STI stakes claims on physics

D

DocFaustus

2[H]4U
Joined
Sep 22, 2002
Messages
2,830
As linked from hardocp mainpage to techreport: http://techreport.com/onearticle.x/8887

This demonstration was more important as a statement of position and direction for ATI than anything else. NVIDIA has been rather enigmatic about any plans for physics processing, and has seemed to step out of AGEIA's way for the most part, welcoming the PhysX effort as another means of improving PC gaming. ATI is clearly ready and willing to take on AGEIA's PhysX card by using its GPUs for physics simulation, and the company believes that the more general programmability of its Shader Model 3.0 implementation in the Radeon X1000 series could make it a viable competitor there. There was talk of a CrossFire dual-GPU config in a split setup, with one GPU handling graphics and the other handling physics for certain games, and somebody present even suggested the possibility of a third GPU on a PCI card dedicated to physics processing. The folks from ATI seemed to like this suggestion.
We haven't heard many specifics yet about how ATI plans to go about exposing the physics acceleration capabilities of its GPUs to developers. One obvious path forward would seem to be cooperation with physics middleware vendors like Havok, but AGEIA has already made extensive inroads into next-gen games development by giving away licenses for its PhysX API. If ATI is serious about making this push, it may have to slog through an API war by pushing developers to use something other than PhysX API. We shall see.
Click to expand...
I snipped those two paragraphs because I do not like what I am reading. Here is why: I am excited about the prospects of a physics card and plan to support it though purchase. However, If ATI pulls some fucked up API war then everyone looses because your physics may only work with titles written for it. If they want to do support physics and make something compatable, fine, competition is good, but if they are not 100 percent compatible with this new technology, they will never see a dime from me again.

If I am incorrect in my interpertation please correct me.
 
DocFaustus said:
As linked from hardocp mainpage to techreport: http://techreport.com/onearticle.x/8887


I snipped those two paragraphs because I do not like what I am reading. Here is why: I am excited about the prospects of a physics card and plan to support it though purchase. However, If ATI pulls some fucked up API war then everyone looses because your physics may only work with titles written for it. If they want to do support physics and make something compatable, fine, competition is good, but if they are not 100 percent compatible with this new technology, they will never see a dime from me again.

If I am incorrect in my interpertation please correct me.
Click to expand...

thats what i got out of it as well.
 
lol
physics is physics

is not like there are different types of physics in this world
 
DocFaustus said:
As linked from hardocp mainpage to techreport:
If they want to do support physics and make something compatable, fine, competition is good, but if they are not 100 percent compatible with this new technology, they will never see a dime from me again.
Click to expand...
Ati has a history of bending standards their way. SM3.0 done right anyone? Maybe incompatibility can be solved by a 'hack'.
 
There is also the other uses for this as well...folding for one... And what if ATI was able to get a cross license for the same physics API? I don't see a big use for me...but I am sure for others it may make good sense to play around with it...

Now when do you think NV/ATI will make their first part to challange the Intel/AMD for CPUs? :)
 
Aztlan said:
lol
physics is physics

is not like there are different types of physics in this world
Click to expand...
I am not bitching about physics, I am bitching about an API.

Go back to school regarding that second comment
 
If I had to guess, I would think we will see vendor-specific API's for a short time, and MS or somebody else will release a version of DirectX or "OpenPL" or something similar (or both) and the physics-processor companies will be forced to jump on board. For the time being, though, if ATi releases a PPU (or enables physics processing on their GPU), we will probably have the PhysX API and the ATi API separate, and games will support one or the other.
 
could be a reason for the increase price for the x1300?

compatability, and the ability to act as a PPU, but for $100 less? if it works out to be able to process physic's as fast as the AGEIA's PhysX, then the card would be fairly priceless
 
Oh and BTW....mods please change title to "ATi" not "STI"
 
I dont think the X1800 series is near efficient enough at performing physics calculations to really compete with the PhysX chip. Their was a chart measuring Gromacs performance of the chip over on XS, it was only 3x faster than a P4. If the type of physics performance is similar (which I would expect) that is tens of times slower at physics than the PhysX chip.

