Maybe we can use the term "Half Physics" ?

[Brent_Justice]

Instead of calling HavokFX's method "Fake", or "Emulation" or what have you.

Effects Physics are still physics, period. It isn't fake, it isn't emulated, it is real physics. It just isn't as in-depth as gameplay physics. But it is very real and very much hardware accelerated physics.

Call it "Half Physics", or honestly just call it by its real name, "Effects Physics".

I don't see any reason to attack it for what it is. Effects Physics are better than NO Physics

In the end whichever way you go is up to you, let's not flame others for what decisions they choose to make. The flames in here are getting a bit heated and I'd like it to calm down a bit, I don't see any reason for it. Right now to much is uncertain with all this, there are different ways of going about Physics Acceleration and the more choices we have the better for us the consumer. It will all work itself out in the end, no need to get worked up about it. Keep an open mind, sit back, and see who wins.

 

[MrGuvernment]

they should have:

Effects Physics
Interactive Physics

 

[WickedWeasel]

Firing a flame thrower against a wall in one of the demos, can the ATI physics setup cause the flames to "bend" around the corner, or will the cpu do the physics?

i.e., can the ATI physics and a Dual Core CPU do the interactive physics, or is it entirely limited to effects physics?

 

[Brent_Justice]

technically nothing is stopping the GPU from doing interactive physics, the hardware is capable

it all comes down to the software and program interfaces, I know for a fact ATI is working with ways for developers to have direct access to their hardware GPUs so that the game developer can decide what they want to do with physics regardless of what physics engine they use, this type of work is in progress

as i mentioned above we need to give this more time, it will work itself out, it is way to early now to jump on any platform for this, there is no current compelling reason to, I firmly believe in hardware physics acceleration, but it is way to early right now to take a stance on anything, we need to be open to all the possibilities, and above all, we need games that make us want this hardware acceleration

 

[SuperGee]

I got a name for it.
Fire and forget Physics.

Do GPU have a Pixel shader opcode to Write Back computation result to Main system memory. If not doesn't matter if PCIE is bi directional. Till ATI/nV implement such thing a Opcode that does so in there Pixelshaders. Or Unified shaders.

Then They could go all tha way for Havok FX if they support those Physics to.

 

[LuminaryJanitor]

I've got a question...

If, as so many people are saying, it's impossible for the GPU to send data back to the CPU, then how does a computer take screenshots? If a GPU can process some texture and geometry data and return a 1600x1200x32 bitmap, why can't it process some physics data and return a 1600x1200 array of 32-bit elements? Even if it's horribly inefficient, that's still a far cry from the "physically impossible" most people are claiming.

 

[MrNasty]

I've got a question...

If, as so many people are saying, it's impossible for the GPU to send data back to the CPU, then how does a computer take screenshots? If a GPU can process some texture and geometry data and return a 1600x1200x32 bitmap, why can't it process some physics data and return a 1600x1200 array of 32-bit elements? Even if it's horribly inefficient, that's still a far cry from the "physically impossible" most people are claiming.

If you want to try and write code that extracts meaningful information from a rendertarget be my guest - it's anything but easy and has the downside of requiring the CPU to piece back what's happening from the final data it's been given. Not a great way of doing it to gain performance

It's a simple efficiency problem with current GPU architecture - I have no doubt DX10 cards will be a whole new kettle of fish.

Good analogy: You have a 6-gear drag racer that's going for a world land-speed record. Now try and put it in reverse - complicated and far away from being as quick as before

 

[cyks]

calling effect physics real is like calling a montior a real computer.

 

[ScYcS]

calling effect physics real is like calling a montior a real computer.

...and paying money for ANY PPU right now is the same as buying a sports car when there are no roads. It's shiny, it's new and you can adore it. But driving it you can not.

 

[MrNasty]

...and paying money for ANY PPU right now is the same as buying a sports car when there are no roads. It's shiny, it's new and you can adore it. But driving it you can not.

Very good point - but why the need for the rolling eyes?

 

[Jason711]

i prefer the term psuedo-physics.

my problem is that ati/nvidia is trying to spin their "effects physics" as something better than gameplay physics. when it isnt and never will be.

