Conroe & Cell Factor Demo?

SuperGee said:
I'am talkin about what CPU do for
Crysis
PPU
and what CPU can mean for PhysX.

And the latest and grates Gaming CPU platform Conroe.

Games are CPU/GPU dependant. How much depends on wich game.
PPU has made a triangle so they called it.

I think CPU GPU is a Chian of setup the next Frame. FPS as frames Per second.
PPU triangle is more a looped chain. Or a chain with a loop
CPU PPU CPU GPU.

CPU must set up a lot more PhysX for offload computation. And compute this massive result back in the game logic wich setups the render offload to the GPU. Plus a lot of other things in between AI sound server etc.

The Time per frame is now split in three. where CPU part is made of two sup parts.

So i wonder a PPU game could be More CPU intensive. and also stress the GPU more.
Each part has less time to do there thing.

So would the most game accelerating CPU could benefit PPU. Oh yes I think there is some latency time lost in transporting the load to and the result back. So the Computing time is smaller.
And a DC can make a PC run smoother. Lot's of system threads.

Did i say your name?
 
There's something about this argument that I don't understand. Why is everyone saying that, with the release of PhysX, developers are going to start improving CPU physics?

It's like someone says "hey cool, this new PPU can do advanced physics processing", then someone else says "good, now developers are going to start doing advanced physics on the CPU."

I don't think the introduction of the PPU is going to make developers say "hey, we don't want this new chip to be successful, so let's figure out ways to make CPU physics just as good." If anything, the PPU is going to make developers move away from CPU physics, not towards it.

And please stop saying that a CPU is fine for physics. Yes, it's fine for today's physics. But, if developers decided to start exploiting the PPU, they would be able to do MUCH more. Even more than a dual-core, quad-core, or eight-core CPU. And what's easier? Trying to multi-thread the physics to do it on three cores, or just doing it on one PPU with much more headroom?

Like I said, now all we need is for Microsoft to come in a set a standard physics API (which they will most certainly do once Ageia gets some competition - such as from ATI or NVIDIA). Then, people without a PhysX will use part of the GPU for physics, and people with a PhysX can get the full performance from their GPU just for graphics.

I'm suprised at the amount of opposition to this major hardware advance. Was it like this when the first real GPUs were first introduced? Were there people saying "CPUs/framebuffers can do graphics just fine"?
 
HOCP4ME said:
I'm suprised at the amount of opposition to this major hardware advance. Was it like this when the first real GPUs were first introduced? Were there people saying "CPUs/framebuffers can do graphics just fine"?

They where actually.
3Dfx got a lot of flak and people stated that the CPU was sufficient...no need for dedicated 3D hardware.
Fast forward 10 years and we have the same situation...this time with the PPU :rolleyes:
What even more funny is that according to Steam survey the number of people with SLI/Crossfire is a "staggering" 0.77%...
No wonder ATI/NVIIDA is putting out this much PR about the GPU, becuase apparently the graphics alone dosn't boost sales very much...

Terra - And I suspect a lot of viral marketing too...
 
http://www.hardforum.com/showthread.php?t=1078795

Developer of Prey noting 3d graphics were going to be a standard back in 1997 I don't hear anyone claiming that right now for Physics cards.

Also it seems that Agiea has had the same amount of press as Ageia has maybe even more so to say ATI/Nvidia has been putting out so much advertisement to boast sales then your being a hypocrite because that is exactly what Ageia is doing but I mean that is the name of the game to make money.
 
psychoace said:
Also it seems that Agiea has had the same amount of press as Ageia has maybe even more

Actually, I would say that Ageia has had more press, because I've never heard of Agiea. :p

(jk man, I know you meant to say ATI/NVIDIA)
 
Starting my post off-topic and ending on topic, :) right now Ageia trying to sell physics card so you can display a piece of cloth in a demo game... Hmmm i wonder if there is no other way to put a similar piece of cloth in there without the PhysX. Wait a minute i think i saw a piece of cloth in quake 1 or 2 in a CTF game like 10 years ago.... :s

Physics is something you can´t see and although i recognize it's value i find it hard to sell to an entertainment audience.