Besides the fact that if I am spending a large amount of money on a graphics card...I wouldn't want half of my card being taken away for physics essentially giving me a cheap video card and an ineffective physics card in one.

The X1800 may be fine for accelerating todays physics, so is a regular CPU. The type of physics capable on the PhysX chip I imagine would make an X1800 piss its pants and run home to momma (like it does to any CPU). This is simply ATi trying to come up with reasons to sell their product, I doubt it will be a viable physics acceleration solution. Just because you can cut meat with a butter knife doesn't make it compete with a steak knife ;)
 
So AGEIA owns the right to do hardware accelerated physics? And I'm not sure I agree with the claim that the company has made "extensive inroads" with developers when the company's president said in a fairly recent interview that there's only going to be a handful of games that benefit from a PhysX board next year. For your average gamer that's probably going to amount to paying out over $200 to see slightly better frame rates and maybe slightly more advanced physics in 1-2 games next year.

And NVIDIA has made some noise about incorporating physics acceleration in a future next-gen. chip, so in all honesty I'm not sure how AGEIA is going to survive as a startup company trying to carve out its place in this niche market.
 
Whats the point of having cross fire setup when only one VPU will handle the graphics and the other VPU handle the physics? I can understand it for those gaming with low res (sub 1600x1200) but the type of gamer with a crossfire setup will be wanting high res.

I'd prefer having a seperate card for the physics.

As far as Ageia surviving I think that either ATI or Nvidia would buy them...or their hardware would be licensed by both.

And dont count out Matrox.
 
John Reynolds said:
So AGEIA owns the right to do hardware accelerated physics? And I'm not sure I agree with the claim that the company has made "extensive inroads" with developers when the company's president said in a fairly recent interview that there's only going to be a handful of games that benefit from a PhysX board next year. For your average gamer that's probably going to amount to paying out over $200 to see slightly better frame rates and maybe slightly more advanced physics in 1-2 games next year.
Click to expand...
Nope I never said that, In fact, I said that competition was good and commended them for that. However, if you read what I wrote again you will see that my beef is with the possibility of an API war. ATi isn't good at sticking to standards, we know that already. Unfortunately, the new crap they come up with instead of the standards normally sucks and makes a mess of everything.

I like rooting for ATi and I have supported them with past purchases, but I really believe they have to stop this "road less traveled" bullshit or make it so compelling that it becomes the way. Rewriting code for negligable performance increase is dumb, and in the case of physics, it would be rewriting code for a LOSS in performance.

Side note: I changed the typo in the title, but it doesnt reflect it on the thread listings.
 
DocFaustus said:
Nope I never said that, In fact, I said that competition was good and commended them for that. However, if you read what I wrote again you will see that my beef is with the possibility of an API war. ATi isn't good at sticking to standards, we know that already. Unfortunately, the new crap they come up with instead of the standards normally sucks and makes a mess of everything.

I like rooting for ATi and I have supported them with past purchases, but I really believe they have to stop this "road less traveled" bullshit or make it so compelling that it becomes the way. Rewriting code for negligable performance increase is dumb, and in the case of physics, it would be rewriting code for a LOSS in performance.
Click to expand...
Wait a minute.
What is your suggestion then? There is no standard API for physics, so developing your own is the only option, at this point.
 
It appears that AGEIA has made very good progress forging relationships with many partners of all different levels and has been busy making sure people understand how to make use of it. Seems if ATi hopped onto that and made it work with AGEIA's API then everyone (expecially the customers) would benifit.

ATi would still deliver strong graphics and light physics capabilities. Whereas if you bought a dedicated card for graphics and a separate AGEIA card for physics you would have a more powerful system. But most importantly, anything written to make use of physics acceleration would work on BOTH systems.
 
