Will hype kill the PPU.

[Ockie]

Personally, I'm quite hyped about this product... but that leaves me with a bad reminder that in the past I've always been burned by hype. When the pci-e version comes out, I'm planning on digging in... however, I guess I'm the type who tosses 300 dollars at something expecting 300 dollars worth of work.


The 7900gt is 300 bucks, most people dont even have a card such high end and you can run all games on cards with lower performance specs...


so will hype kill the ppu? if there is no killer app on the official release date i'm sure it's not going to survive due to the fact that people expected the moon and got a rock so to speak.

 

[reddhaus]

I am hyped for the PPU!!!!

Yee Ha!!

I will oder one up as soon as I can find one online.

 

[Master [H]]

It's hard to say. This is the first product from Ageia, so I think it's premature to say that anything will kill the PPU.

It's got some things going against it:
1.)Brand new technology. The unfortunate side here is that we're dealing with a new company's first product. This is new territory for everyone because it's not been done before (at least for gaming on this scale). There's going to be bugs and other issues with drivers and the hardware itself.

2.)New market. Ageia has created a new market, and while it's a fascinating one, if the product doesn't intrigue enough people to purchase, they could be SOL. They need to be in it for the long haul.

3.)Apps. As stated, it's a new product. It has new capabilities, but the market for the product is so new that we don't really have software (mainly games) to take advantage of it.

What PhysX does have going for it:

1.)Hardware. Granted, it's 1st gen technology here, but it's pretty much an accepted law that hardware performance > software performance. Havoc may be a good thing, but if the PPU was designed for a specialized operation, like the GPU or hardware RAID controllers with dedicated XOR processing.

2.)First to market with the hardware. This is kinda redundant, but Ageia is first to market with a hardware physics processing solution. Even if another company starts making a rival PPU, Ageia still has them beat.

3.)The product is available. It's not just vaporware---you can actually purchase a PPU card.


Right now, it's really too early to tell how Ageia will fair. It's pretty easy to see it going very well or very poorly for Ageia.

 

[ATIIsBetter]

I am hyped for the PPU!!!!

Yee Ha!!

I will oder one up as soon as I can find one online.

http://www.hardforum.com/showthread.php?t=1053126

 

[Adidas4275]

interesting

 

[NulloModo]

Personally, I hope the whole idea of a PPU does a swift death. Games are games, they don't need realistic physics. Maybe flight sims and the like, but those are hardly enough of a market to support a new technology. Integrating the technology into DX10 cards to that they can use unused pipelines for other purposes is a far better idea. Computer buyers in general want less cards to pay for, not more.

 

[Big Fat Duck]

Personally, I hope the whole idea of a PPU does a swift death. Games are games, they don't need realistic physics. Maybe flight sims and the like, but those are hardly enough of a market to support a new technology. Integrating the technology into DX10 cards to that they can use unused pipelines for other purposes is a far better idea. Computer buyers in general want less cards to pay for, not more.

personally, i dont want to pay another 300 just to enable some physics settings that developers designed into their games that could have been perfect using Havok

 

[Terra]

personally, i dont want to pay another 300 just to enable some physics settings that developers designed into their games that could have been perfect using Havok

Havok?
Perfect?
"Perfect" as long as your CPU dosn't get bogged down...
But please, tell me...are you claiming that the CPU is faster than the PPU for phycis?


Terra - Never knew "perfect" meant "limited"...

 

[SuperGee]

Personally, I hope the whole idea of a PPU does a swift death.

If the Dev community keep a strong support for PhysX and Havok FX stay behind. this means a lot of games you can play only in low Physx settings.
There will alway's be !!!!!! wich are for GPU camp and other for PPU camp and you choose your side. But PPU have a lot of games behind it. Will Havok fX catch up maybe.
I let the games deside. What if there are a lot Of nice PPU games like UT2007 and a few nice Havok FX game and you want to play both in optimum. You need both a PPU and a extra G-Card for Physics.

Games are games, they don't need realistic physics.
Maybe flight sims and the like, but those are hardly enough of a market to support a new technology.

Well Physics is not only for simulation and realism in that field.
Realisme can be fun Physics to. Like special ability's Or PhysX weapons. Or for more interaction with the enviorment in a realistic way for much more emersion. Gameplay Physics, More Physx computational eyecandy. effect Physx.
The game industry is going the way of more Physics for a lot of years now. But due to ageia it get a big bump by the use of Hardware accelerated PhysX by PPU and the competion wich jump on this wagon to Like GPU firms and supporting API HavokFX.
So Physx will grow drastacaly with or with out ageia. The ball is rolling and can't be stopped anymore. The question is who will win in the shortrun, longrun and in the far future. To early to tell.

Integrating the technology into DX10 cards to that they can use unused pipelines for other purposes is a far better idea. Computer buyers in general want less cards to pay for, not more.

Oh but GPU physics is al about more cards and more cash flow to ATI or nV.
And yes your are in the GPU !!!!!!'s camp. So didn't you know for optimum GPU pleasure you need a extra card.
But DX10 is not about Physics. MS will put Physics support much later in a later DX version. The market must settele first for requiering a standard. A revision of DX10 like DX10 b or c or d or even just in DX11 this is far in the future. HavokFX will be the first choice for devs in the near future to use GPU PhysX in games. And those SM4.0 later give it a bit more power.
But ageia is in the lead. Much Dev support and Ageia is a free hardware ppu support solution Havok + FX not. That feed PPU support drasticly.
In the end its all about the games. They deside who wins.

personally, i dont want to pay another 300 just to enable some physics settings that developers designed into their games that could have been perfect using Havok

So your a Havok fan so in the GPU camp don't you mean Havok FX.
Because Havok you can put against Physx without PPU wich means just like Havok runnin on the CPU. The normal low Physics setting.

Havok FX with extra GPU $$$ for Physics you can put agianst PhysX with PPU $$$

But The games desides who wins.
Plus that $300 Price is now. If the PCI-E version will be there the price of the PCI2.2 will get down. Price in the near fututere is unknow also when PPU2 will be there.

And at first GPU and PPU Physix will be at first something more for the Gamer enthausiast. In both way it will cost a bit more.
In the future the whole situation could be different also in cost and produkt line.

But cost is just one thing. Not everething.

I wait for the games I like and then deside for a extra GPU. next to my PPU for Havok Fx games

BTW Hyped I think not. Hype is wenn the public is to over enhausiast about something in this case PPU. The feeling that more people are GPU PRefering means no hype at all. Wich is good. Because Hype hurt if it does not deliver. But A Killer app. Like
UT2007 or CF revelutions could be could make the public do a 180 degree turn on the isue. Because its all about the games so it what the devs are doing with it.

