Flashpoint 2 And The PPU

[kegetys 2]

The commercial license cost of the Novodex SDK is said to be 50 000 dollars for one title on one platform, which is quite expensive considering the SDK itself is nothing that special compared to the other choices. For BIS that money would propably be better used to hire a extra coder or an artist to work on more important things...

Quote[/b] ]Speaking of SDK's though, BIS has been using a licensed SDK from Immersion for some functionality in OFP and VBS, although I'm not sure that they take full advantage of all the libs offered in that SDK. M$'s Dependency Walker doesn't show a whole lot of hooks going into the ifc22.dll.

The IFC22.dll is the Immersion Foundation Classes library, and as far as I know it only does force feedback effects.

[SpecOp9 0]

I think BIS will have no problem with their own physics engine in this game.  They do everything on their own from scratch.

This PhysX engine seems like it's more for games like Halo, Half Life, and other shooters.  Using this for Game2 would be pointless, knowing Game2's architecture is going to be more different than any other games we have seen.

It would be like making a Super Mario engine where it looks pretty when you collect a star, and using that engine for Doom 4.  Just doesn't seem right.

Game2 should have its own coded game, with its own coded physics, with it's own everything. And, that's the way Game2 will likely turn out in the end.

[Pyronick 21]

I think BIS will have no problem with their own physics engine in this game.  They do everything on their own from scratch.

This PhysX engine seems like it's more for games like Halo, Half Life, and other shooters.  Using this for Game2 would be pointless, knowing Game2's architecture is going to be more different than any other games we have seen.

It would be like making a Super Mario engine where it looks pretty when you collect a star, and using that engine for Doom 4.  Just doesn't seem right.

Game2 should have its own coded game, with its own coded physics, with it's own everything.  And, that's the way Game2 will likely turn out in the end.

So basically you want to have 20-30 objects real-time in a huge island of 25kmË› with over 400.000 objects by itself.

That means lag, unless the physics would be as crappy as Op. Flashpoint...

Which means loosing potential buyers.

I think BIS would have to make their physics engine compatible with the PhysX PPU.

I mean not using fancy NovodeX features just like in Meqon or Havok ordinary rigid body simulations.

But not like in Flashpoint.

Your comparison isn't right, look at Age of Empires 3 and Company of Heroes, totally different games also Havok physics powered and look what the effect is!

Totally different (in a positive manner) compared to the ancient gameplay of Age of Empires 2.

And PhysX isn't a engine but a processing unit next time look what you are facing

And coded physics? :S do you know what a physics engine is and how to make one?

You can make a physics engine but coding it?

[shinRaiden 0]

Found this today. John Carmack's take on the PPU at Quakecon.

Physics acceleration's prospects

Carmack said he considers the prospects for the upcoming physics acceleration chip on the PC iffy, because physics presents a very fundamental problem that graphics doesn't have: it isn't easily scalable for level of detail. Either an object in the game is a true physics object with which other objects can interact, or it isn't. Carmack predicted this constraint would lead to a number of physics-accelerated titles where acceleration affects only elements, such as flowing water, that are peripheral to the core gameplay experience.

Another of his concerns about physics acceleration was speed, not in the sense of peak processing throughput but in terms of the immediacy of real-time interactions. Carmack recalled that the first pre-3dfx graphics chips made Quake feel slower due to lag between user input and visual output. He worried that the first generation of physics chips could cause similar problems, leaving us with games that are fragile and slow. He also happily conceded that he is more of a graphics guy than a physics guy, and admitted that his worries about the bar being raised in physics are probably similar to the worries other have had about his own standard-setting work in graphics.

Well, I hate to be the one to throw pointy barbs at John Carmack of all people, but here goes.

First off, consider the analogies to graphics systems design. CG fixed on triangular polygons eons ago. No matter how 'smooth' something maybe, it's still triangles. Even nurbs, which are merely another process to ultimately arrive at triangular polys. With the understanding that virtual worlds are made up of polys instead of other surfaces in RL, that changes the whole mindset of systems programming.

Look at really old games. There's the very basic triangular constructs. From those elemental levels, everything now is just padding more triangles onto the base. The key point though is you're starting from the minimal LOD, and stepping up to the maximum LOD.