*calming down* .....

 

[ScYcS]

Very good point - but why the need for the rolling eyes?

Because i just don't understand why people would call Havok "pseudo"...if an apple fall from a tree, it's physics that makes it fall down. If the apple bursts on the ground, it's physics. If pieces fall all over from the shattering, it's physics. It's all physics. Only because one physics processor does calculate different (simple?) physics, it's not to say it's not physics...

 

[cyks]

...and paying money for ANY PPU right now is the same as buying a sports car when there are no roads. It's shiny, it's new and you can adore it. But driving it you can not.

Why did you quote me?

 

[ScYcS]

Because it was directed to you. Why can't we call effect physics real? Only because it's effects? Newsflash! Effects are part of physics in a game and therefore "real".

This forum is called "PPU/Physics Processors". Everytime i go into this section of the forum (mind you, i AM interested in this part of computing), all i see is praise for the AEGIA PPU and bashing all other approaches. Maybe we should simply call this subforum "AEGIA PPU" and be done with it.

Honestly, i think that most people here that bash the ATI approach are people who are in fear that they bought a $300 piece of hardware (that does nothing at the moment?) that all of a sudden can be had for less than $100 from a different company.

Why not simply accept that there are and will be different apporaches and see how they turn out instead of bashing one approach before it even comes to the market? Is it fear? Is it brand loyalty? I just don't get it.

If i see an Intel processor performing better or equal to AMD for the same price or less, I'll buy it. Same vice versa. Same with ATI vs. nVidia and you know what? Same with AEGIA and Ati.

If ATI comes up that is similar in performance for less money, then hey!!!! I'll buy it. Great for consumers, bad for AEGIA "pioneers" (or fans).

 

[LuminaryJanitor]

If you want to try and write code that extracts meaningful information from a rendertarget be my guest...

I'm not suggesting actually using the print screen function to extract data from the GPU...

It just seems to be the general consensus around this board that there exists no data path from the shader units to the CPU, making GPUs incapable of interactive physics, and the fact that you can take screenshots seems to be a direct contradiction of that.

And even if the transfer rates are laughable, isn't this still better than nothing? I would have thought a GPU could get away with far less communication with the CPU (compared to a PPU). Most of the time it could treat the physics purely as effects, and render the objects straight away. It'd only need to notify the CPU about a small number of interactions, like something hitting the player, or an explosive barrel involved in a collision. I guess you'd also need a bit of info on each collision for the sounds, but not much (I'd think the positions could be pretty vague for the purposes of audio).

 

[LuminaryJanitor]

if an apple fall from a tree, it's physics that makes it fall down. If the apple bursts on the ground, it's physics. If pieces fall all over from the shattering, it's physics. It's all physics.

In theory, on a PPU, you can build a house out of individual bricks and have it act (and more importantly, smash ) just like a real house.

On a GPU, build the same house and you can walk straight through the wall, 'cause the GPU can't inform you that there's something in your way. That's not physics. That's just fucking retarded.

Same with AEGIA and Ati ... If ATI comes up that is similar in performance for less money, then hey!!!! I'll buy it.

If ATI came up with a real PPU, or modified their GPUs to act like real PPUs, then you'd have a valid point. But this is nothing like the competition in the CPU/GPU markets. They're not equivalent products, and furthermore, there isn't room for two companies selling incompatible solutions. It's not just a bunch of ATI-bashing !!!!!!s - the Havok solution is inherently crippled, almost everyone here can see it, and no one wants to see an inferior product win this through marketing alone.

This isn't another VHS vs Betamax. This is more like VHS vs Books on Tape. The choice should be obvious.

 

[ScYcS]

If that's the case, I suggest we completely rename this forum to AEGIA and move it over to their website.

 

[cyks]

Because it was directed to you. Why can't we call effect physics real? Only because it's effects? Newsflash! Effects are part of physics in a game and therefore "real".

This forum is called "PPU/Physics Processors". Everytime i go into this section of the forum (mind you, i AM interested in this part of computing), all i see is praise for the AEGIS PPU and bashing all other approaches. Maybe we should simply call this subforum "AEGIS PPU" and be done with it.