 

[Chernobyl1]

I'm sticking my neck out here to try and describe everything I have learnt so far about the PhysX card. If anything is incorrect, I am happy to learn about it.

The PhysX card is having to create its own market which is slowly gaining ground so its very difficult for AGEIA to get going and Kudos to them for managing it with a product which hasnt shone yet!
The question is, do we really need the PhysX card when the CPU and/or GPU can handle the job just as well if the PhysX API is coded for them?

The PhysX card will not do Physics calculations that are much better than that we have seen already. It needs higher precision on many more objects than the PhysX card is capable of. A second CPU is a better choice for precision maths and is a neater solution.
However when working with lots of objects, a CPU can get bogged down. More CPU cores or passing some of the iterative work to a programmable GPU will work.

AGEIA allow their hard coded algorithms to be used but do not allow the coder to program the PhysX hardware, so you only get to use the routines that AGEIA supply when using the PhysX card. You cannot write your own Physics routines and get the PhysX card to process them.
It has been demonstrated that an extra 'Physics' graphics card on the PCI-E bus is better suited to the task of Physics processing due in part to its ability to communicate more directly with the graphics card and it is programmable unlike the PhysX card.

To do true Physics simulations with rotating, stacked, heavy and light etc objects interacting takes an immense amount of processing which the current PC isnt capable of.
Even to get it to look realistic with fixes and patches rather than true algorithms is hard work and is how games have to be programmed.
The horsepower AGEIA have brought to the table is a tiny bit extra compared to what we already have/can get so we arent going to see a big difference from the current PhysX card.
The PhysX card also requires a non insignificant amount of CPU power to drive it so it needs to be able to offload other processing from the CPU to compensate. The current implementation doesnt succeed in doing this.
Of course this problem goes away if the minimum spec for the card is with a Dual core CPU, but do we really need a PhysX card then?

There are 2 basic sides to Physics processing, Visuals and Precision movement/collision detection. Both are described as Physics but they are very different.
If the card is used to display only more visuals, those visuals will be for looks and will have no impact on gameplay. This is mostly what we have seen so far as it is the most effective to demonstrate and takes the least processing power.
The interaction between physical objects gets vastly more complex the more there are and the more types of shape, size and movement they have. To accurately obtain natural behaving movement from these objects is one of the goals of the PhysX processor but it isnt capable of handling much more than we have already seen.
We would like the PhysX card to draw complex objects with what appears to be true to life Physics in movement. Unfortunately, more complex objects add an exponentially larger workload to the PPU if movement physics is also applied to the object.
Maybe the next gen card built on PCI-E will be more capable?

Physics processing has been around a long time and has been improving without any external processing solutions.
With Dual core here and Quad core coming/improving for both CPU and GPU, its going to be a tough ride for AGEIA to sell the PhysX card.
Now that the focus is on Physics processing, CPU and GPU manufacturers can optimise parts of the processing cores to improve on this. Bear in mind that they compete head on with the PhysX card without any improvements so far!

My bet is AGEIA will make their money from licenses to use their algorithms in software, which will no doubt improve.

 

[nhusby]

I dont think its too hyped... I have never heard anything about it on tv or even in magazines (except a paragraph from E3 reviews.)

I am personally hyped... I think alot of people see the potential of what the PPU could do for games and say "this is exactly what I was waiting for." When CF comes out I'm going to build a new rig just for it. (I really need a new rig) I justify it because I'm going into 3-d rendering and simulation programming at school (basicly game design with professional name) but my real motivation is to play CF because it is something new. I built my last system when half-life came out, because that was something new (to my anyhow), and I havnt seen anything really new since then.

but anyways... I admit it... I am hyped...

 

[Terra]

AGEIA allow their hard coded algorithms to be used but do not allow the coder to program the PhysX hardware, so you only get to use the routines that AGEIA supply when using the PhysX card. You cannot write your own Physics routines and get the PhysX card to process them.
It has been demonstrated that an extra 'Physics' graphics card on the PCI-E bus is better suited to the task of Physics processing due in part to its ability to communicate more directly with the graphics card and it is programmable unlike the PhysX card.

I will just adress some of the worst misconception you have:
Show me where(and how) PCI-E is better for physcis?
The physcis is calculated internally, via a MASSIVE parrallel architechture.
No tecttures ect. are being transferred, only the result of the calculations.
So even if calculation physcis is a BIG numbercruching game, PCI is more than enough...for the moment.

To do true Physics simulations with rotating, stacked, heavy and light etc objects interacting takes an immense amount of processing which the current PC isnt capable of.

Yes it does...

The horsepower AGEIA have brought to the table is a tiny bit extra compared to what we already have/can get so we arent going to see a big difference from the current PhysX card.

Actually we have ween nothing yet.

The PhysX card also requires a non insignificant amount of CPU power to drive it so it needs to be able to offload other processing from the CPU to compensate. The current implementation doesnt succeed in doing this.

Acutally that is very wrong.
The PPU giving lower FPS are not due to the CPU, but the GPU's.
And the lower FPS ahd been adressed by driver updates and is about NIL today...

Of course this problem goes away if the minimum spec for the card is with a Dual core CPU, but do we really need a PhysX card then?

You think dual, quad or even more cores is going to be better for graphics than the GPU?
If not, please aplly the same thinking to the PPU...

There are 2 basic sides to Physics processing, Visuals and Precision movement/collision detection. Both are described as Physics but they are very different.
If the card is used to display only more visuals, those visuals will be for looks and will have no impact on gameplay. This is mostly what we have seen so far as it is the most effective to demonstrate and takes the least processing power.
The interaction between physical objects gets vastly more complex the more there are and the more types of shape, size and movement they have. To accurately obtain natural behaving movement from these objects is one of the goals of the PhysX processor but it isnt capable of handling much more than we have already seen.
We would like the PhysX card to draw complex objects with what appears to be true to life Physics in movement. Unfortunately, more complex objects add an exponentially larger workload to the PPU if movement physics is also applied to the object.
Maybe the next gen card built on PCI-E will be more capable?

I'm tellig you, we have barly tapped into the PhysX possibilities.
Stop thinking in boxes only...fluids, cloths, destructable architechture, trigger points, friction ect...we havn't seen nothing yet

Physics processing has been around a long time and has been improving without any external processing solutions.
With Dual core here and Quad core coming/improving for both CPU and GPU, its going to be a tough ride for AGEIA to sell the PhysX card.
Now that the focus is on Physics processing, CPU and GPU manufacturers can optimise parts of the processing cores to improve on this. Bear in mind that they compete head on with the PhysX card without any improvements so far!

Show me a physcis engine (besides PhysX) that has real time destructable cloth..and more than 5 FPS?

My bet is AGEIA will make their money from licenses to use their algorithms in software, which will no doubt improve.

