giving away the SDK for free (you can dl it off their site) while Havok charges hundreds of thousands of dollars per game

Aegia already has BFG and ASUS in the bag, and these are heavyweights in the industry with excellent reputations, not just some shcmucks

The intial cards are supposed to retail at $250, but as video, sound, network cards and CPUs show, technology becomes cheaper after it's adopted. If PPUs take off, there is no doubt that cheaper solutions will become available.


I'm all for this initiative, and I'll be one of the early adopters for sure.

someone tell me what to think about this. i'm too unitiated to research it myself.

honestly, I don't want to spend an extra $50 - $100 on something like this.

Video cards do specifically video card stuff.

They do polygon transformations, texturing, lighting, and now they do those same things on pixels. (Shaders).

Their architecture was designed to do those kinds of things very fast. The state of video game graphical detail, (# of polygons on scene, texture resolution, all the new shader stuff) actually outpaced the video chip design. So the cards grew very quickly. Now the 'video subsystem' (chip, memory, cooler, etc) is the most expensive part of a PC. Even more (ironically) than the mobo, cpu, RAM subsystem.

I got off track.

My point, and only The Carmack knows (kidding), is whether physics has that same level of separation. Can you so effectively separate world physics from game logic as you can the rendering? Are there simple definable functions (like matrix multiplication) that you do over and over with physics calculations to the point where designing a hardware solution makes sense?

I guess my thought is that now that processors are going multi-core, wouldn't it be simpler and just as effective to simply put the physics on a seperate core/CPU and have the benefit of residing in the main system RAM, and being close to the game logic?

I'm not a physics programmer, but it seems to me it's closer to general purpose calculations than it is to rendering

And if it's not...then I have to vote for Nvidia/Havok. You'd be doing lots of things with surface and vertex collisions, and the GPU is best at that kind of work.

Of course, there's also weight/mass, velocity, viscosity, friction. All of those things are completely unknown to a GPU.

All in all, it's good for the industry.

so now it looks like havok/nvidia are competing over accelerated phys with AGEIA? on the one hand that could make things very messy... on the other competition tends to spur things forward very fast.

Actually the GPU physics solution could be more expensive and less powerful/effective. SLI required for good physics performance adds quite a lot of expenses to the equation.

im all for accelerated physics.. not for 300 bucks tho..