Same thing goes for vegetation. The most complex systems modeled on L-System mathematical constructs merely consist of polygonal and vector subdivision from basic to complex.

Physics on the other hand is still thought through top-down. Think of a RL dust cloud, with each particle of dust as a seperate indexed particle. How do you determine which to cull, how to cull, and how to even process something at that scale? There's your problem, you're taking it backwards.

This is not supercomputing, this is gaming. As such, not only can we look at things differently, imho to some extent we should.

Instead, what I think should be done is simple expand existing models. For example, moving bodies have two vectors, inertial and moving. Those are fairly simple to model vectors, that you can then make more complicated in terms of time and space domain LOD'ing.

Fluid physics such as particles, air, and fluids can be done similarly. Simply define some basic models - like a mass with various types of simple vectors, then set parameters to control the scaling. Do you need to do full particle modeling on every water drop in a river? Absolutely not. Make the river as a model, then decide how many water drops you are willing to support and in what circumstances.

Now speed is a different subject. Using the traditional model of input -> physics -> visualization, you can never crunch physics fast enough because it is inherently late in that model. This is why physics modeling is much better suited to dual-core or multi-sub-core systems like Cell. A drop-in card with a chip like Aegia's would help, but syncronization could be tricky.

The point is that physics has to be done and calculated before the graphics can be accurately rendered, and the rendering and input need to be concurrent to be realistic and not laggy. So in that sense, there needs to be some banging of heads against the wall to figure out how and where to crunch the physics to have the results all ready to go for input and visualization.

BIS's tendency has been to do everything in-house, including graphics and physics and netcode and what-not. I suppose that would preclude them licensing some external libs, and tbh, integrating external libs of that sort can be as much pain as it is a help.

HOWEVER, massive physics is still in very infant stages in gaming. If we immediately demand physics performance levels comparable to the current state of PC 3D graphics - which have taken 20 years and countless billions of dollars of development and sales - that's shaking the baby, and we'd get a lot worse in the process.

[quiet_man 8]

Found this today. John Carmack's take on the PPU at Quakecon.

...

- that's shaking the baby, and we'd get a lot worse in the process.

agreed on this, simulating the destructing of a few objects within an optimized environment is one thing, maybe suited to some FPS or wathever games. A full scale tank/Infantry battle in a city is something different and even if the PPU could handle it, graphics and gamelogic would be overloaded by change of the battlefield.

The destruction modeling the demos show is mostly eye candy. IMHO improved simulation of objects that are already single graphical/gameplay instances would be more helpfull for OPF style games.

e.g. how does the PPU support simulating a tank suspension or bullet flight? Calculation of correct bullet flight with penetration and deflection at surfaces or simulating the effect of a single tank wheel destroyed would improve the simulation without overloading other parts. IF the PPU supports this better than a multicore cpu and people are willing to pay the price.

on part of eyecandy it might be usefull to integrate PPU functions at the graphic cards!?

quiet_man

[Scrub 0]

Sorry, I put this in NGG2 Suggestions - game physics, but it seems the actual discussion was here.

I hope this isn't pushing a dead horse, but I thought you all might want to see these two pieces of info.

BFG to retail a range of AGEIA boards

AGEIA acquires Meqon

(I'll delete the origional post)

[Tigershark_BAS 0]

Shin and quiet man. Enjoyed reading your posts!

Nice to have great discussion on such topics

[blackjack-VS- 0]

couple of days ago i got an email from a teacher recruting plp with interest in 3dsmax,photoshop,3dsound,c++,and other mechanical/technical specs.so today i was invited to a kind of project briefing.i'v been 1h watching and listening a master degree student presenting both Novodex and Ogre 3dengine. i was amazed with both, and just cant wait to start working with that team. after reading in Ogre website (free open source 3d engine btw) i presume that is out of question for a game like ofp... but for other 3d aplications it's a blast!

about novodex engine,it is something really ahead.. i really want to test a game(dont care what style) that uses this kind of technology.its the 1st time i see this kind of software, so i went nuts with it!

for those who want to see what i'm talking about just go to both websites and download the small files with demos. not videos! u really intereact and able to do a lot of stuffs.

if this is oldnews, i'm really sory...

ogre 3d engine

Novodex rocket 1.1

enjoy it!