WRONG!!!
The PhysX SDK is FREE...if you implement support for the PhysX hardware...
Where did you get all your misinformantion?!

Terra...

 

[nhusby]

I'm sticking my neck out here to try and describe everything I have learnt so far about the PhysX card. If anything is incorrect, I am happy to learn about it.

The PhysX card is having to create its own market which is slowly gaining ground so its very difficult for AGEIA to get going and Kudos to them for managing it with a product which hasnt shone yet!
The question is, do we really need the PhysX card when the CPU and/or GPU can handle the job just as well if the PhysX API is coded for them?

The PhysX card will not do Physics calculations that are much better than that we have seen already. It needs higher precision on many more objects than the PhysX card is capable of. A second CPU is a better choice for precision maths and is a neater solution.
However when working with lots of objects, a CPU can get bogged down. More CPU cores or passing some of the iterative work to a programmable GPU will work.

AGEIA allow their hard coded algorithms to be used but do not allow the coder to program the PhysX hardware, so you only get to use the routines that AGEIA supply when using the PhysX card. You cannot write your own Physics routines and get the PhysX card to process them.
It has been demonstrated that an extra 'Physics' graphics card on the PCI-E bus is better suited to the task of Physics processing due in part to its ability to communicate more directly with the graphics card and it is programmable unlike the PhysX card.

To do true Physics simulations with rotating, stacked, heavy and light etc objects interacting takes an immense amount of processing which the current PC isnt capable of.
Even to get it to look realistic with fixes and patches rather than true algorithms is hard work and is how games have to be programmed.
The horsepower AGEIA have brought to the table is a tiny bit extra compared to what we already have/can get so we arent going to see a big difference from the current PhysX card.
The PhysX card also requires a non insignificant amount of CPU power to drive it so it needs to be able to offload other processing from the CPU to compensate. The current implementation doesnt succeed in doing this.
Of course this problem goes away if the minimum spec for the card is with a Dual core CPU, but do we really need a PhysX card then?

There are 2 basic sides to Physics processing, Visuals and Precision movement/collision detection. Both are described as Physics but they are very different.
If the card is used to display only more visuals, those visuals will be for looks and will have no impact on gameplay. This is mostly what we have seen so far as it is the most effective to demonstrate and takes the least processing power.
The interaction between physical objects gets vastly more complex the more there are and the more types of shape, size and movement they have. To accurately obtain natural behaving movement from these objects is one of the goals of the PhysX processor but it isnt capable of handling much more than we have already seen.
We would like the PhysX card to draw complex objects with what appears to be true to life Physics in movement. Unfortunately, more complex objects add an exponentially larger workload to the PPU if movement physics is also applied to the object.
Maybe the next gen card built on PCI-E will be more capable?

Physics processing has been around a long time and has been improving without any external processing solutions.
With Dual core here and Quad core coming/improving for both CPU and GPU, its going to be a tough ride for AGEIA to sell the PhysX card.
Now that the focus is on Physics processing, CPU and GPU manufacturers can optimise parts of the processing cores to improve on this. Bear in mind that they compete head on with the PhysX card without any improvements so far!

My bet is AGEIA will make their money from licenses to use their algorithms in software, which will no doubt improve.

I think you need to do more research...
when terra see's this he's gunna have a hay day LoL EDIT: he already did, while I was typing...

I'll point out a few things to you.

1) The CPU is a superior general processor, but it is not very paralell. In game physics require lots and lots of paralell interactions to be processed all at once. If a CPU does the job of a PPU then it has to string all of the calculations into a line. The result in a game designed with alot of physics is a second CPU with a 20% utilization and a game running at 10FPS. PPU's do their job better than CPU's.

2) The GPU does graphics better than a CPU. The GPU does physics better than a CPU. But a PPU still does physics better than a GPU. GPU's are designed for paralell processing, but not to the same scale as a PPU. Havok is going to use a second GPU to do effects physics because GPU's are redily available, most people have them, and Havok is betting that when games with their API come out that alot of people will have an old GPU that can do the job. They dont want to gamble on better, but new technology.

3) CPU's and GPU's can do all of the physics in current games. That might have something to do with game devs designing games to work on current spec's. The point of the PPU isnt that it will do faster what your PPU does. The point is that the PPU will make a different kind of game possible. Current games are limited by the potential of CPU's. Hopefully future games will not have that limitation because of the PPU.

4) Ageia gives away their API. Ageia doesnt plan on making money on their API. Their cash is going to come from PPU sales or bust.

 

[Chernobyl1]

wow, I didnt expect to be ridiculed.
Please re-read what I have said as most of your points have already been explained.

 

[nhusby]

wow, I didnt expect to be ridiculed.
Please re-read what I have said as most of your points have already been explained.

Is this intended for me?

I think the points I made conflict with yours. I didnt intend to ridicule you, I am simply in disagreement. Maybe we are in agreement on some points, but the way I interpret what you posted I bilieve you are wrong. I expect you may think I am wrong, and thats entirely within your rights. I have no new evidence to support my points that is not already on this forum in old threads.

 

[HOCP4ME]

Please re-read what I have said as most of your points have already been explained.

Okay, I'll try.

The PhysX card is having to create its own market which is slowly gaining ground so its very difficult for AGEIA to get going and Kudos to them for managing it with a product which hasnt shone yet!

That's right, the PhysX card is entering a new market. It's suprising that AGEIA has gotten so many developers to offer even minimal support for a new device that may or may not make it in the future.

The question is, do we really need the PhysX card when the CPU and/or GPU can handle the job just as well if the PhysX API is coded for them?

That is just wrong. A GPU isn't as good at physics as a PPU, and a CPU can't even come close. A GPU can manage it, but a CPU can't. Not even an eight-core CPU.

The PhysX card will not do Physics calculations that are much better than that we have seen already.

That is not true. A PhysX card can do much more physics than we have already. That's the whole point of it.

A second CPU is a better choice for precision maths and is a neater solution.

Hmm, okay, $300 for a second CPU that can do around 3000 objects, or $300 for a PPU that can do around 32000. Tough choice...

However when working with lots of objects, a CPU can get bogged down. More CPU cores or passing some of the iterative work to a programmable GPU will work.

There are several problems with that. First off, multiple CPU cores still can't come anywhere near the PPU's abillities. You would need around 11 CPU cores to equal the PPU. We won't see that for a long time.

Second, combining the CPU and GPU to do physics is going to be tough. Imagine having a calculation be split up by one core, then sent to other cores and the GPU over the PCI-E bus, then sent back to the first core, and put back together again. Talk about lag...

Third, why would developers want to take the time to do this? Hmm, okay, we have a new chip here called the PPU. Should we take the easy route and use the chip for physics processing? Or, should we optimize the API, multithread the physics, figure out a way to spread it across multiple cores as well as the GPU, then optimize the drivers to prevent lag? Tough decision...

