Posted on 08/08/2006 11:43:45 AM PDT by Sofa King
Physics has always been part of our games, as the term describes how objects relate to each other and their surroundings. But to render physics in a realistic way, a large number of extremely advanced calculations is required. This demands serious horsepower, which puts more and more strain on the CPU. And as the CPU already handles a lot of other tasks, the idea of a separate hardware unit to take care of physics - just as the GPU handles graphics - has been around for quite a while. Based on that thinking, Ageia created and released their PhysX card. Nvidia and ATI, though, have a different take on the problem, saying that physics can be done just as well, if not better, by adding another GPU to the equation rather than a separate physics processing unit (PPU).
So it seems that opinions differ concerning which technique handles physics best. That's why we here at THG had a talk with a couple of scientists, in an attempt to try to shed some new light on this subject. Is physics really that hard to do? Why? And which technology works best?
(Excerpt) Read more at tomshardware.com ...
Personally, I think an "AI Card" might have been the way to go. Seperate game AI from the rest of the calculations and that would have given developers the luxury of making far more intelligent and involving games.
OTOH, you do realize that if you have a physics processor in a machine, that some basic rules of that physics processor will apply to all games. Unfortunately, this may mean that all games will have a certain basic feel to them. Then again, maybe not. It's hard to say since we don't yet know exactly what tasks the physics processor will take over from the CPU.
IMO, as I said on [H], this is a silly article.
The only thing it demonstrates is why engineers build things and not scientists (speaking in the pure sense of each profession). Further the author and his sources seem to have a knowledge of gaming that advances no further than the shooter (FPS) genre.
These scientists seem to think no card will allow truly accurate real-time physics, and they're right; but neither does any CPU, past or present.
Nonetheless, ever since Sublogic Flight Simulator for Apple ][, people have been building and enjoying game simulations of real physics.
What changes is that as we've gone from a 1Mhz 6502 to 2+Ghz machines, the level of accuracy that we're able to enjoy increases and the number of shortcuts or kludges decreases.
SubLogic FS 1.0 featured no rendered plane, a 6x6 mile wireframe 'world,' ridiculously simple physics, and, well, the framerate was maybe 5-10 fps.
In 1990 we had a colored world, sprite-driven cockpits, approximate but still table-driven physics, and multiplayer dogfighting in Kensai's Air Warrior.
Now we have lush 3d worlds and cockpits, 6 degree-of-freedom physics emulating the control surfaces and effects of engine and radiator controls, accurate ballistics and online campaigns in Ubisoft's IL-2 (and it's still not perfect).
A dedicated physics processor is not a cure-all, and I'm not certain it's the best approach (for reasons having to do with flexibility as well as timely communication between CPU, GPU, and physics processors). However, the argument that because a physics chip is not perfect, it's useless, is a red herring at best.
Gaming has always been about doing as much as you can within the constraints of a playable timing cycle, which is why it's an engineer's playground and a scientist's nightmare.
The scientist interviewed in the article is exactly right. These physics cards are essentially worthless. The problem that he mentioned is that the physics algorithms are built into the hardware and that they cannot be changed. I have to agree with him and say that this makes the card almost worthless.
Sure, these cards may take some load off of the CPU and they may help developers get around the issue of having to cope with the usage of difficult formulae and writing their own algorithms, but for those of us who work in the fields of physics and numerical analysis, the nature of the card is just unacceptable.
Maybe these cards would be fine for simple Newtonian physics used in games ... who knows. But games are not physics simulations. In a game, things just need to appear correct, but they don't necessarily have to be correct. In a physics simulation, things must be correct, or else your results will be wrong.
Don't count on me buying one of these cards anytime soon. I'd rather use my own physics code and numerical methods like Runge-Kutta (of any order that I want) to do calculations than have some preprogrammed unextendible junk code locked into a $300 physics card.
Digital is for people that can't handle reality. ;-)
Off-topic, here's joke I heard that I thought was amusing:
A physicist, an engineer, and mathematician were all staying in a hotel.
The physicist walks into his room, and sees that there is a fire on his bed. He walks up to the fire, takes some measurements, and does some calculations. He then walks into the bathroom, fills up the garbage can with the exact amount of water the he calculated would be needed, and pours it on the fire, with the last drop of water just extinguishing the fire.
The engineer walks into his room, and sees a fire on his bed too. The walks into the bathroom, fills the garbage can with water, says “this looks about right”, and dumps it on the bed, extinguishing the fire.
The mathematician walks into his room, and sees the same thing. He thinks to himself for a moment. The walks into the bathroom, turns on the sink, and pulls out a match. He strikes the match and puts it under the sing, and sees it extinguished. “Ah,” he says, “a solution exists.”
Video tech geek overload ping!
I can handle reality. It just doesn't have enough zombies for my tastes.
The physics processing in the Half-Life 2 engine is awesome.
"Maybe these cards would be fine for simple Newtonian physics used in games ... who knows. But games are not physics simulations. In a game, things just need to appear correct, but they don't necessarily have to be correct."
Well, that's what they're being primarily developed for. Nobody would have dreamed of trying to use one of the early graphics cards for the type of CGI used in movies. A CPU will always be able to do things more exactly and with more flexibility as long as time isn't critical.
Imagine what a physics coprocessor could do for Solitaire?!?!? Man, I soil myself just thinking of it... :-) Actually, BF2 would run much smoother I think...
"Maybe these cards would be fine for simple Newtonian physics used in games ... who knows. But games are not physics simulations."
Apparently you don't indulge in simulation gaming. I've been involved in long, heated, angry conversations about, for instance, the proper coefficient of drag of German 20mm cannon rounds (of various types) and whether the resultant kinetic energy was inflicting the proper results (and/or causing the cannon shell to fuse properly) at various ranges and/or angles of impact to target aircraft.
As I said above, the authors of the Tom's article (and you as well) appear conversant mainly with what I would generally term "light" gaming.
Personally I don't know what these cards can and cannot do exactly, but I do know that with DX10 on the horizon processing is becoming more and more CPU-aspecific. Between the CPU (or CPUs for dual core users), the GPU, and possibly this sort of physics chip, we're not so much running *a* computer any longer but really a localized computing cluster.
A B.C. comic from years ago:
Picture one poor schmuck with his head in the guillotine, up on a platform in the public square.
King: "Executioner, cut the rope!"
Executioner swings axe down on rope, cutting it.
Nothing happens. There's an awkward silence.
Priest: "It's a miracle!"
King: "In that case, free the prisoner."
Poor schmuck twists his head around and looks upward. "Uh, it's just a rock caught in the release mechanism."
Jester: "Once an engineer, always an engineer."
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.