Can Ageia's PhysX Card Bring Real-World Physics to Games?

pschmid

Distinguished
Dec 7, 2005
333
0
18,780
Ageia says its new physics processing device and engine will do nothing less than revolutionize PC gaming. We gauge just how well Ageia's PPU (physics processing unit) can handle flying debris and shrapnel in Ghost Recon Advanced War Fighter--which you can also see for yourself by downloading our video.
 
as i stated in the 3d gaming notebooks thread ->> http://forumz.tomshardware.com/mobile/modules.php?name=Forums&file=viewtopic&p=192302#192302

its much a case of, whos technique will be most successful, in this "physics war", the main problem, if it persists, will be that there will be X% games sporting physx and X% sporting havok, if nvidia only support havok with sli mode (and ati do the same) then you will need a physx card to get the most of out the physx games and sli/xfire to get the most out havok games, a very costly solution

worse would be if ati go with physx support for xfire, meaning an ati based system would have no havok support

possibly the best outcome would be one of the physics companies going under (not likely havok seeing an already fairly large base will support its physics rendering when the forceware 90 comes out) or a buy out resulting in a merging of technologies

which would come back to the point i made in the gaming laptops thread, with then 3 potential chips in your computer able to do the physics work would you be able to set a primary/secondary chip for physics? if you have a dual core cpu, sli and physx card could you set the physx card as primary physics device with cpu as secondary and extraneous gpu usage as tertiary

perhaps this sort of setup would help prevent some of the slow downs involved by helping prevent the "primary" physics card getting bogged down and using other available resourses to do other work
 
I think the whole Physics fad is really interesting, and may eventually make gaming more fun. How it all shakes out is not really important to me right now.
Here is my issue. "Real-World Physics". This is not what people are looking for , or expect. They are looking for Hollywood action movie physics. Please don't get me wrong, My intention is not to burn anyone- this is just my observation. I have a friend contracting in Iraq. I am reasonably certain that if he gets into a sticky situation, he will be aiming for center of mass, and not paying homage to Atari's Combat with trick shots. At least, I hope not.
That's my 2 cents, Thanks
 
Yeah, PPU is bad idea right now. Got some major problems and only causing the game to slow down. Ageia better get a solution fast before Nivdia or ATI can come up with something better.
 
What I'm failing to understand is how Ageia sees the adoption scenario unfolding in their favor.

Extra CPU cores are becomming the norm very quickly and it's inevitable. Far faster than custom physics hardware which is too new to even call a fad.

A solution that uses extra cores will be incredibly more successful than a solution that requires a $300 card. Devs will not be that interested in pidgeon-holing themselves into a solution that restricts sales to 500 people due to a low saturation of dedicated cards.

This is what will kill Ageia. And it's simply so obvious that it makes my head hurt that anyone thinks that they can succeede on their current path.

Ageia has the mindshare right now, due to a ton of work and money spent in the marketing channels. But their product is too much of an uphill battle.

Frankly, Havoc is the solution most likely to win the battle here, but Ageia is the current darling due to focusing more heavily on the marketing and having the "only PPU" out there.

But if Havoc can get their message out to the audience that they offer the same as Ageia does and, without the $300 card, then Ageia is done for.

Ageia's only chance of success is if they start selling the cards for $50, start getting it added as an on-board chip on new motherboards. Only then will it become widespread enough that devs will lock their games to needing a PPU. Then Ageia can focus on selling for the upgrades to better units.

But even this is a weak stance ultimately, because unlike GPUs, people want their PPU to CHANGE gameplay, not just make it prettier. And the amount of gameplay change attainable is limited to A) the presence of a PPU and B) the capabilities of the PPU present. The likely hood of ANY game taking full advantage of a non-ppu system playing alongside a PPU motherboard solution playing alongside a second generation PPU system is too far fetched. The games will have to play to the lowest common denominator and thus we have the trap that makes PPUs a troubled market before it even really exists.

Ageia has been spending a LOT of money to build it's name and create the market for PPUs so that they can in turn provide the solution for the problem. But it's a false problem. Now and in the future games = online. And Online means you have to accomodate all the players, not just the one with the PPU.

Just because Ageia can build a PPU does not mean that it's the right way to bring physics to gaming. The solution that can succeede is the right solution, not the one with the bigger marketing budget.

Ageia's sales pitch to devs: "Hey, over 1,000 people have Ageia cards now, use our system to revolutionize your game and you might generate as much as $20,000 gross sale. And even more might spend $300 out of the blue to get our card."

Nvidia/Ati sales pitch to the devs: "Trust us, we can talk gamers into paying an extra $500 for a second or third GPU. I know that less than 1% have it now, but trust us, we have big names"

Havoc's sales pitch to devs: "A third of your potential customers have multi-core systems and more buy them every day. Use our system and you can revolutionize your game for the masses and achieve maximum sales."