DocFaustus said:
Nope I never said that, In fact, I said that competition was good and commended them for that. However, if you read what I wrote again you will see that my beef is with the possibility of an API war. ATi isn't good at sticking to standards, we know that already.
Click to expand...

Not sure I agree with that comment.

Unfortunately, the new crap they come up with instead of the standards normally sucks and makes a mess of everything.
Click to expand...

I think you're getting ahead of yourself in this blame game. Neither ATI nor NVIDIA have really made any commitment to hardware physics acceleration. The Gromacs/folding stuff is just marketing at this point in time.

I'm honestly not sure what you're referring to with the whole road less travelled angle. Are there examples you're thinking of when writing this?

Edit: Though I'm essentially of the mind that when I shell out $400+ for a GPU I want it doing just that, accelerating graphics. I want its design, its die space, spent on logic that enables usable features.
 
DocFaustus said:
Seems if ATi hopped onto that and made it work with AGEIA's API then everyone (expecially the customers) would benifit.
Click to expand...
I'm no engineer, but it seems to me that, if we are really talking about enabling physics processing on existing GPU's, it would be exceedingly difficult if not impossible to get those GPU's to work with the AGEIA API.
 
John Reynolds said:
I'm honestly not sure what you're referring to with the whole road less travelled angle. Are there examples you're thinking of when writing this?

Edit: Though I'm essentially of the mind that when I shell out $400+ for a GPU I want it doing just that, accelerating graphics. I want its design, its die space, spent on logic that enables usable features.
Click to expand...
3Dc comes to mind.

If I am reading your edit correctly, I agree with that. I have no problems shelling out the cash for a topend vid card and a top end physics card to go with it. But when I do, I would hope it runs everything.
 
jebo_4jc said:
I'm no engineer, but it seems to me that, if we are really talking about enabling physics processing on existing GPU's, it would be exceedingly difficult if not impossible to get those GPU's to work with the AGEIA API.
Click to expand...

I don't see why. An API is a "Application Programming Interface" like OpenGL or DirectX. ATI hardware is totally different from Nvidia hardware, or S3 hardware, or Intel Graphics hardware, but both OpenGL and D3D work on all of them.

API's are not hardware specific, the MFG's have to make their drivers work with the API, not the other way around.

==>Lazn
 
jebo_4jc said:
I'm no engineer, but it seems to me that, if we are really talking about enabling physics processing on existing GPU's, it would be exceedingly difficult if not impossible to get those GPU's to work with the AGEIA API.
Click to expand...
So you are saying that they didn't add anything to the card for physics but instead are taking away from graphics processing power to perform these calculations? Hrmm, I need to learn more about that.
 
The cool thing about the ATI gpu being able to do physics processing is this.

Picture the scenario.

Its 2007. Your X1800XT is getting a bit long in the tooth now and has trouble with the latest games but ATI or Nvidia have just released their new cards with all the new gizmos and you decide to upgrade. So you buy a nice new card and enjoy smooth gameplay once again.

But wait. What to do with the trusty old X1800XT?. Then you remember ATI's drivers have support for their physics engine, GravATI (see what i did there? i bet they end up using that name :D) which allows cards to do physics processing.

Saves having to fork out for a new physics card AND your old card wont need throwing out.
That $$$ investment may last a few more months yet.

Sure it may sound a bit stupid at the moment. But a couple of years down the line it could make a lot of sense.
 
DocFaustus said:
3Dc comes to mind.
Click to expand...

In all fairness, I believe ATI made that open so anyone can add hardware support if they choose to. And it doesn't break regular use of normal maps since any sane developer is going to provide non-3Dc maps in their game. Honestly, I think that's a somewhat poor example to use to claim a company isn't following standards. Is NVIDIA's DST a DirectX spec? No, it isn't, but it's a feature that helps performance with shadow maps so why knock the company for supporting it?

Don't get me wrong, I'd love to see AGEIA succeed since I think physics have been so rudimentary for so long that your typical gamer thinks physics = ragdoll behavior. What's probably best for the PC game industry is to have Microsoft incorporate a physics API into DirectX (and Havok would probably 'win', IMO).
 