Although much of what you have to say borders on irrational flaming, I feel that you miss the more important issues and actually need to be talked to.



PhysX is a new technology (PPU) with new potentials. Effect physics is based on old technology (GPU) with limited potential.



Ageia clearly wants games to evolve into new dimensions of gameplay. Nvidia & ATI want to sell more and more GPUs.



Physics is all about interactions (scientifically). PhysicX offers interactions-- Interactions between gameplay, the player, and the physics. Effects physics doesn't. For example: when I play UT2004 and shoot down a flying opponent, [real] physics take place-- the jet explodes and begins to fall. As this is happening, I interact with the jet even further (the rules of the game slightly change, I now have a new opponent- gravity, inertia, the jet etc...) Because it is real physics, that jet has the possibility to kill me or my teammates, so we either try our best to get out of the way, or we shoot the falling jet even more to force it to explode into smaller, less damaging parts.



If all of the former were effects physics, there would be no further interaction, just additional eye candy. If the latter was done with effects physics, there would be no reason to break it up into smaller pieces. With effects physics, the jet can fall on you and nothing would happen. The debris can pierce your player, but he/she wouldn’t lose health. You believe that to be real? Pretty much, all interactions between the player, the gameplay and the physics end when effects physics takes place.



If you think I am here just to praise the name of “AEGIS?” you are mistaken. I see effects physics and Physx as two totally different aspects in gameplay. Although I hope that one day both will be integrated into a single solution that offers amazing graphics with the in-depth interactions between the player, the game, and the physics, meanwhile I must not become a tool to Nvidia and ATI and allow them to decide for me what in game physics really is.



http://www.hardforum.com/showthread.php?t=1063010



Since you claim to not quote me on accident, I will address your previous post….



As far as I can tell, there are far less applications for effects physics than physx as of now. And if you think you are telling me something new when you tell me there are few applications for physx to begin with, you are, again, mistaken.

If that's the case, I suggest we completely rename this forum to AEGIA and move it over to their website.

And if what you say is the case, we should just move you over to a Nvidia or ATI webpage... two can play your silly OT game.

 

[ScYcS]

This was not meant as flame bait at all. But i am slowly developing a gripe against this forum, more so against people who own an Aegia PPU. Nobody knows (except maybe some ATI engineers) what the ATI solution is capable of. Everything is really just speculation at this point. And still yet, it get's bashed by the people who bought a PPU from Aegia. What is the point?

I can understand that we talk about the solutions from both companies and even critizie them in certain ways, but i totally fail to see the reason why people already, without any knowledge of the real product, say that it's not a real physics card.

THAT is beyond me and that is something that makes me irate about this forum and the people here sometimes.

 

[cyks]

I just don't get it.

QFT

 

[ScYcS]

See, your QFT is such a typical answer for this forum. This is really getting worse than any nVidia vs.ATI vs.3dfx (back in the days) or Intel vs. AMD.

 

[Menelmarar]

Although much of what you have to say borders on irrational flaming, I feel that you miss the more important issues and actually need to be talked to.

Everything he said was rational and valid.

PhysX is a new technology (PPU) with new potentials. Effect physics is based on old technology (GPU) with limited potential.

Effects physics on a GPU is old technology? What? It's not even out yet. Put down the pipe.

Ageia clearly wants games to evolve into new dimensions of gameplay. Nvidia & ATI want to sell more and more GPUs.

Yes, because clearly Ageia is not a company who wants to turn profits, they are a righteous beacon of light in the corporate world! /sarcasm

Physics is all about interactions (scientifically). PhysicX offers interactions-- Interactions between gameplay, the player, and the physics. Effects physics doesn't. For example: when I play UT2004 and shoot down a flying opponent, [real] physics take place-- the jet explodes and begins to fall. As this is happening, I interact with the jet even further (the rules of the game slightly change, I now have a new opponent- gravity, inertia, the jet etc...) Because it is real physics, that jet has the possibility to kill me or my teammates, so we either try our best to get out of the way, or we shoot the falling jet even more to force it to explode into smaller, less damaging parts.