Now honestly, where do you think success lies?
 
even "holywood" physics are based off "real world" physics

for example, in "holywood" physics if a door is blown off a car it will very unlikely end up standing on its edge (poor example of physx attempt), it may fly further than you would generally expect but its movement, bouncing, spinning, etc would be what you would expect from something that size, shape, weight traveling at that speed, trajectory

change the weight, speeds, shapes, trajectory and the resulting movements should be what is expected as a result of those changes

just from that it sounds complex, but thats nothing compared to the number of calculations required for that sort of movement to take place, especially considering the fact that where the main force impacts that hits it, what part of it hits the ground and bounces, and resulting aerodynamics are all variable and will impact the number and complexity of the calculations that are taking place

going back to your real world example, the physics involved in those situations is fairly immense. a bullet has an arching trajectory, currently as far as i know there are no games that take this into account, bullets just fly straight, i would find this alone very interesting and would make games much more enjoyable as, as in real situations, you would have to take into account wind speed and distance for long distance shots

also how bodies react when collisions occur would be another real world example, in the real world you may not be watching and going "wow that looks so realistic and cool" but in a game it would be just another thing that, when paying attention, would hold the players attention and captivate them

all these "real world" physics examples are the base of "holywood" physics, drop the weight of an item and suddenly it flies twice as far, explosions blow bodies higher into the air and over greater distance, all while running the same set of physics calculations causing the actors to react as you would expect

the idea is to make things happen as you would expect them to happen, not for items to react unnaturally, if its unnatural then generally it sticks out like a sore thumb, if its natural then you wont notice it as much, you will see it, it will do what you expect, and you will be satisfied and forget about it, unless youre paying attention and watching what it does to make sure it does what it should. whether you realise it or not your mind is doing a huge amount of physics calculations itself while watching these games, which is why bad physics is so obvious, your mind calculates one rough outcome, something completely different occurs and its like an error pops up in your mind

one prime example i would say would be the collision calculations in oblivion, whenever you run into an object thats moveable it reacts in an extreme manner, like fly across the room and slam into the wall at a force that makes it bounce about 10 feet back from the wall, even one time a piece of wood spun around on one end for about half a minute before falling over! you notice it because its not what you would expect to happen, you laugh because it looks funny, but in the end its a loss for the physics engine (or the values going into it) because there is an incorrect calculation going on thats causing unnatural effects

i suppose to concisely sum up this rant id say that this should give you some idea why physics is so costly (atleast resource wise!) to implement and how when properly implemented the results are not stunning unless youre paying a lot of attention to what is going on, otherwise it will just be what you expect and nothing out of the ordinary

--edit--
silly spelling mistakes
 
Why dont ya people just give it a chance... if you dont have money... dont buy... if you have money and want to spend it on something esle... do it... its pointless to complain what it can or cant do... its just new technology... you cant expect much from it... ya havent given it some chance to mature or become optimize... why dont ya wait till UT2007 comes out to see if its good to buy or not... before that happens... just sit and wait instead of complaining how bad it is
 
I actually like ATI's products a lot but am disappointed with their idea of using a 3rd dedicated GPU for physics processing. I thought the AGEIA PhysX card was such a bad idea and now ATI wants me to buy a 3rd videocard for physics? I don’t care what you want to call it, it’s just as bad as buying the dedicated physics card.
 
we arent complaining as we are discussing, as you say, so far there is too much unknown, too much untestable, to either justify purchasing a physx card, or, if you dont have it already, a second graphics card or multicore cpu for the physics

the fact of the matter though is that physics calculations are going to get more complex and one of these solutions will be required to keep up, so far it doesnt look like the physx card will cut it because of, what appears to be, some serious limitations in implementation

it will take more than just one game to show the potential of this, and other, solutions, it will also take maturity of the system, both of which will take valuable time. the havok solution will have an easier implementation, if you have sli cards just upgrade driver and it goes, assuming all havok engine driven games are supported then immediately there will be serveral testable titles to run against

i just hope that nvidia offer an option of primary/secondary physics device, if i get a quad core cpu im sure id rather its cores do more of the physics work than the sli cards, let them deal with the rendering!
 
physics is the future for all gaming, not as big as gfx mindu, but a seperate physics solution is need to provide the best quality. obviously i would like to see ageia make a break through and it is way too early to tell if it will. consider that it does hav some bigs games, ut2007 for example. dedicated solutions hav always worked better then software or driver solutions and atis and ageias look promising, but just like quad sli, the software for them has to be worked out
 
The issue here is that it is not even bleeding edge tech, it's just bleeding - the one game that uses it doesn't represent well, there are few others on the horizon, and it is very expensive to boot. Not a good combo - it almost seems like Ageia thought they could get by on the novelty factor until they improved the product and the number of games supported. Almost like they are charging $299 to be a beta tester - ouch...
 
excellent way of putting it. why should I spend money to be a guinea pig for a product that does not yet work (essentially)?

PhysX seems to be not a niche market, but a non-market. What niche does it fill? Nobody can use it effectively, and we don't know if it will actually start being used significantly in the future.