Finally, all of that is unnecessary. Developers could save the required time and get even more power if people would just adopt the PhysX instead of trying to do it using every other chip in their system. What's next, sending some of the physics to the RAID controller?

AGEIA allow their hard coded algorithms to be used but do not allow the coder to program the PhysX hardware, so you only get to use the routines that AGEIA supply when using the PhysX card. You cannot write your own Physics routines and get the PhysX card to process them.

The PhysX can be programmed. AGEIA just hasn't allowed that yet, because they want to see a more basic utillization of the technology first. Programming the PPU isn't required to simply go beyond what we have today. That's something that we'll see in the future.

It has been demonstrated that an extra 'Physics' graphics card on the PCI-E bus is better suited to the task of Physics processing due in part to its ability to communicate more directly with the graphics card and it is programmable unlike the PhysX card.

The PhysX doesn't even max out the PCI bus. There's no reason why PCI-E would help it. And it doesn't need a more direct communication with the graphics card. The physics calculations have to be sent back to the CPU anyway.

To do true Physics simulations with rotating, stacked, heavy and light etc objects interacting takes an immense amount of processing which the current PC isnt capable of.

You're right, true physics is very hard to do. But the PPU gets us much closer to "true physics" than a CPU.

Even to get it to look realistic with fixes and patches rather than true algorithms is hard work and is how games have to be programmed.
The horsepower AGEIA have brought to the table is a tiny bit extra compared to what we already have/can get so we arent going to see a big difference from the current PhysX card.

No, we don't already have anything that can compare to the PhysX, and it does make a big difference. That's the whole point.

The PhysX card also requires a non insignificant amount of CPU power to drive it so it needs to be able to offload other processing from the CPU to compensate. The current implementation doesnt succeed in doing this.

The PPU offloads all physics from the CPU. It's actually the GPU thats slows it down. The PPU introduces new objects, and the GPU has to draw those objects. And guess what? Physics done on a CPU core or a GPU would have THE SAME LIMITATIONS!

Of course this problem goes away if the minimum spec for the card is with a Dual core CPU, but do we really need a PhysX card then?

First off, a dual-core CPU doesn't help the PhysX much. Second, a dual-core CPU isn't nearly as powerful as a PhysX.

And like Terra said, if you think CPU's can do physics just as good as a PPU, you must think CPU's can do graphics just as good as a GPU, which is totally incorrect.

There are 2 basic sides to Physics processing, Visuals and Precision movement/collision detection. Both are described as Physics but they are very different.
If the card is used to display only more visuals, those visuals will be for looks and will have no impact on gameplay. This is mostly what we have seen so far as it is the most effective to demonstrate and takes the least processing power.

Yes, this has been used for demostration because people like to see eye candy.

he interaction between physical objects gets vastly more complex the more there are and the more types of shape, size and movement they have. To accurately obtain natural behaving movement from these objects is one of the goals of the PhysX processor but it isnt capable of handling much more than we have already seen.

LOL! Look, if the PPU can't do it, the CPU can't even try it. That's like saying "there's no reason to get a Viper instead of a Civic, because it still can't do 300MPH."

We would like the PhysX card to draw complex objects with what appears to be true to life Physics in movement. Unfortunately, more complex objects add an exponentially larger workload to the PPU if movement physics is also applied to the object.

Once again, that argument can be completely thrown out, because if the PPU can't do it, neither can the CPU or GPU. Don't complain that the PPU isn't equivalent to a quantum CPU. It's not that good, but it's better than anything we've gotten so far.

Maybe the next gen card built on PCI-E will be more capable?

"The next-gen card will be more capable". Why thank you, Captain Obvious.

Physics processing has been around a long time and has been improving without any external processing solutions.

Yep, but it can improve much faster if it does have an external processing solution.

With Dual core here and Quad core coming/improving for both CPU and GPU, its going to be a tough ride for AGEIA to sell the PhysX card.

Multi-core CPU can't even come close to the PPU. GPU physics can come close, but who wants to give up GPU power for physics? If I had two video cards, I would want to use SLI, not give one up for physics.

Now that the focus is on Physics processing, CPU and GPU manufacturers can optimise parts of the processing cores to improve on this. Bear in mind that they compete head on with the PhysX card without any improvements so far!

LOL, they compete "head-on"? Let's see:

Number of PhysX games out: 6
Number of Havok FX games out: 0

And as for multi-CPU core physics, we don't even have a single company supporting that yet.

My bet is AGEIA will make their money from licenses to use their algorithms in software, which will no doubt improve.

AGEIA's API is free.

 

[Chernobyl1]

I found this article which may help you understand where I'm coming from
http://tomshardware.co.uk/2006/08/03/the_scientists_opinions_on_gaming_physics_uk/

 

[Terra]

I found this article which may help you understand where I'm coming from
http://tomshardware.co.uk/2006/08/03/the_scientists_opinions_on_gaming_physics_uk/

Toms' are a bunch of retards
Their first article was lame (wich I pointed out) and they keep on getting more and more lame...
They ased scienctist abut a piece of gaming hardware
Next I presume they will ask a photographer about the 3D of a GPU...in case it's not only physcis they understand nothing off...

Terra - Don't use Tom's...dosn't make you look good

 

[Chernobyl1]

I was expecting that kind of response from you.
The article is a very good read, especially this part

"So, what hardware will be best for physics? We still have to wait and see. The optimal solution will probably be some sort of compromise, where a multi-core CPU does the complex calculations, and some kind of hardware (be it the GPU or a PPU) takes care of easier calculations where a larger number of objects is involved. And of course, you wouldn't go wrong having dual graphic cards to render it all on the screen with a decent frame rate."

 

[Terra]

I was expecting that kind of response from you.
The article is a very good read, especially this part

"So, what hardware will be best for physics? We still have to wait and see. The optimal solution will probably be some sort of compromise, where a multi-core CPU does the complex calculations, and some kind of hardware (be it the GPU or a PPU) takes care of easier calculations where a larger number of objects is involved. And of course, you wouldn't go wrong having dual graphic cards to render it all on the screen with a decent frame rate."

The article is bullshit, they should have talked to gamers and developers.
Tom's are again speaking out their asses and understading nothing.

Just to sugest that the CPU does physcis is lame...
Time and time again it ahs been stated that CPU's SUCK at physics...what part of that don't you understand?!
Or do you claim that the CPU also could render graphics?
THE CPU SUCKS AT MASSIVE PARRALLEL CALCULATIONS...PERIOD!

No wonder you are wrong if you listn to Tom's....CPU for physcis...come on
Not even Intel or AMD would make such a claim....

Terra - CPU's are not good for either physcis or graphics...be they dual or quad-core!