In a racing simulation forum a long time ago i saw someone write something like this, "what if we could simulate the loss of aerodynamic efficiency when an F-1 racing car is in front of the other, or program the cars underperformance around the airflow instead of simple mathematic equations." Now that's where a PPU might radically change the world of gaming. I imagine they would need their own programming language and a whole set of knowledge that would allow programmers to model the world's movement through the PPU. But selling the concept to a 14 year old gamer who is saving his allowance and working sundays, it may be a bit harder. He wants to see GORE!!! he doesn't give a shit about the airflow.

So what has Ageia done, find an obscure game developed by a small company and launch a demo with all the physics programming you can put in it in say... 6 months time, that is Cellfactor. The thing is that Cellfactor showcases 90% physics programming and 10% physics hardware from what i have read. The complex solid object movement can be done by a normal GPU with a 3 fps loss!!!! The only things that bring things to a standstill are the cloth and the fluid.

So to the topic of the thread PPU vs. Conroe, uhm... in todays world yes, Conroe does the same, i don't know in 10 years time.
 
Quakes flag was pre-animated and thus was static and didn't conform to body interaction. Although Prince of Persia had a great cloth simulation with there game all in real time all not pre animated just it wasn't destructable.
 
Alien said:
In a racing simulation forum a long time ago i saw someone write something like this, "what if we could simulate the loss of aerodynamic efficiency when an F-1 racing car is in front of the other, or program the cars underperformance around the airflow instead of simple mathematic equations." Now that's where a PPU might radically change the world of gaming. I imagine they would need their own programming language and a whole set of knowledge that would allow programmers to model the world's movement through the PPU. But selling the concept to a 14 year old gamer who is saving his allowance and working sundays, it may be a bit harder. He wants to see GORE!!! he doesn't give a shit about the airflow.

Yes, that's why the PPU is having a hard time getting off the ground. People aren't going to say "hey, look at this, my car handles worse when it's behind another car!" That's also why the demos of PhysX so far have been of cloth and boxes, which are things you can see, but not the things the PPU was meant to do.

But it will get off the ground eventually. Too much commotion over the physics thing has been generated for it to just die. The interesting thing will be seeing how it happens. Will there be one "killer app" that launches it into popularity? Or will more Cell Factor-like games be made, gradually implementing more and more hardware physics? But it can't be done on a multi-core CPU, and some gamers just aren't going to give up GPU power for physics, so the PPU will at least always be an option.

The point is, however, that the PPU is much better at physics than a CPU. Whether the developers want to exploit that is up to them. But a lot of people say "the CPU can do physics just fine", but that's just because the PPU hasn't been nearly used to it's full potential. When it is used to it's full potential, regular CPU physics will seem like crap in comparison.
 
Yeah but the question really should be will there be a standard. I hope there will be because I would hate to have to choose my games out by what Physics accelerator drivers are supported. I hope also it isn't going to be a monopolized situation by Agiea. If they really wanted to make it better for the consumer everyone would follow the standard. CPU's follow a standard (x86 and soon x64), Hard Drives follow a standard, Cd/DVD burners follow a standard why make it more difficult for the consumer by limiting what they can play? Why not boycott them all until there is a standard used by all Physic accelerated hardware.
 
I say it again.
Very important and very all sayin.
Ageia is just starting. Market adoption and sucses depends on what those 65+ Games will bring in 1 or 2 years. Not CF Demo or graw and some patched games.
That just how it stands now, But very important in Ageia introduction fase.
Just 1 game out.
 
HOCP4ME said:
Yes, that's why the PPU is having a hard time getting off the ground. People aren't going to say "hey, look at this, my car handles worse when it's behind another car!" That's also why the demos of PhysX so far have been of cloth and boxes, which are things you can see, but not the things the PPU was meant to do.

I havn't had the time to look indepth at the SDK yet but this is what I spotted right on as possibilities with the PhysX :

  • AI sensor
  • Fluids
  • Joints
  • Surfaces(friction)
  • Raycasting(shadows)
  • Componds
  • Trigger events
  • And much more

IMHO opinion AGEIA has done piss poor marketing...or they have some "sinister" plan I know nothing about.
Cause the PhysX is WAY more than boxes and cloth.
But perhaps the are using their time on the only one that really matters: The develpers...
They must have done something right to covince +60 game developers...
Then it really dosn't matter what us consumers think, if it catches on at the devs, that will be the future...