Lazn_Work said:
I don't see why. An API is a "Application Programming Interface" like OpenGL or DirectX. ATI hardware is totally different from Nvidia hardware, or S3 hardware, or Intel Graphics hardware, but both OpenGL and D3D work on all of them.

API's are not hardware specific, the MFG's have to make their drivers work with the API, not the other way around.

==>Lazn
Click to expand...
I'm sure you are right, but there is definitely an aspect of designing hardware to operate under a certain API. Otherwise, why couldn't Intel just design a driver to enable P4's to run DirectX?
 
DocFaustus said:
So you are saying that they didn't add anything to the card for physics but instead are taking away from graphics processing power to perform these calculations? Hrmm, I need to learn more about that.
Click to expand...
I don't know if I'm right or not, but my understanding is, you could slap a x1300 in a x16 slot, and run it as a (let's call it) a "general processor."
My impression is that this general computing power is a byproduct of the architecture, not the original intent of the designers.
 
John Reynolds said:
What's probably best for the PC game industry is to have Microsoft incorporate a physics API into DirectX
Click to expand...
QFT. I bet we see this sooner rather than later.
 
John Reynolds said:
Don't get me wrong, I'd love to see AGEIA succeed since I think physics have been so rudimentary for so long that your typical gamer thinks physics = ragdoll behavior. What's probably best for the PC game industry is to have Microsoft incorporate a physics API into DirectX (and Havok would probably 'win', IMO).
Click to expand...
Once again, I agree. I think we are getting to the same conclusion here. As long as everything works no matter your upgrade path. I am happy.
 
jebo_4jc said:
I'm sure you are right, but there is definitely an aspect of designing hardware to operate under a certain API. Otherwise, why couldn't Intel just design a driver to enable P4's to run DirectX?
Click to expand...

Actually it should be possible for a P4 to run DirectX, In the Quake years (up to quake3 I think) it was possible to run it in under openGL in software. Boy it was slow :)

What ATI need is to License the physx engine like Sony has done for the Playstation3, that way we will have a standard physics API.
 
jebo_4jc said:
I don't know if I'm right or not, but my understanding is, you could slap a x1300 in a x16 slot, and run it as a (let's call it) a "general processor."
My impression is that this general computing power is a byproduct of the architecture, not the original intent of the designers.
Click to expand...

There are developers experimenting with a form of ray tracing via Pixel Shaders, and I've heard of a media player that should handle real time video effects (ala ffdshow) also via PS. Even sound processing and image processing (like photoshop) could be done.

Actually I wonder why ATI and nvidia have a dedicated engine for mpeg2/divx/H.264 decoding. They should be able to accelerate it via PS/VS.
 
DocFaustus said:
As linked from hardocp mainpage to techreport: http://techreport.com/onearticle.x/8887

I snipped those two paragraphs because I do not like what I am reading. Here is why: I am excited about the prospects of a physics card and plan to support it though purchase. However, If ATI pulls some fucked up API war then everyone looses because your physics may only work with titles written for it. If they want to do support physics and make something compatable, fine, competition is good, but if they are not 100 percent compatible with this new technology, they will never see a dime from me again.

If I am incorrect in my interpertation please correct me.
Click to expand...

Well, if DirectX were to make a standard, a lot of the inconsistencies would be resolved, or atleast minimized.

Aztlan said:
lol
physics is physics

is not like there are different types of physics in this world
Click to expand...

There's only one kind of physics, but there are infinite ways of actually implementing it. There's only one polygon, but ATI and nVidia are different in how they display them, and neither displays them correctly (Unless you're using the professional cards).

At worst we'll have atleast two different physics API standards competing. Just like the situation with OpenGL and DirectX.


