PPU's a fad that will never happen?

chrisf6969

[H]F Junkie
Joined
Oct 27, 2003
Messages
9,011
I think its a BIG waste of money, time & even a sub-forum.

having a PPU with the upcoming multi-core chips will be like having a math co-processor. IE: a BIG waste

By the time PPU's could get "mainstream", Intel & AMD will probably both have quad core's as their mainstream GAMING processors.

And with programmers inifinitely more familar with coding for x86 CPU's (not PPU's) I think the choice will be clear in a year or 2 from now. IE: Physics will be done on the extra cores.
 
Well just to be clear a PPU (That is Physics Processing Unit) is not a fad.

An add-in card version my very well be unsuccessful given what AMD/Intel are doing along with the raw power of a G80 card. But I think we will see a PPU in some form or another. While CPU's continue to grow in cores they are ultimately general processors. And in time as we see more and more advanced physics in games (as developers get a grasp on what works and what doesn't) I think the need will definitely arise for a specialized PPU.

Again, to be clear it may not exist in the form we know it today. It may become a standard feature on video cards to have an extra PPU chip embedded on its board.

Or we may very well see high end motherboards have integrated physics as opposed to integrated graphics.

The thing to focus on is that in the beginning of 3d graphics things started slow and were kind of quirky (pre-direct X days). You needed a card that was supported by the game and if you were really cool you could select your Sound blaster 16 from the sound menu OOOOOOHHHHH.

All it takes is Microsoft to add physics to their direct X API and bam! when the standard is set PPU's will come into their glory (Just not on their own dedicated card)
 
Why do you have a GPU. Just run it on an extra core on your CPU!!!!!111!

:rolleyes:

A PPU is significantly more powerful than a CPU. Just as a GPU is vs a CPU

A processor built around a very specific function will always be faster than a processor that tries to cover all the bases.
 
Unless the programs or drivers (in this case the physics) are coded in such a way as to take advantage of the architecture of the CPU/GPU/PPU or whatever is doing the physics.

I think the PPU will become more mainstream among enthusiasts when you can grab an old 7XXX series video card and have drivers that let you use it as a physics card.

PPU wouldn't be a fad if more games supported it and if the price was "reasonable" (say around $100 with good game support).
 
Vashypooh said:
Why do you have a GPU. Just run it on an extra core on your CPU!!!!!111!

:rolleyes:

A PPU is significantly more powerful than a CPU. Just as a GPU is vs a CPU

A processor built around a very specific function will always be faster than a processor that tries to cover all the bases.


Yeah, but I'm having such trouble finding games that use my graphics card!
 
Fad implies popularity, which is non existent.

Maybe in the future it will be a Fad when people actually want it, but not right now.
 
Why a PPU subforum? well it could be come just like Grapics a more common place for Phsyics hardware.
Where CPU PPU GPU for Phsyics are discussed.

PPU is just the first one. Wich come to a point.
Hardware physics competition is PPU & GPU. vs CPU


If PPU doesn't fullfill GPU can. nV and ATI are putting large efforts to storm that new market. So it doesn't depend on ageia if HArdware physics sucseed it is all in the hands off Ageia and ATI and nV.

it's more a question if ageia will be a part of it in the future like say over 2 á 3 years.

Now I put a PPU in my rig. Then comes a time you have next to a PPU a G-card for Phsyics. If Ageia doesn't make it. It will after some time be only a G-card for Physics.
However that still far down the road. How it unveils.
In the mean time PPU get's more games out in the field. So PPU get more usablity however that takes time. Within a year. Ageia must do it thing.
 
Remember math co-processors? What ever happened to those? Did cpu's just out grow the need for them or what?
 
urbsnspices said:
Remember math co-processors? What ever happened to those? Did cpu's just out grow the need for them or what?
yes and no....

yes... they redesigned CPU's to do floating point...

no... now they are called graphics cards and ppu's...
 
chrisf6969 said:
I think its a BIG waste of money, time & even a sub-forum.

having a PPU with the upcoming multi-core chips will be like having a math co-processor. IE: a BIG waste

By the time PPU's could get "mainstream", Intel & AMD will probably both have quad core's as their mainstream GAMING processors.

And with programmers inifinitely more familar with coding for x86 CPU's (not PPU's) I think the choice will be clear in a year or 2 from now. IE: Physics will be done on the extra cores.

You don’t really sound like your all that familiar with how a game works with the PPU. Games that have the potential to work with the PPU aren’t “Coded for the PPU”, they use the PhysX engine. The PhysX engine existed prior to the PPU, and is actually fairly widespread; only newer versions of the engine work with the PPU.

Given that the PhysX engine has a long history that precedes the PPU even existing, it makes great use of the CPU, even to the extent that it is multi-threaded to make use of multiple cores.