Terra...
 
like i said before. its all about the games.
Ageia must convince Developers. That is sucses full. enough support. Developers convince Player with there Games.
Well still waiting on those 65+ games. it takes some time 6 month to year or more so.

Developers supports is just the first step to sucses. Utumatly the games decide if it has sucses potentional.
 
I am intreged by what terra said about Ageia's PPU doing Raycasting (shadows)...
to me, that would be a huge offload, with a high potential for benefit to online FPS's. If hardware shadows can be offloaded from the CPU's and GPU's (I am ignorant to where shadows are processed) then there could be a significant acceleration in performance that we have not considered yet.

also, combine fluids with surface friction, could be interesting... the old spill on isle 3 could make a terrorist slip and take head injury... Or send him unwillingly into line of fire... (I know its been done before, but never dynamicly)

I do still bilieve that a faster processor may help increase the performanc of games with PhysX. I think that even though PhysX games offload the PhysX to the PPU it might still be bottlenecked at the CPU due to the increased objects and complexities created by PhysX.
Thats why I started this thread. I am interested in Cell Factor, and I'm probubly going to build a new rig soon, and I'd kinda like to know whats holding Cell Factor back.
 
nhusby said:
I am intreged by what terra said about Ageia's PPU doing Raycasting (shadows)...
to me, that would be a huge offload, with a high potential for benefit to online FPS's. If hardware shadows can be offloaded from the CPU's and GPU's (I am ignorant to where shadows are processed) then there could be a significant acceleration in performance that we have not considered yet.

Picture of it(Physcis and Raycasting at once):
image4kd1.jpg


also, combine fluids with surface friction, could be interesting... the old spill on isle 3 could make a terrorist slip and take head injury... Or send him unwillingly into line of fire... (I know its been done before, but never dynamicly)

Different levels of fricton in action:
image5qj8.jpg


I do still bilieve that a faster processor may help increase the performanc of games with PhysX. I think that even though PhysX games offload the PhysX to the PPU it might still be bottlenecked at the CPU due to the increased objects and complexities created by PhysX.
Thats why I started this thread. I am interested in Cell Factor, and I'm probubly going to build a new rig soon, and I'd kinda like to know whats holding Cell Factor back.

I belive the biggest factor holdning CF back right now is the GPU's...
CF uses everything from loads of shaders, to HDR to per pixel motion blur ect...

Terra...
 
True CF demo demostrates PhysX within a full game code situation.
wich means all whats in a game is also in this demo.

Its a Feature rich Game engine that Reality engine. Have seen the Tutorial video for making Pixel shaders effects.
 
nhusby said:
I am intreged by what terra said about Ageia's PPU doing Raycasting (shadows)...
In the case of shadow volumes (which seems to be the popular method of shadowing right now) raycasting isn't something that can benefit them specifically.

Certainly a fast vector processor is great for calculating shadow volumes, which a PPU is, but then so is a GPU.

I'm not really familiar with other techniques of shadowing, and certainly there are methods that involve raycasting, but they tend to be fairly slow AFAIK (I'm not a game/graphics coder or anything).
 
Currently my only grief I have with the whole "which physics way will win"
arguments is that until Developers comes out with products that showcase
what ATI/nVidia/Physx offers we all will not know which is "king of the physics" so to speak.

Granted CellFactor will be released sometime and maybe that will be the
"big hit" that is needed by Ageia is hoping for, for those who bought PPU's so far.
And when the remaining 60+ odd companies impliment physx into either existing
games or new ones will we be able to see something concrete as to what Ageia
has to offer in gaming.

When ATI / nVidia hurry up to make their physics versions of their drivers/software,
etc so we can at least see how well it works with or without crossfire/sli setup.

Ageia will not go out of business as the Physx PPU itself can be used in many
other applications in the private and goverment sectors. Ageia is probably
already promoting it to universities, cgi companies, aerospace, automotive, the
list could be endless.
 
Whether or not a PPU can do pseudo-physics quicker than a CPU is of absolutely no value to me.

Both the game devs and Aegia are going to have to do a LOT more to convince me spending $300 on a PPU is anything but a complete waste of money.