 

[R1ckCa1n]

Is the PPU overhyped? In this forum we get a HELL YEAH.

Until something more than a tech demo, 1/2 assed game patch, or a killer game come out supporting features that can only be done with a dedicated PPU, it is all hype.

I went for my weekly visit to Frys and see the same amount of BFG PPU's sitting on the shelf and has been so for month. I at least expected to see return tags on them but no, they just are not selling.

 

[Chernobyl1]

The article is bullshit, they should have talked to gamers and developers.
Tom's are again speaking out their asses and understading nothing.

Just to sugest that the CPU does physcis is lame...
Time and time again it ahs been stated that CPU's SUCK at physics...what part of that don't you understand?!
Or do you claim that the CPU also could render graphics?
THE CPU SUCKS AT MASSIVE PARRALLEL CALCULATIONS...PERIOD!

No wonder you are wrong if you listn to Tom's....CPU for physcis...come on
Not even Intel or AMD would make such a claim....

Terra - CPU's are not good for either physcis or graphics...be they dual or quad-core!

Calm down and read the article again.
I read Toms articles and make my own mind up whether they are based on sound knowledge. I dont need you to do that for me.
The article is based on sound knowledge and input from Physicians whos word I am more likely to believe

I cant understand how you can be so enthusiastic over a product which is costly and doesnt perform at all well.
The PhysX card has been out for 6 months and there still isnt a compelling reason to buy one other wise I would have one.

 

[Terra]

Calm down and read the article again.
I read Toms articles and make my own mind up whether they are based on sound knowledge. I dont need you to do that for me.
The article is based on sound knowledge and input from Physicians whos word I am more likely to believe

Then take this word of advice:
There are not going to be physcis simulation on a nucelar level in games in uor lifetime.
Wich means the article is a waste of time...

I cant understand how you can be so enthusiastic over a product which is costly and doesnt perform at all well.

If you base that on articles from Tom's, I say goodbye to you.
If you based it on anything else, feel free to contribuate...

The PhysX card has been out for 6 months and there still isnt a compelling reason to buy one other wise I would have one.

Godd thing I ain't you then, or read Tom's...

Terra - Good bye...

 

[Chernobyl1]

lol
goodbye

 

[piako]

It would be cool if say they had some type of waterfall blood like get killed in a waterfall and see it or something. Or maybe like have the wind f with your bullets. I'm thinking FPS here.

But seriously. How many game developers are going to sit down with the SDK and retool their precious game code to support the PPU? Wonder if it's easy to implement? It would make sense to me if a game developer sat down and said, ok, about x % of the gaming population wants ppu support compared to xx% wanting a new game. They'll make the new game. At least from a business decision view the would make more money.

 

[Chernobyl1]

Hopefully Microsofts upcoming Physics implementation will be the answer to our prayers.
As with DirectX, it is likely to become part of everyday game coding.
No doubt this will develop into a multi hardware capable solution as well (ie CPU, PPU, GPU).
Good times are a coming

 

[R1ckCa1n]

No doubt this will develop into a multi hardware capable solution as well (ie CPU, PPU, GPU).

Or could develop into CPU, GPU, GPU or CPU, GPU, GPU, GPU

 

[HOCP4ME]

Or could develop into CPU, GPU, GPU or CPU, GPU, GPU, GPU

This is getting out of hand. I don't understand why everyone wants to go with the absolute most complex solution possible just to ensure that they don't eventually have to buy a PPU.

We get this new card that can do awesome physics simulations, beyond anything we've seen so far.

Next, everyone says "but wait, we'll just use extra CPU cores for that."

Then, after realizing that CPU's can't do parallel processing, everyone says "oh, then we'll just move it to an extra GPU"

But since that wouldn't be powerful enough, everyone says "hey, let's put some physics on the extra CPU cores and some on the extra GPU, that will equal the PPU."

So, what should we do? Split up the physics equations, send some of them to a GPU, send some of them to extra CPU cores, and piece them together again? Or, should we just apply Ockham"s Razor here and use a PPU?

It's like everyone is desperate to prevent this thing from getting off the ground. Why should we do physics on a single PPU? We could just use three CPU cores and a GPU, couldn't we? And why not use that RAID controller chip, or that sound card, or that NB for physics processing? After all, we have to make sure we are 100% utillizing every piece of silicon in our systems before we'll buy a PPU, right?

 

[drizzt81]

There is a simple reason why people would rather have more CPUs than a PPU: The CPU can increase the PPD in Folding At Home. GPUs are likely to be used in F@H at the end of the year, which allows them to help fight cancer. The PPU won't!

 

[Chernobyl1]

This is getting out of hand. I don't understand why everyone wants to go with the absolute most complex solution possible just to ensure that they don't eventually have to buy a PPU.

We get this new card that can do awesome physics simulations, beyond anything we've seen so far.

Next, everyone says "but wait, we'll just use extra CPU cores for that."

Then, after realizing that CPU's can't do parallel processing, everyone says "oh, then we'll just move it to an extra GPU"

But since that wouldn't be powerful enough, everyone says "hey, let's put some physics on the extra CPU cores and some on the extra GPU, that will equal the PPU."

So, what should we do? Split up the physics equations, send some of them to a GPU, send some of them to extra CPU cores, and piece them together again? Or, should we just apply Ockham"s Razor here and use a PPU?

It's like everyone is desperate to prevent this thing from getting off the ground. Why should we do physics on a single PPU? We could just use three CPU cores and a GPU, couldn't we? And why not use that RAID controller chip, or that sound card, or that NB for physics processing? After all, we have to make sure we are 100% utillizing every piece of silicon in our systems before we'll buy a PPU, right?

If you have a nice new dual core system with a PhysX card, wouldnt you like your PC to use spare GPU and CPU cycles to improve your physics experience?

There is a lot more to the story than you are mentioning, CPU's and GPU's can be used in tandem to perform PhysX.
This is where the PCI-E bus would come in handy as it can send and receive data simultaneously, allowing much better integration of separate components to perform larger tasks.

Why waste hardware?
The complexity of what hardware to use isnt an issue, the Physics drivers are sure to come along for each type of hardware.
If most of the public has unused hardware and a developing game can benefit from the use of it, the developer would be daft not to use it.

 

[RAutrey]

Yea, yea, yea... PPU this and PPU that. Give me ONE app that decently showcases the TRUE potential of a PPU... That's right, there isn't one. They are barely worth discussion at this point in everyone but an Agiea employee's life.

It is simply a product that can't really be used at this time. It should not have been launched. At least not yet.

 

[Terra]

Yea, yea, yea... PPU this and PPU that. Give me ONE app that decently showcases the TRUE potential of a PPU... That's right, there isn't one. They are barely worth discussion at this point in everyone but an Agiea employee's life.

It is simply a product that can't really be used at this time. It should not have been launched. At least not yet.