Your right in that in-game physics at this time is mostly about the interactions between objects. Using your example of UT2004. When you kill someone their body flops against walls, or tumbles down stairs or a ramp. This is the body interacting with the environment in a realistic manor and has no effect on the player. Does that mean it isn't real physics, no that would be retarded. There are real calculations going on here to produce those effects. Effects physics are real physics calculations between objects often taking into account velocity, weight, shape, etc. There is nothing pseudo, fake, false about this.

Interaction with the player is only one dimension and it's absense doesn't mean that real physics calculations aren't taking place. It's much better to see a player fall limp over the edge of a platform than to lay stiff over thin air. Gameplay was improved...

Interactive physics AND effects physics are real physics calculations and both improve gameplay, but to different levels.

If you think I am here just to praise the name of “AEGIS?” you are mistaken. I see effects physics and Physx as two totally different aspects in gameplay. Although I hope that one day both will be integrated into a single solution that offers amazing graphics with the in-depth interactions between the player, the game, and the physics, meanwhile I must not become a tool to Nvidia and ATI and allow them to decide for me what in game physics really is.

You clearly don't grasp the concept of physics, so maybe you should take note of what Havok, ATI, and nVidia are trying to accomplish. Again, interaction with the player is only one dimension and does not invalidate any physics calculations done from the interaction between objects.

As far as I can tell, there are far less applications for effects physics than physx as of now. And if you think you are telling me something new when you tell me there are few applications for physx to begin with, you are, again, mistaken.

Huh, far less applications of effects physics? What do we have on the market right now for accelerated physics games. Let me count them.

Acclerated Effects Physics:
1. GRAW

Acclerated Interactive Physics:
Other than a LAN only multiplayer technical demonstration. I haven't seen any.

hmmm oops....what? You say that I am mistaken for saying there are few applications for PhysX. Can you show me more? Please do.

 

[wollyka]

Call it "Calculated" Physics and be done with it. lol (just kidding)

 

[MrNasty]

I'm not suggesting actually using the print screen function to extract data from the GPU...

It just seems to be the general consensus around this board that there exists no data path from the shader units to the CPU, making GPUs incapable of interactive physics, and the fact that you can take screenshots seems to be a direct contradiction of that.

And even if the transfer rates are laughable, isn't this still better than nothing? I would have thought a GPU could get away with far less communication with the CPU (compared to a PPU). Most of the time it could treat the physics purely as effects, and render the objects straight away. It'd only need to notify the CPU about a small number of interactions, like something hitting the player, or an explosive barrel involved in a collision. I guess you'd also need a bit of info on each collision for the sounds, but not much (I'd think the positions could be pretty vague for the purposes of audio).

I know you didn't actually mean using prtsc to do anything, but essentially this is what would happen, as the results of pixelshaders can only be applied to frames so any calculation of what has happened as the result of the pixelshader has to come from within the GPU unless you export the whole frame - and then you have to figure out what to do with that frame, the solution of which will eat CPU cycles wholesale.

So yeah interactive physics with DX9 GPUs is possible, but it would cost a lot of memory and CPU time - most probably to the point of being inefficient.

To the guy trolling about how this is an AGEIA forum now - all I see on this forum are people who have knocked AGEIA at some point, and usually with no technological understanding or basis. In fact I remember hundreds of posts about 3 months ago which stated ATi's solution will conquer all and that it was 20x faster, and still many more that just liked the idea more because they didn't quite get what it meant (they thought it was the same).

ATi's turned its usual marketing knob to 10 and is jumping in wholesale - if AGEIA's product was hyped (and I believe it was in the eyes of the general enthusiast population) then ATi's is OMGWTFBBQ hyped - to the point where they're not even being transparent about what hardware they're using to do all the physics.

 

[merlin704]

See, your QFT is such a typical answer for this forum. This is really getting worse than any nVidia vs.ATI vs.3dfx (back in the days) or Intel vs. AMD.

Sorry man, but I am going to have to agree with all you say. I've been a member of these forums since 2001 and I've been there for all the Vs. battles and this is just a repeat of the same ol' schematic. It's nice that we have a forum for Physics Processors but when I come in here, all I see are AEGIA threads. God forbid another solution for physics is spoken because it will just get bashed.