In other words, the only thing that will ever convince me a PPU has any value is when it can make games move beyond the current limits of polygonal geometry - IE, realistic near-atomic scale simulation of fluid, gas and solids based dynamics/interaction.
 
hfx392 said:
Both the game devs and Aegia are going to have to do a LOT more to convince me spending $300 on a PPU is anything but a complete waste of money.
[...]
IE, realistic near-atomic scale simulation of fluid, gas and solids based dynamics/interaction.
Something tells me you won't be getting a PhysX card anytime soon.
Near-atomic scale simulation of fluids is unecessary, and far too computationally difficult to be done on a PC.
 
hfx392 said:
Whether or not a PPU can do pseudo-physics quicker than a CPU is of absolutely no value to me.

Both the game devs and Aegia are going to have to do a LOT more to convince me spending $300 on a PPU is anything but a complete waste of money.

In other words, the only thing that will ever convince me a PPU has any value is when it can make games move beyond the current limits of polygonal geometry - IE, realistic near-atomic scale simulation of fluid, gas and solids based dynamics/interaction.

I'm not waiting for it, but that would be cool...

Everything in game environment built of molocular units and simulate chemical properties... That would be sweet...
 
hfx392 said:
Both the game devs and Aegia are going to have to do a LOT more to convince me spending $300 on a PPU is anything but a complete waste of money.

No, it's not worth $300 now. But it has the potential to be worth $300 if developers will exploit it. ;)
 
hfx392 said:
In other words, the only thing that will ever convince me a PPU has any value is when it can make games move beyond the current limits of polygonal geometry -
Looks like you have a beef with GPUs, not PPUs.

hfx392 said:
IE, realistic near-atomic scale simulation of fluid, gas and solids based dynamics/interaction.
hahahahahaha, you want a $300 PPU to perform the job of a multimillion dollar supercomputer?
 
What he means is he wants a PPU that will do physics calculations outside the constraints of traditional polygon based games (he mentioned "near-atomic").
 
hfx392 said:
...beyond the current limits of polygonal geometry - IE, realistic near-atomic scale simulation of fluid, gas and solids based dynamics/interaction.
Do you have any idea what kind of scale near-atomic implies?

A modern CPU can model a couple of thousand objects in real time.

A PPU, if all Ageia's claims are true, can handle 30,000 objects.

A grain of sand contains around 10,000,000,000,000,000,000 atoms.

If you constructed a game down to this kind of scale, you'd be lucky if the combined efforts of every computer on the planet could churn out a single frame within a trillion years.
 
LuminaryJanitor said:
Do you have any idea what kind of scale near-atomic implies?

A modern CPU can model a couple of thousand objects in real time.

A PPU, if all Ageia's claims are true, can handle 30,000 objects.

A grain of sand contains around 10,000,000,000,000,000,000 atoms.

If you constructed a game down to this kind of scale, you'd be lucky if the combined efforts of every computer on the planet could churn out a single frame within a trillion years.

QFT.
This must be one of the weirdest "attacks" I long have seen on the PPU.
"Sorry PhysX, you are not faster the the supercomputers that the US uses for simulating nuclear explosions, you are not worth anything! :confused:

If I think there has been the following "why the PPU suck":

"No games"...with 6 games supported and more comming that point is MEEH

"CPU is lightyears ahead of the PPU"...not even Intel or AMD would make such a claim...that point is MEEH too.

"No 64-bit drivers"..comming in September....point MEEH.

"PCI-bus is to little(bandwith) for physics...this is a PPU...no textures, just cordinates, no calculations is transfeered over the PCI bus...local storage(128MB RAM) and a MASSIVE parrallel core(both in core calcuations and memory bandwith)....point MEEH.

"No PCI-E cards"...they should be comming...but if you if you got SLI...whats the worry...NVIDA/ATI say they can do physics on their GPU's...don't you trust them? :)

"Price"...$252.99 including GRAW(FULLL version)..subtract the price for GRAW at $42.99 (252.99 - 42.99 = 210)...$210.
This is what you get of videocards in the same price range: Link...

"DOS graphics"...people blaming the PPU for the GPU's work... :rolleyes:

Terra - Did I forget any?
 