So because there is nuch more untapped potential in the PPU than even CellFactor can show they should not have relaeased it?
I guess ATI was very bad in releasing the 9700 when there where no DirectX9 out?
Or NVIDIA are morons for releasing the G80 before DX10 is ut?

It's quite simple actually..no hardware, no support....

Terra - What a lame point

 

[Terra]

Or could develop into CPU, GPU, GPU or CPU, GPU, GPU, GPU

Nice, you still cling to ATI/NVIDIA unprocven soultion.
A slolutuon they don't even dare release a tech-demo for...
(Wich I think is a fiirst for those 2 companies)
Why would I want more CPU's/GPU'S?
I have a CPU to orchestrate it all.
A PPU to calcualte physics.
A GPU to render graphics.
A soundchip for sound and a nic for netwok.

I don't need a new mobo for this.
You solution is both unproven and more costly.

Terra - Nice PR though

 

[quadnad]

I don't think hype will, I think no reason to purchase it could.

I'm still laughing all the time about people writing things about how the "cpu sucks at physics calculations" meanwhile I don't see a need to simulate 100,000 boulders rolling down a hill. I still don't see why people wouldn't prefer just to offload the calculations to the generally unused second core of a CPU. Far more people have dual core processors, and they're much cheaper than the PPU.

I know it won't be able to simulate those boulders rolling down a hill, or 50,000 bricks flying around, but apparently CPU physics is more than enough to add the immersion shown in the Crysis demos (and I'm sure is capable of more), and that is what I'm looking for. I want immersion, not a novelty.

Once game physics progresses (I see this as a progression, like the graphics realm is) then I think PPUs will make more sense. They'll be less expensive, graphics solutions will be more able to render all those objects flying about, etc.

 

[Terra]

I don't think hype will, I think no reason to purchase it could.

Like with all other hardware, point being?

I'm still laughing all the time about people writing things about how the "cpu sucks at physics calculations" meanwhile I don't see a need to simulate 100,000 boulders rolling down a hill.

Physcis is more than boulder and boxes.
AGIEA's PhysX API is already multicore, but still it can bog down the CPU.
Destructable cloth are not just possibile on toays CPU...Unless you like 2 FPS?

I still don't see why people wouldn't prefer just to offload the calculations to the generally unused second core of a CPU. Far more people have dual core processors, and they're much cheaper than the PPU.

Due to the nature of physics calcualtion the CPU would be a piss poor contender.
But if you think the second core is powerfull enoguh to act like a PPU or GPU...then you seriously need to do some reading up...
I have a dualcore CPU and it get slammed to the ground when trying to run hardcore physcis.

I know it won't be able to simulate those boulders rolling down a hill, or 50,000 bricks flying around, but apparently CPU physics is more than enough to add the immersion shown in the Crysis demos, and that is what I'm looking for. I want immersion, not a novelty.

Lets see how fluids, cloth, friction, detructable architechture ect. get done in Crysis....
So far I have alreday posted videos that show funny physcis on LSD and bullet's doing nothing to the enviroment of Crysis...
The physcis in Crysis dosn't impress me so far...perhaps you have other video to share than the ones I have seen?
(Where bullets don't interact, trees break at 2 fixed points and the stub just dissapers when shot, tree's falling, acting like they are on LSD, and then suddenly get VERY immobile, explosions that casue no physcis damage to the surroundings ect...?)

Once game physics progresses (I see this as a progression, like the graphics realm is) then I think PPUs will make more sense. They'll be less expensive, graphics solutions will be more able to render all those objects flying about, etc.

You are so locked in thinking boxes, that you can't the the forrest for trees...
Again that is a VERY small part of the PhysX API...important yes...but far from the whole story...

Terra...

 

[quadnad]

Like with all other hardware, point being?

Point being that I answered the question the thread posed.

Physcis is more than boulder and boxes.
AGIEA's PhysX API is already multicore, but still it can bog down the CPU.
Destructable cloth are not just possibile on toays CPU...Unless you like 2 FPS?

To tell you the truth, destructable cloth and what can only be described as gloop ("liquid") aren't on my short list of things that help immersion, certainly not high enough to justify the cost.

Due to the nature of physics calcualtion the CPU would be a piss poor contender.
But if you think the second core is powerfull enoguh to act like a PPU or GPU...then you seriously need to do some reading up...
I have a dualcore CPU and it get slammed to the ground when trying to run hardcore physcis.

Trying to run hardcore physics? This is my point, I'm not saying that the CPU is as capable, I'm saying that it doesn't need to be. I'm not looking for the 100,000 boulders, I'm not looking for "destructable cloth," I just want some better physics and I think the CPU is capable enough to bring us to the next level with a core dedicated to this.

Lets see how fluids, cloth, friction, detructable architechture ect. get done in Crysis....
So far I have alreday posted videos that show funny physcis on LSD and bullet's doing nothing to the enviroment of Crysis...
The physcis in Crysis dosn't impress me so far...perhaps you have other video to share than the ones I have seen?

I think that the general environment in Crysis looks great, and the amount of physics that are in the game help to enhance it. So why do I need a $250 PPU to calculate friction? How is that going to enhance my gameplay in a way that my CPU can't, or at the very least to a level that helps justify the cost? I used Crysis as an example, and perhaps it wasn't the best, but we have not A. seen the final game or B. seen any other game even attempt using the second core on a CPU for physics. Thats why I think it should move in that direction.

I just want people to take a step back and focus on how any of this relatively new physics area is going to become more mainstream. I really think it needs to start with the power we've got, which is the CPU, and I'm sure with some proper code we can have some immersive physics taking advantage of multicore processors. I know it isn't close to what the PPU is capable of, but think about this: with a $250 barrier of entry, how will this tech evolve? Maybe if titles start appearing that use the Physx API or whatever the nVidia or ATi tech will be called to their full potentials then it will seem more attractive, but this hasn't happened yet because the tech is still in it's infancy.

 

[Terra]

To tell you the truth, destructable cloth and what can only be described as gloop ("liquid") aren't on my short list of things that help immersion, certainly not high enough to justify the cost.

First time I played PREY and met "GrandPa'" I thoght 2 things.
His clothes(cloth) looked bad(the downside f seeing what can be done, compared to todays games) and his hair looked bad(hair is physics too)
HAve you even seen the cellfactor cloth/fluids realtime?

Trying to run hardcore physics? This is my point, I'm not saying that the CPU is as capable, I'm saying that it doesn't need to be. I'm not looking for the 100,000 boulders, I'm not looking for "destructable cloth," I just want some better physics and I think the CPU is capable enough to bring us to the next level with a core dedicated to this.

You "want" better physics, but not the physics?
Sorry, you statement is a paradox, please refrase.