Of course there are other factors, such as that the engine is being basically given away to developers whereas other similar engines can cost hundreds of thousands of dollars.

You also seem to be fairly comfortable comparing discrete processing solutions with general purpose processors as if they were really comparable in any way. Multiple Core processors open up a lot of new doors but it’s not going to mean the end of discrete processing solutions. Imagine trying to run a game in software mode (no 3d acceleration on the GPU) using a brand new quad-core Processor. Let’s even go as far as to say somehow you were able to use 3 out of the 4 cores just for graphics processing. How fast do you think it would be? Do you think it would even be faster than a budget IGP? Likely not… In this similar way, the potential for physics on a PPU drastically exceeds what can be done on a general purpose processor and it doesn’t really matter how many cores you have.

You’ll have games (unreal engine 3 based games, etc) out that use the PhysX engine, that work fine on the CPU and run great, and this will be what most people use. You’ll also have the potential to drop in a PPU and have that physics acceleration offloaded with a very likely performance increase. Given the sheer amount of money people pour into things that only give you a very small performance increase, what makes you think some people won’t also buy PPU’s in this same scenario. People will drop hundreds on a top-end X-Fi just to get X-Ram.
 
I had a weitec it's competitor was the 80387.
I think they moved on "die" as intel where lookin for what to do with those extra transistors posible with a dieshrink far in the past around that switch.
Now CoProcs or FPU is evoluted from 3Dknow to SSE 1 to the upcoming SSE4 unit's.
To The Streaming Procs in the G80 and PPU.
A dedicated chip yield more power then using a part of a chip for it.
 
The PPU market needs two things to really take off.

1 - A killer app. For 3D accelerators it was GLQuake. Sure the move to 3D cards was happening and would have happened eventually without GL Quake and a lot of other programs played their part. But that game and it's engine is where 3D really got momentum. Anyone that saw it said wow, that IS a big difference. Any gamer that saw it said I have to have it. It didn't happen in 2006, we'll see what 2007 brings.

2 - Better standardization. Is a game written for Physx going to benefit from Nvidia or ATI's upcoming solutions, or a game written for Nvidia going to benefit from a Physx board? Right now my impression is the current solutions are like 3dfx's glide. Maybe they are great, but there doesn't seem to be an OpenGL or Direct3D to tie it all together.

Any one of these two things will get the market moving, and if PPU's live up to the hype both will happen and the PPU will just be part of gaming. Either as part of the graphics card or a seperate card, it's to far off to say which will really succeed and really both could. I don't see dual or quad core CPU's really stopping this, dual and quad core (SLI/Crossfire/GX2/Etc) graphics cards MIGHT stop the seperate add on cards though, or push them into the hardcore gamer market like add on sound cards are now. It's to early to tell IMO.
 
SuperGee said:
I had a weitec it's competitor was the 80387.
I think they moved on "die" as intel where lookin for what to do with those extra transistors posible with a dieshrink far in the past around that switch.
Now CoProcs or FPU is evoluted from 3Dknow to SSE 1 to the upcoming SSE4 unit's.
To The Streaming Procs in the G80 and PPU.
A dedicated chip yield more power then using a part of a chip for it.



It was Weitek http://cpu-museum.de/?m=Weitek&f=Math+CPUs
3Dnow never really got anywhere,it did a little to help Quake2 back I the
day though.As for the PPU getting a killer app,it will not happen in the way
some here hope it will,like the way Quake did,with Glide and OpenGL.New
Modern cpu's and the stream procs in the 8800/R600 series will give the
average gamer,and even the 'enthuasist gamer' all they need,as far as
hardware accelerated physics go.
 
I don't see any reason it couldn't happen, well I would be shocked if it was quite THAT dramatic but if you can't see a real difference it's worthless. That doesn't mean such an app will be created, I just don't see the PPU ever taking off without it.
 
I didnt read any of this thread, here is my response to the title alone:

The PPU is FOR SURE a good idea. A client for the company I work for built some crazy workstations to run some type of 3D software that they use to seperate tissue (cancer, diseases, I dont know much about it) on a Dell 3007FPW to show clients of theirs. It suposedly renders the tissue doing whatever, and then they can cut/split it in real time.

They heard about the PPU then wrote the software to take advantage of it and he said their frames per second increased by 20-40%... and they were running DUAL 8800GTX, 4gbs of ram, and a Core 2 Duo.

If they can do it then gaming companies can do it... although Im sure a game would need to be worked with a little differently then the program they have that does one specific thing.
 
Manny Calavera said:
As for the PPU getting a killer app,it will not happen in the way
some here hope it will,like the way Quake did,with Glide and OpenGL.New
Modern cpu's and the stream procs in the 8800/R600 series will give the
average gamer,and even the 'enthuasist gamer' all they need,as far as
hardware accelerated physics go.
I think you forget the API dependancy games will be.