In hfx392's defense, s/he wasn't attacking the PPU, but just saying s/he wasn't interested in it at this point in time.
There's no need to be so defensive.
 
jimmyb said:
In hfx392's defense, s/he wasn't attacking the PPU, but just saying s/he wasn't interested in it at this point in time.
There's no need to be so defensive.

The "reasoning" was very stupid.
.oO(The PPU dosn't preform faster than any supercomputer on the face of the planet, then I don't wan't it!)
Come on...this is Earth, year 2006...get real...

On a technical point, he/she wouldn't notice the difference between atom-physcis and molecular physcis if he/she tried...
But that won't be in our lifetime that those kind of detailed simulations will appear.

Terra - So that point is REALLY meeh....
 
I agree that you could get a simulation virtually indistinguishable from real life without have to go down to such a microscopic scale. I also don't share hfx392's feelings on how much a PPU has to do before it becomes useful to me (I'm far less needy btw).

He probably just wants to see a level of simulation that's not available right now. In some senses I agree. I'm quite pleased with the physics of rigid bodies and cloths, but my impression of fluid simulation is that it doesn't look very good. And by "look" I mean the actual fluid in motion, not the graphics of it. To be fair though, this is not so much the fault of the hardware, but the lack of a realistic model to simulate fluids in realtime.
 
Pretty much, if it were a 'real' physics processor, there wouldn't be much need for a GPU anymore, eh?

But that's not quite what I'm looking for.

I want my rockets to make craters in whatever they may hit. I want to see earth fly, concrete crumble, metal beams bend/break, wood splinter, water splash. I'm sick of static worlds with relatively few objects tied to pseudo-physics systems. I want to be able to rip my way through that stupid chain-link fence between me and my objective, with anything I possibly can - whether it be bullets, rockets, bricks, swords, a blowtorch, or monkeys from my ass - I want it to behave realistically - all of it.

Is that really so much to ask from the gaming-hardware-world's supposed next coming of Christ?

Just being able to play with some boxes, even large amounts of boxes - isn't quite worth $300 to me.
 
hfx392 said:
Pretty much, if it were a 'real' physics processor, there wouldn't be much need for a GPU anymore, eh?

But that's not quite what I'm looking for.

I want my rockets to make craters in whatever they may hit. I want to see earth fly, concrete crumble, metal beams bend/break, wood splinter, water splash. I'm sick of static worlds with relatively few objects tied to pseudo-physics systems. I want to be able to rip my way through that stupid chain-link fence between me and my objective, with anything I possibly can - whether it be bullets, rockets, bricks, swords, a blowtorch, or monkeys from my ass - I want it to behave realistically - all of it.

Is that really so much to ask from the gaming-hardware-world's supposed next coming of Christ?

Just being able to play with some boxes, even large amounts of boxes - isn't quite worth $300 to me.

You should read up more on the capabilities of the PPU, you seem to be very ill informed...
And please don't mix human superstion (christ) into a real world hardware debate.
I don't use the easterbunny as an argument either...

Terra - BTW, none of you points have the need for molecular/atomar level physics?
 
I think your X-TREME XRAM soundcard and PPU will both be hanging on your wall in a year, terra

you've got the money to buy gimmicks, keep them to yourself and only talk when they show some practicality
 
Big Fat Duck said:
I think your X-TREME XRAM soundcard and PPU will both be hanging on your wall in a year, terra

you've got the money to buy gimmicks, keep them to yourself and only talk when they show some practicality

My XRAM get used in all Doom3 engine based games from Doom3 to Prey and I enjoy support for 128 hardware voices in all other games, and I won't upgrade untill the next gen of soundcard that offer sme much imporve is here...welcome to the real world...

My PPU also gets used in games, and will be in service as long as this PC is still kicking, wich will be for years to come, again welcome to the real world...

If you don't like my posts or my purchases, here is the fix for you: Linky

Terra - You sound awfully bitter...
 
hfx392 said:
Is that really so much to ask from the gaming-hardware-world's supposed next coming of Christ?
What you're asking for has little to do with hardware. Fully destructible environments are a problem for developers, not semiconductor companies.