[QOUTE]I think that the general environment in Crysis looks great, and the amount of physics that are in the game help to enhance it.[/QUOTE]

The physcis in Crysis actually had the opposite effect on me.
I was dissapointed...and still am.

So why do I need a $250 PPU to calculate friction? How is that going to enhance my gameplay in a way that my CPU can't, or at the very least to a level that helps justify the cost?

Again, get in your head:
The CPU uscks at physics due to the massive parrellel dataflow
Something a CPU sucks at.
And you are trying to pin the price on 1 single segment.
Like saying: "Why dio I need a +$300 GPU for AA?"...see it?

I used Crysis as an example, and perhaps it wasn't the best, but we have not A. seen the final game or B. seen any other game even attempt using the second core on a CPU for physics. Thats why I think it should move in that direction.

Not even a quad-core CPU will be able to go anywhere near the physcis( collisions, rigid bodies, cloth, fluids, smoke, friction ect...) that the PhysX can deliver..why don't you get this?

Do me a favour, try and calulate how many core you would need for a CPU to do 530.000 convex-convex collision per second..and then come back.

I just want people to take a step back and focus on how any of this relatively new physics area is going to become more mainstream. I really think it needs to start with the power we've got, which is the CPU, and I'm sure with some proper code we can have some immersive physics taking advantage of multicore processors. I know it isn't close to what the PPU is capable of, but think about this: with a $250 barrier of entry, how will this tech evolve? Maybe if titles start appearing that use the Physx API or whatever the nVidia or ATi tech will be called to their full potentials then it will seem more attractive, but this hasn't happened yet because the tech is still in it's infancy.

Physcis isn't "new"...havok been doing it for years.
Limitied by the CPU.
Now that bottleneck is being removed...that is all...
Hell even Havok know that the CPU can't keep up with the PhysX...why do you think they launched HavokFX?

Terra...

 

[quadnad]

HAve you even seen the cellfactor cloth/fluids realtime?

Yes, I've seen cellfactor, and while I thought the cloth looked very cool I thought the "liquid" looked...terrible.

You "want" better physics, but not the physics?
Sorry, you statement is a paradox, please refrase.

No, you're ignoring my qualification: I'm saying that although the CPU is not as capable as a PPU, I believe it has far more potential to improve with the multicore processing.

The physcis in Crysis actually had the opposite effect on me.
I was dissapointed...and still am.

Ahh, but then you're carrying the banner for Aegia.

Again, get in your head:
The CPU uscks at physics due to the massive parrellel dataflow
Something a CPU sucks at.
And you are trying to pin the price on 1 single segment.
Like saying: "Why dio I need a +$300 GPU for AA?"...see it?

I disagree; graphics cards have been around for a long time, and have evolved from originally being done on the CPU. The CPU eventually reached it's limits, so the GPU was created. I don't believe the CPU (specifically multicore CPUs) have reached their limits, and I think this is an issue for the acceptance of the Physx card. I know the CPU isn't as powerful, but I don't think it's usefullness has been exhausted. To me, justifying a $300 video card is far easier than a $300 PPU because I can see the benefits with video accellerator support accross the board. A $300 videocard does a shitload more than just antialiasing...

Not even a quad-core CPU will be able to go anywhere near the physcis( collisions, rigid bodies, cloth, fluids, smoke, friction ect...) that the PhysX can deliver..why don't you get this?

You don't have to act condescending about it, and I do understand that, as is evidenced by what i've said in this very thread:

• This is my point, I'm not saying that the CPU is as capable, I'm saying that it doesn't need to be.

• I know it isn't close to what the PPU is capable of...

Physcis isn't "new"...havok been doing it for years.
Limitied by the CPU.
Now that bottleneck is being removed...that is all...

Physics as a concept in games isn't new, but physics as a focus in gaming certainly is. I still don't think that the usefullness of the CPU has been exhausted; I just don't believe that cloth and "liquid" are worth $250. I think it makes the most sense to start by using what it can do, and then move from there rather than being convinced by companies like Aegia that what we need right now is a PPU.

HavokFX was designed to improve the graphical effects, not the "interactive physics." Different ballgame, as you well know.

 

[SuperGee]

I think that the general environment in Crysis looks great, and the amount of physics that are in the game help to enhance it. So why do I need a $250 PPU to calculate friction? How is that going to enhance my gameplay in a way that my CPU can't, or at the very least to a level that helps justify the cost? I used Crysis as an example, and perhaps it wasn't the best, but we have not A. seen the final game or B. seen any other game even attempt using the second core on a CPU for physics. Thats why I think it should move in that direction.

I just want people to take a step back and focus on how any of this relatively new physics area is going to become more mainstream. I really think it needs to start with the power we've got, which is the CPU, and I'm sure with some proper code we can have some immersive physics taking advantage of multicore processors. I know it isn't close to what the PPU is capable of, but think about this: with a $250 barrier of entry, how will this tech evolve? Maybe if titles start appearing that use the Physx API or whatever the nVidia or ATi tech will be called to their full potentials then it will seem more attractive, but this hasn't happened yet because the tech is still in it's infancy.

Yey sure but consider this. That Crysis is a specific game with other needs for its design in requierments. There are a lot of Physics Features and every game design, weights wich one are more important or less for there game and implement some off those.
Crysis will not be the Destructable enviorment game. Yes cool jungle with some interaction never seen before. But PhysX is more. For the Crysis Physics needs a DC will suffice. Other games have the option to put other more heavy PhysX features in where DC CPU comes short.
Like Graw went for a bitty slapped on eyecandy effect physX.
UT2007 will put a lot more optional effects Physix and a bit gameplay physx in a few maps hopefully.
CF revolution will pack physX to the max. As it will be a very Physx driven game.
it will be also a very different game then crysis. Where this flood of Physics makes more sense. It's intergral to the game.
Surely every dev is looking what could this PhysX power mean for there Project. And what Physics features could enhance to there game. What it will cost in price time and resources and risk.

But there is more.
Time, and inhouse expertise is also important.
Crysis is a class AAA title. They make a inhouse game engine wich will be nativly supports there Gamedesign requierments for Crysis. Because they got the time and money to do it from there publishers. Also with that the time to make a inhouse software Physics API wich supports this Game design for Crysis nativly to complete the Game engine in featureset.

Other Dev's with a more limmit or aiming at a differnt kind of FPS style of game. The use of already avaible middleware High level API's and tools. They often fullfill the Game design needs for there projects.
Plus today it gives the option for dedicated hardware acceleration support. Wich makes more Heavy PhysX possible option.
Whitout the need of a expensive physics guru in the team wich has to deal with low level physics at CPU level, needs a lot of time and money and resourse to do his thing.
just a choice a dev makes.