When Aegia was first announced, I was like " Yeah I want one", glad I didn't buy one though. I played the GRAW demo and no physics card can make that a good game.

Coming to this section of the forums is almost like being force-fed religion!

 

[cyks]

You say that I am mistaken for saying there are few applications for PhysX. Can you show me more? Please do.

Am I surprised that another member misread what I typed?


I know there is more explanation to be done (you guys misread a lot, so I must adapt my rhetoric to your type of audience), but I don’t have the patience to deal with type of behavior as of now.



Let me just leave it at the following:

: The advantage here is accelerated physics can be handled by already existing hardware. Havok has stated that a GPU can only do effects physics, meaning that it will not affect the environment with things like flying shrapnel. You will see the shrapnel, but it won't be able to hit players, go through walls, deform the terrain...etc.

Based on your logic, Already existing hardware = New



Shrapnel not causing damage, miraculously flying through walls = Real Physics

 

[ScYcS]

"I know there is more explanation to be done (you guys misread a lot, so I must adapt my rhetoric to your type of audience), but I don’t have the patience to deal with type of behavior as of now."

<edit> you know what? I had a comment for this but i decided to delete it and just leave your quote standing as is. I think it shows enough without comment.

 

[cyks]

"I know there is more explanation to be done (you guys misread a lot, so I must adapt my rhetoric to your type of audience), but I don’t have the patience to deal with type of behavior as of now."

<edit> you know what? I had a comment for this but i decided to delete it and just leave your quote standing as is. I think it shows enough without comment.

So your post was for what purpose? Although I am glad I finally made myself clear, your original agenda was nothing above any implication you may have thought I was trying to get at with that last sentence that you have just quoted in vain.

Lets all agree that this thread is pointless-- there is nothing to dispute. We can continue to misquote each other and sidestep the current facts/speculation for our own facts/speculation, but it only brings out the worst.

 

[ScYcS]

Amen to that.

And back to what i originally wanted to point out in first: Let's wait and see what the future brings. Be it AEGIA or Ati or nVidia or whoever. Without bashing any of them.

 

[MrGuvernment]

I know ATI / NV dont have their products out yet but from all of the information THEY have released does that not give us enough direction as to how they are incoporating their "engines"

i dont own a PPU card and dont plan to anytime soon, sure i got dual SLI 16x slots, but i just ordered my x1900 XT and that will keep me happy enough.

i play alot of games, like UT2004, or other games where if a barrel blows up beside u - u get injured.

or a car gets grenaded and blows - you get injured from the car, not the nad... is that not interactive physics... if so, it has been around for a VERY long time already........


time shall tell.

Althought i do like ATI's implementation of this, of being able to use an 2nd video card for it so you dont need to buy a single card that can only do one thing.

 

[Bicarus]

I have a BFG Physx Card and damn proud of it. -bash me cause I have one - I don't care.

I do not have a Crossfire setup - nor an SLI setup. And even if I did I would still purchase a PPU for the simple fact that I would rather have a seperate PPU than have a game use my extra GPU(s) to process Physics stuff.

Now perhaps one day (in two or so years) BFG will come out with a new breed of cards that impliment the PPU on the same card as the GPU. -it is bound to happen if it can be done.

It just seems to me that there is anomosity towards PPU's in general because they cost money - and in 2007 or so ATI & Nvidia will release their own version of GPU phyiscs processing drivers and or cards. But like I stated above I just do not want my GPU to do complex physics calcs when those cycles can be used to get a bit better fps/graphics, etc.


So please, just relax - it will all be ok. and hopefully the PPU and non PPU users can live together in harmony.

 

[ahlknysvgmwso]

I'm sick of physics. Take it away! *cries*

 

[Firebat]

I'm sick of physics. Take ot away! *cries*

QFT, It was cool for about 3-4 minutes.

Honestly, if everything had as good physics as oblivion, I would be all set.

 

[Jason711]

QFT, It was cool for about 3-4 minutes.

Honestly, if everything had as good physics as oblivion, I would be all set.

im glad your not in charge.

 

[Terra]

Instead of calling HavokFX's method "Fake", or "Emulation" or what have you.