And not something you're likely to find in the foreseeable future. Walls serve a very important purpose in level design. Give the player the ability to remove them, and the designer loses a lot of control. The capability has been around at least since Red Faction, and developers are still avoiding it. New hardware isn't going to change this.

In any case, the answer is Yes. That is far too much to ask of any piece of modern hardware. You're talking about, at best, only millions of physically simulated voxels. This is a few of orders of magnitude above the capabilities of current hardware. Even a supercomputer couldn't pull it off in real time.
 
No i think not. The way i see it.
You got
Red Fraction as a very limited example. Level A
Of that Destrucable enviorment feature.
And
Startrek Holo Decks Level Y (Z would be a high-end HoleDeck :) )
Where Level P is possible with the high-end Cray super computer with aditinal Co-PU.
So there a lot in between. the point of my vision.
What hardware can make posible.
Like
DC could be Level C
Quadcore level D
PPU Level E
A R800 DX11/12 from ATI abused for gameplay PhysX in 2010 would be like Level G

So there is more possible with this new hardware but it is a long road.
This mean This power can be use to step Up the Scale and or the Physix detail.
Like 10 Wall's in Redfraction could be 10000 wall's in same detail.
Or with Higher detail 1000.
Or more structures and walls and objects with some more detail like 100.

So with each level It's all gameplay witch counts. Level A was a bad use of it. Could be better if it was merge to create a gameplay where this feature excels. But its more some icing on the cake. it doesn't add to the gameplay.

So Level E will not be all that. But put to good use it could make a new kind of gameplay witch is very fun to do and more people are willing to add new special hardware for it.

Those 65+ games. There must be a few who put it to good use. Hopefully.

Cellfactor a unofficcaly word about the retailgame. Would implement destructable enviorments to. You got now a hunche that it would be like a more very souped up Redfraction. But that not bad that evolution. For where its hading to in the long run. the in between steps. Important is will such game be fun. That all in the hands of the dev's. But don't expect like level M when that still far and away. But what they do with is posible with it.
 
SuperGee said:
No i think not. The way i see it.
You got
Red Fraction as a very limited example. Level A
Of that Destrucable enviorment feature.
And
Startrek Holo Decks Level Y (Z would be a high-end HoleDeck :) )
Where Level P is possible with the high-end Cray super computer with aditinal Co-PU.
So there a lot in between. the point of my vision.
What hardware can make posible.
Like
DC could be Level C
Quadcore level D
PPU Level E
A R800 DX11/12 from ATI abused for gameplay PhysX in 2010 would be like Level G

So there is more possible with this new hardware but it is a long road.
This mean This power can be use to step Up the Scale and or the Physix detail.
Like 10 Wall's in Redfraction could be 10000 wall's in same detail.
Or with Higher detail 1000.
Or more structures and walls and objects with some more detail like 100.

So with each level It's all gameplay witch counts. Level A was a bad use of it. Could be better if it was merge to create a gameplay where this feature excels. But its more some icing on the cake. it doesn't add to the gameplay.

So Level E will not be all that. But put to good use it could make a new kind of gameplay witch is very fun to do and more people are willing to add new special hardware for it.

Those 65+ games. There must be a few who put it to good use. Hopefully.

Cellfactor a unofficcaly word about the retailgame. Would implement destructable enviorments to. You got now a hunche that it would be like a more very souped up Redfraction. But that not bad that evolution. For where its hading to in the long run. the in between steps. Important is will such game be fun. That all in the hands of the dev's. But don't expect like level M when that still far and away. But what they do with is posible with it.

Out of those 65 games only 2 are guaranteed to support PPU hardware beyond offloading from CPU to PPU. Most games listed aren't big budget titles so I doubt they would put a lot of effort and money in utilizing it's features for a small community of gamers.
 
SuperGee said:
...
Red Fraction as a very limited example. Level A
...
Where Level P is possible with the high-end Cray super computer with aditinal Co-PU.
...
What hardware can make posible....
You've completely, totally and utterly missed my point.

It doesn't matter how realistic you can make it look. If the gameplay doesn't work, it's not going to happen. This is not a question of hardware. Developers have the capacity to build fully destructible environments right now. And they're not doing it. What makes you think they will in the future?
 
Back
Top