Thats the cost of using what is avaible. Guru's for crancing out everylitle computing out of excisting hardware. cost a lot of time and money and a expensive Guru specialist in optimizing.

Don't know for sure if they do inhouse or middware Physics for Dualcore. But keep in mind. Crysis will use a subset off all what can be done with physX. they use what Crysis need the most. While other Game projects see now the oppertunity to put new or larger scale PhysX for those Physics Features that requier a lot of Physics power.

I disagree; graphics cards have been around for a long time, and have evolved from originally being done on the CPU. The CPU eventually reached it's limits, so the GPU was created. I don't believe the CPU (specifically multicore CPUs) have reached their limits, and I think this is an issue for the acceptance of the Physx card. I know the CPU isn't as powerful, but I don't think it's usefullness has been exhausted. To me, justifying a $300 video card is far easier than a $300 PPU because I can see the benefits with video accellerator support accross the board. A $300 videocard does a shitload more than just antialiasing...

* This is my point, I'm not saying that the CPU is as capable, I'm saying that it doesn't need to be.

The market doesn't work like that. The Chicken egg thing way comes up very strongly. First between hardware and Devs. Then Devs to Gamers and Hardware to gamers.
The triangle of dependence.

Sure games now utilise for physics only a midrange Single core. That's enough for you. And enough for requerments from older already released Game projects.Dualcore spice it up a bit more. Devs will go for that to. chicken egg thing. Wich runs a lot smoother then new introduced hardware.
But to Break this Chicken egg think they must just put the hardware out. Like Voodoo did. like Geforce 3. EAX1/2 card for 3D sound.And DX10 will be.
If its avaible and a small target platform emerges Dev are going to support it.

But here the deal The Chicken egg between Hardware PPU and Dev's is no Problem.
PPU already aviable, so the first step. Dev pick it up pretty well 65+ game supporting it. Some already released. Now it waiting for more games. And the target platform will grow with the more games are coming wich attract gamers for buying PPU.
The Chicken egg thing.
While for basic Gameplay not much Physics is needed But Games Compete and sale depend how the game stands out and making use of the latest and gratest new hardware like. Dualcore, DX10, PHysX or havokFX can help a lot to stand out of the Pack.
The rest depends on a good execution of a nice next gen game design to make a good game fullfilling gamers expectation or even more. orginality help to.
PhysX can help with that so do DX10 and Dual core.

Physics as a concept in games isn't new, but physics as a focus in gaming certainly is. I still don't think that the usefullness of the CPU has been exhausted; I just don't believe that cloth and "liquid" are worth $250. I think it makes the most sense to start by using what it can do, and then move from there rather than being convinced by companies like Aegia that what we need right now is a PPU.

I agree that there is a overly focus on Physics due to the DC vs PPU vs GPU thing going on now that AI is running at the background.
Both Cloth and liquids are just some of the features of physx. Wich stands out as very PPU computive in Cellfactor Combat training.
But you forget that it is a limited PhysX game demo it doesn't let you see all wich is posible. Cellfactor Revolutions is not about Cloth and Fluids alone that just a part of all the PhysX features. And even a other few will not be used.

HavokFX was designed to improve the graphical effects, not the "interactive physics." Different ballgame, as you well know.

Thsi is just a good thing for ageia and a bad thing for HavokFX. Havok FX by not give the option to support Gameplay physics. its a Halv solution. Oh yes later they will, after two years or more if the market is ready. In between Cellfactor revolutions a true gameplay PhysX game have hit the streets. Pushing the devs due to competness to force to go for more Gameplay Physx.
Especialy if the gamers community pick this game up. That it become the must anticipated killer app.
this Half solution of Havok FX. Wich makes the price point isue a way more complex.
while a X1600 is a lot cheaper it's a lot weaker then other GPU like X1800XL or X1900. But forget about gameplay PhysX effects only.
Its just a choice.
Ageia have a strong selling point that it supports gameplay Physx now.

 

[R1ckCa1n]

This thread is evidence of how over hyped the PPU is on this forum. We have yet to see anything other than a few tech demos, promises of games, and a whole lot of nothing but hype.

[H]ard|OCP published some great articles around where physics is today and what it MIGHT mean in the future. For everyone in denial or simply missed the articles, look below.

Agei's PhysX Needs a Killer App 03-29-2006

Bottom Line:

Forgetting all the important business components like install base, distribution, technical support, game publisher and developer partnerships…..the list goes on and on. Without a couple of killer apps soon that create their own buzz and solidify a PhysX community, Ageia is simply allowing Red Team and Green Team time to play catch-up. And if Red and Green catch up, I think Ageia’s lifetime will be short or soon ripe for the picking.

Ageia PhysX & You - 05-07-2006

Bottom Line:

With the current titles available, PhysX delivers nothing worth the purchase price and PhysX has to be solely valued by the gaming content that it supports. Sites that are dragging the framerate wars into this are not telling the gamers what they need to know. Sadly this is exactly what I expected. PhysX is not about framerates, synthetic benchmarks, or PR catfights; PhysX is about the gaming immersion and experience it provides you. PhysX will live to be judged again another day as we hopefully see games that truly take advantage of the promises that Ageia makes about their hardware. Until then, save your PhysX pennies because the cards are not going to be getting any more expensive, unless of course they go the way of the Dodo. We will be exploring this topic more as PhysX content is released.

Effects Physics & Gameplay Physics Explored - 07-10-2006

Bottom Line:

There is a difference between effects and gameplay physics and having dedicated hardware acceleration for one or the other or both can lead to an improved gameplay experience. The current GPUs and PhysX processor are capable of both effects and gameplay physics acceleration although gameplay physics not being taken advantage of. Accelerating effects physics is the logical place to start moving towards utilizing full gameplay physics acceleration. Effects physics will not impact the gamestate; therefore it can be performed on single player and multiplayer games right now to provide a more immersive experience for one or more players without crippling a title’s install base. In the end though, accelerating both effects and gameplay physics have the potential to evolve PC gaming. It all comes down to the game content developers and what support they implement for hardware physics acceleration as this market moves forward. We feel that 2006 is simply the re-introduction of gaming physics as game physics have been around for years now. This is the year we are introduced to the idea of pure hardware acceleration, people talking about it, the technology being hashed out, deals being made, and the industry learns what is possible and probable. There is already one dedicated hardware processor on the block; Ageia with their PhysX processor. On the GPU side of things we believe that Windows Vista’s DX10 and DX10 GPUs will be the real start of physics acceleration on the PC using GPUs. Even though today’s GPUs are capable of effects and gameplay physics acceleration DX10 GPUs are really better suited for them. Hopefully 2007 and its released gaming titles will give us a much better look at just how hardware enabled physics will push gaming evolution.

What is the common theme here?

edit: I left the ATI article from [H]ard|OCP so Terra's head wouldn't explode