$300 US is insane for such a product, I can buy a video card which will reasonably run any game on the market for that price. This product is only currently supported on a game I'm not going to buy. I knew the result of this article before it was written, because I thought to myself "what possible need do I have for this product?". At least with the Voodoo add-on cards, I saw immediate significant tangible increases in performance.
 
The Physix card is just a waste of money as well as power, and space in ones rig. Besides it only supprts a number of games, but it would be much better if it were backwards compatible with older games, such as CoD2. It just seems that Ageia are trying to take over the "real world physix" cenario with games, but it just seems that it hasnt spent enough time in creating this piece of hardware. My verdict wait until Ati Or Nvidia come out with something better.
 
While it's not worth it right now, I'm hoping the PhysX card succeeds because it's the only implementation that's not just eye-candy. The problem with Nvidia and ATI's solutions is that the resulting physics calculations don't affect gameplay. PhysX has the possibility for in-game physics to alter the gameplay and I think that could be really awesome.
 
They should have used the PCI-e x1 slot insted since you could get close to twice the bandwidth then just the PCI slot.
 
In my opinion this is just another idea as how to get more money from pc users.I fear that if Ageia's popularity will rise then it's just a matter of time before you'll have to buy 10 different cards just to play anything more then "minesweeper" Motherboard producers will follow, and try to imagine the size of your "desktop pc" then. ;-)
 
Isnt this the kind of technology that could be added to a graphics card or even built into the mother board? This just sounds like another way people can waste their hard earned money and get very little in return. Advertisements can be very seductive and very deceiving don't believe the hype!
 
Isnt this the kind of technology that could be added to a graphics card or even built into the mother board? This just sounds like another way people can waste their hard earned money and get very little in return. Advertisements can be very seductive and very deceiving don't believe the hype!

read the whole article...
 
I actually liked the article. Thanks. :)
I was waiting for this article for a long time.

It had some important issues like the old "clipping" problem that never seem to get fixed. The more realistic weight causes very unrealistic leaning angles. Frame rates are slower, etc.
I think all the PhysX card hype is a small step in the right direction, but there is a LOT more work to do that just adding some hardware. The problems also seem bigger than "Hovok vs. Ageia" or "1 card vs. 3 cards", etc (they're all far from ideal)

I am hoping developers can start taking physics more seriously if they know the hardware can handle the extra load and users are expecting more realistic physics while also expecting more object interaction.
In a few years there might even be very usable phyics hardware on the motherboard like AC97 or cheap video with shared memory.
 
Seeming how there is a hack for Ghost Recon that allows the same effects without using a PhysX card, I don't see their solution coming out on top.

It will be interesting to see what UT2007 gets from the PPU.

Actually, I'm a bit surprised that MS hasn't come out with a Direct X standard for physics.
 
The solution always seemed obvious to me, ATI and Nvidia need to slap on a PPU to all their new cards then all game developers will support it. ATI proved with All in Wonder series they could combine 2 hardware components so why can't they do it now?

I honestly thought ATI's developers were retarded when I saw they wanted us to buy 3 video cards to get physics.
 
The "old clipping problem" points out another weakness with the PhysX card, or any other type of hardware solution. If the game isn't written to allow it, the hardware can't fix it. Doors that don't fall over, clipping, and other oddities are part of the software in the game. The way I see it, and I may be wrong, is that the PhysX card will just do more to show the bugs or poor software codes then it will do anything that actually fixes the base problem and give realistic explosions, etc.

I think that the solutions from Nvidia and ATi will offer more in the long run than the the PhysX card, and their solutions will force the game companies to write games that use the hardware. Time will tell, so we can only wait and see at this point. For now, I'm not buying anything.
 
A particle effects accelerated, yeah. Maybe if software comes out that really shows something interesting I would consider one, but sure as heck not for a few extra bits of shrapnel.
 
First of all this card debuting at $300 is not just expensive...it's utterly insane. Paying $300 for something that at present is little more than a paper weight is mind boggling to me.

I don't know what the marketing geniuses at ageia are thinking, but this company is not going to go far with this kind of planning.

This is not a graphics card. This is not an essential necessity for gamers. At present it’s a novelty accessory at best.
Most people are not willing to pay $300 for a graphics card, what makes you think they will pay that for a useless accessory like this?

This item should've debuted in the $50-$100 range, and even then it would've had a hard battle ahead (just look at how many people actually spend that kind of money for sound cards...not many as most people just go for the onboard option.)

Second, although ageia is desperately awaiting the release of UT2007, I don’t think that it will give it any sort of a boost in sales.
UT2007 is a multiplayer game, and many of the lan problems outlined in the article, will be present there too. All physics eye-candy will be client side, which means most people will turn them off anyway because it will most likely put them at a disadvantage. Nobody wants their view to be obstructed by a pile of smoke and debris while their opponent has a perfectly clear view.

The only way I see this product succeeding is if either ageia slashes their prices to reasonable levels(around $100), or even better if they go for the integrated option, either on the motherboard or video card.
Otherwise they should start looking for new jobs…