For hardware accelerated Physics you need the hardware for the API being used.

A g80 Quantum does nothing with it's quantum or its more its stream procs for Physics,if the game uses PhysX as API. While a PPU will do nothing with a game using Havok FX.
The API will use whatever CPU is used instead.
So it's a matter if any or even more PhysX game supporting PPU will get populair or not.

If you even like a lot of different games. It could be
Quad core, HavokFX, PhysX games, you need all solutions to be able to play every game in optimum as it was ment to be played. That will take a QC, PPU and a second G80/R600. Because the enthausiast wont deliver in on rendering power. And wenn DX10 games come out they will make good use of the G80/R600. It will not be overkill anymore.
So sharing rendering and Physics comes with a cost. With DX10 titles that needs a haevy DX10 Card to render optimal.


So I'am planning a Vista rig for begin 2007 wenn DX10 games become avaible.
4GB mem
A QC
R600
X1950 for Physics.
PPU

I could buy that Physics G-card later because first there will be a growing need for PPU. So it will have a PPU from the start. Later the needs comes for a Extra G-card for HAvok FX games. That would be possible mid/end 2007.

Till there comes a time where almost every game will use a standard PhysX API DirectXPhysX. Then it will be the Performance or support for that standard API wich make the difference. But that situation is still far away.

Now and in the near future, it is what Physics API those dev uses for your favo game.
 
Blimey is to release a racing simulation. A real racing simulation that is using the Ferrari license. They are going to use the Ageia chip for it´s physics calculations.

If you don´t know who Blimey is it´s the developers of the GTR titles together with Simbin. They are working on their own game with their own game engine now.

AGEIA for me has the most use in racing and flight simulators which is extremely demanding on the CPU for all the physics work.

Seeing how hard game developers have ot optimize for dual cores even if it´s easier with the AGEIA it may kick off :)
 
oqvist said:
Blimey is to release a racing simulation. A real racing simulation that is using the Ferrari license. They are going to use the Ageia chip for it´s physics calculations.

If you don´t know who Blimey is it´s the developers of the GTR titles together with Simbin. They are working on their own game with their own game engine now.

AGEIA for me has the most use in racing and flight simulators which is extremely demanding on the CPU for all the physics work.

Seeing how hard game developers have ot optimize for dual cores even if it´s easier with the AGEIA it may kick off :)

If they were to seriously do that, they'll alienate the rest of the non PPU carrying community since the suspension and handling system (which is basically what defines the car) of the same car would be drastically different between PPU and non-PPU users.
 
In what way would it be different? What stops you from doing the same calculations with the CPU even if it may perform less ;)
 
The reason why specialized co-processors go the way of the dinosaur is pretty apparent: They're too specialized. Any sort of highly specialized co-processor will be effectively wasted except when doing its particular task. If the demand for that sort of processing is sufficient, then over time, general purpose processors will subsume the abilities of the co-processor into its retinue. Considering the rapidly increasing spare processing power from multi-core cpus, there's soon going to be a wealth of currently-unused processor power (in games) just waiting to be tapped. There's little reason to consider a ppu at that point.
 
Sly said:
If they were to seriously do that, they'll alienate the rest of the non PPU carrying community since the suspension and handling system (which is basically what defines the car) of the same car would be drastically different between PPU and non-PPU users.

I wouldn't think they'd be different, just one would perform better than the other. Obviously non-PPU systems would have a harder time doing the same things a PPU system would.

In any case, companies need to worry a little bit less about alienating certain computer users. With that kind of thinking 3D Acceleration and multiple CPU cores would never have been invented, much less have become mainstream.

At some point, when more games start using some kind of off CPU physics hardware, gamers will start buying that hardware. Eventually, all gaming machines will come with that hardware standard. Eventually, it will become a popular upgrade item, and even longer in the distant future, all machines may have it or some form of it. It will get cheaper and easier for the masses to have it.
 
So Dan you think the PPU will become very mainstream, maybe even so mainstream that it could becoma an onboard component like LAN & sound?

I just find it hard to phathom people buying in to it.

LAN & sound are required by pretty much 100% of users now-a-days.

Graphics cards are such a "HOT" (pun intended) item b/c the games are constantly changing & getting better and the CPU's would have NEVER been able to keep up with them. In fact its taken the GPU's redesigning & refreshing every 6 months to keep up.

BUT I think with physics, once we have quad core processors as a mainstream (gaming) CPU in 2008, it should be able to handle nearly any physics processing needed easily. IE: not as dire of a need for massive power.
 
chrisf6969 said:
So Dan you think the PPU will become very mainstream, maybe even so mainstream that it could becoma an onboard component like LAN & sound?

I just find it hard to phathom people buying in to it.