Effects Physics are still physics, period. It isn't fake, it isn't emulated, it is real physics. It just isn't as in-depth as gameplay physics. But it is very real and very much hardware accelerated physics.

Call it "Half Physics", or honestly just call it by its real name, "Effects Physics".

I don't see any reason to attack it for what it is. Effects Physics are better than NO Physics

In the end whichever way you go is up to you, let's not flame others for what decisions they choose to make. The flames in here are getting a bit heated and I'd like it to calm down a bit, I don't see any reason for it. Right now to much is uncertain with all this, there are different ways of going about Physics Acceleration and the more choices we have the better for us the consumer. It will all work itself out in the end, no need to get worked up about it. Keep an open mind, sit back, and see who wins.

Wasn't this better directed against NVIDIA/ATI/Havok?
They are the one responsible for this "mess"...not me

/offtopic
I will not be very much online the next few weeks.
Just fried my P4 2.4Ghz and my Asus P4T-E
But since they have been running like 24/7 since mid 2002, I got my moneys worth, out of them so goodbye old hardware...

Terra - I'll be back when my new Asus P5800 SE, Pentium D 940 and 4 x 512 DDR400 arrives...

 

[psychoace]

I have a BFG Physx Card and damn proud of it. -bash me cause I have one - I don't care.

I do not have a Crossfire setup - nor an SLI setup. And even if I did I would still purchase a PPU for the simple fact that I would rather have a seperate PPU than have a game use my extra GPU(s) to process Physics stuff.

Now perhaps one day (in two or so years) BFG will come out with a new breed of cards that impliment the PPU on the same card as the GPU. -it is bound to happen if it can be done.

It just seems to me that there is anomosity towards PPU's in general because they cost money - and in 2007 or so ATI & Nvidia will release their own version of GPU phyiscs processing drivers and or cards. But like I stated above I just do not want my GPU to do complex physics calcs when those cycles can be used to get a bit better fps/graphics, etc.


So please, just relax - it will all be ok. and hopefully the PPU and non PPU users can live together in harmony.

I don't think most of us hate PPU's or anyone who's bought one. Neither do I atleast hate people who have Ageia cards. It's your money you do whatever you want with it. I just hate the people who say that Ageia is the end all be all of PPU's and ATI and Nvidia are just a joke imitation slapped on in fear of Ageia. I think most people arn't going to go Ageia way is it's 300 bucks and takes up 1 pci slot which is very precious in a sli/crossfire setup.

 

[cyks]

Just fried my P4 2.4Ghz and my Asus P4T-E

Trying to OC the PPU?

 

[Terra]

Trying to OC the PPU?

No, it was just old age I guess.
Just finished testing the other parts in my PC, they all work fine.

Terra - 4 years running 24/7, mostly under load...I don't complain

 

[Skirrow]

These discussions are getting ridiculous. The fact of the matter is that BOTH the Physx and Havok API's do the exact same thing. The are both software. Both do rigid/soft bodies, liquid, cloth, the whole shebang and they BOTH have roughly the same performance. The only real difference is that Physx has a hardware card out.

Havok/ATI/Nividia announce that they can use GPU's to accelerate HavokFX, which is basically another version of the Havok API that uses GPU hardware, and everyone starts flinging shit.

All this crap about only Physx being 'true' physics is bull. Havok has the same functionality.

As for HavokFX only being limited to 'effects physics'. Give me a break.

There is no reason why HavokFX cant do everything that physx can. None whatsoever. It might be an idea for people to read up on these things before posting crap.

As for the people who say ATI/Nvidia are only doing this to sell more cards. These are the same people who buy a new gfx card every 6 months. They say they'll have to buy 2 cards just to get physics accelleration. No you wont. Just use your old one. Its not fucking rocket science.

Another thing. Both the Physx card and GPU's that will be able to handle HavokFX are pretty cheap anyway.

Let me see. A BFG phsyx card is £200. An ATI X1600 Pro is £130.

But which will i choose???? I'll buy the Physx.

I already have an X1600 sitting on a shelf doing nothing. Cant go wrong either way now can i.

 

[psychoace]

Here it's a little different Ageia's card is 300 bucks in America while the x1600 is around $100 so it's a much different decision.