EDIT: BTW, i'm not too sure if it's really possible or practical to make a GPU run physics algorithms. There was a study before where an independent party used a GPU for processing audio data. They didn't actually use the whole GPU, they used the shaders to process it. I dunno just how much of that can be used in physics calculations, but it seems wasteful.
 
jebo_4jc said:
I'm sure you are right, but there is definitely an aspect of designing hardware to operate under a certain API. Otherwise, why couldn't Intel just design a driver to enable P4's to run DirectX?
Click to expand...
Stoly already hit on this, but you can software render graphics, before the Voodoo, (well technically the Virge3d was first, but it was more of a graphics decelerator than accelerator) this was the only way to do it.

Edit: in fact now that 3dFX is dead, there are now glide wrappers (emulators) to let you software render apps written for the Glide API. You could write one for OpenGL or DirectX too, but games that use them are so much more intense graphically than any glide game was, your CPU would choke.

The thing is it would be slow, just like physics can run on the CPU now, but it is only fast enough for ~30 objects at once. ATI said thier GPUs are good for ~300. Agea said the PPU will be able to do 30000. So while I could see this getting some use, the Agea PPU is much more optmized for physics and will still be the better option. (for now)

==>Lazn
 
Lazn_Work said:
Edit: in fact now that 3dFX is dead, there are now glide wrappers (emulators) to let you software render apps written for the Glide API. You could write one for OpenGL or DirectX too, but games that use them are so much more intense graphically than any glide game was, your CPU would choke.

The thing is it would be slow, just like physics can run on the CPU now, but it is only fast enough for ~30 objects at once. ATI said thier GPUs are good for ~300. Agea said the PPU will be able to do 30000. So while I could see this getting some use, the Agea PPU is much more optmized for physics and will still be the better option. (for now)
Click to expand...
This makes the most sense to me.

Here's hoping for OpenPL to come to the rescue soon before we see too many physics API's flood the market.
 
im not a big fan of one company dominating a whole category of API, like microsoft with DirectX, so i think ATi's physics push would help lower prices overall and allow developers to choose what they like
 
Big Fat Duck said:
im not a big fan of one company dominating a whole category of API, like microsoft with DirectX, so i think ATi's physics push would help lower prices overall and allow developers to choose what they like
Click to expand...
How would an arguably propiertary API benifit the end user in regards to price?
 
I think ATi is trying to highlight the versatility of the programmable parts of their gpu's more than anything. If they can make them into fairly good physics processors then they would make up the middle part of the PPU market. Ageia only seems to have one product and it is not cheap. Not to fault them, but as a small company it would be hard to have stratified products like in the CPU and GPU market. If I could take some of the load off of my CPU and get better performance for $150, I would be more likely to do that than drop $250 on another part for my computer... but then i'm a generally mid-range, not bleeding edge buyer.
 
If they ultimately make it a good versatile compatible card for the midrange to upper-midrange buyers, I think it would be well recieved.
 
Hmm, Guys's guess what...

Look up the "Open Dynamics Framework" .. it is an Open Source toolkit that Ageia fully supports on their PPU. While not quite a full API, if both ATI and Ageia support it, it would go a long way twords getting some standards set that a full API would have to support.

http://www.ageia.com/novodex_downloads.html

==>Lazn
 
Lazn_Work said:
Hmm, Guys's guess what...

Look up the "Open Dynamics Framework" .. it is an Open Source toolkit that Ageia fully supports on their PPU. While not quite a full API, if both ATI and Ageia support it, it would go a long way twords getting some standards set that a full API would have to support.

http://www.ageia.com/novodex_downloads.html

==>Lazn
Click to expand...
Thank you very much Lazn_Work. That is all that I am looking for and more. Open Source is a fantastic start.
 
tornadotsunamilife said:
Well if they can get it done right (physics wise) they could compete with Aegia's PPU ($250) with a X1300 ($150)
Click to expand...
IF indeed.

I'd be willing to spend $250 on a dedicated PPU if it would do a good job rather than buy an ATi card and get half ass Physics. Also a dedicated PPU would mean I could use either nVidia or ATi cards, no?
 
You must log in or register to reply here.
Back
Top