LAN & sound are required by pretty much 100% of users now-a-days.

Graphics cards are such a "HOT" (pun intended) item b/c the games are constantly changing & getting better and the CPU's would have NEVER been able to keep up with them. In fact its taken the GPU's redesigning & refreshing every 6 months to keep up.

BUT I think with physics, once we have quad core processors as a mainstream (gaming) CPU in 2008, it should be able to handle nearly any physics processing needed easily. IE: not as dire of a need for massive power.

I don't know if people will buy them or not. Remember the [H] doing an article on PhysX needing a killer app? Well I agree with the sentiment of the article. PhysX needs, a game like Quake or something else that can jump start PhysX card purchases. People won't buy them if there isn't a need for them. Right now there is not ONE compelling reason to buy such a card.

Now, if a game came along that actually really took advantage of the PPU, much like early 3D Acceleration, then adoption would begin to occur. Eventually, I can see PPU's being built into CPU's as an additional core, like the mathco, or residing on video cards. They could end up going the way of the video card and get larger, louder and more powerful and always be in card form, but I really can't say. I can't predict the future.

Its really in the hands of the game developers at this time. What I suspect is that until they create a game that simply isn't worth a shit without a PPU, people won't buy them.

Much like comparing games like Quake on 3D Accelerated computers, and those without. The difference in visual quality was HUGE, and compelling. Even with that it took years for 3D Acceleration to become mainstream. Even to this day the video hardware in 95% of retail machines still isn't worth a shit. I think that for PhysX to become popular they are going to need more than particle and debris effects in a second rate game like GRAW in order to get people excited about their product. I don't even think the promise of frame rate increases alone would be enough. A game is going to have to look different, and play different in order for it to happen for Ageia.
 
If only they developed the hardware to run on any engine instead of their own. Would love to see games like HL2 with improved physics.
 
WarriorX said:
If only they developed the hardware to run on any engine instead of their own. Would love to see games like HL2 with improved physics.
This is what I had hoped would happen. I don't have enough tech knowledge to know why it wouldn't work, but I had hoped support for a PPU could be implemented in a patch, as opposed to having to build the game from the ground up.
 
This reminds me of the RDRAM "fad." Pricy and supposed to offer advantages over regular options. Eventually Rambus got rejected by everyone.

Maybe the new video card PPUs will bring the PPU back. I mean, you upgrade your computer, you put your old video card into the PPU slot, and enjoy the benefits.
 
Night Black said:
This reminds me of the RDRAM "fad." Pricy and supposed to offer advantages over regular options. Eventually Rambus got rejected by everyone.
Ageia is more like 3DFX and some of them are EX 3DFX.

Rejected by everyone? In the early day the only way to reject Rambus was by rejecting iNtel. So to go for AMD. As iNtel forced Rambus on you. The only one able to reject rambus directly was iNtel wich it did in the long run.
Rambus was not so populair because of it's patent royalty's and royalties hunting in the bussness. Worked bad for marketing for rambus. Reminds constumers that the pay extra cash to make a firm richer and get not much return for it.
Its like extra taxes.

RDRAM is a better solution it offers easier channeling then DDRram.
We are stick at 2 channel. while RDRAM could be easy extended to 4 or 8 channel.
Before that iNTel dropped the pushing for Rambus.
And Rambus did give The early P4 the band wide it needed But at a high cost.
And the reason why it is so high partly because off royalties to rambus, people don't like at all.

Ageia PPU PhysX is like 3DFX Voodoo Glide.
Maybe the new video card PPUs will bring the PPU back. I mean, you upgrade your computer, you put your old video card into the PPU slot, and enjoy the benefits.
Why do you call a GPU a PPU. So a G80 is a PPU?
 
SuperGee said:
I think you forget the API dependancy games will be.

For hardware accelerated Physics you need the hardware for the API being used.

A g80 Quantum does nothing with it's quantum or its more its stream procs for Physics,if the game uses PhysX as API. While a PPU will do nothing with a game using Havok FX.
The API will use whatever CPU is used instead.
So it's a matter if any or even more PhysX game supporting PPU will get populair or not.

If you even like a lot of different games. It could be
Quad core, HavokFX, PhysX games, you need all solutions to be able to play every game in optimum as it was ment to be played. That will take a QC, PPU and a second G80/R600. Because the enthausiast wont deliver in on rendering power. And wenn DX10 games come out they will make good use of the G80/R600. It will not be overkill anymore.
So sharing rendering and Physics comes with a cost. With DX10 titles that needs a haevy DX10 Card to render optimal.


So I'am planning a Vista rig for begin 2007 wenn DX10 games become avaible.
4GB mem
A QC
R600
X1950 for Physics.
PPU

I could buy that Physics G-card later because first there will be a growing need for PPU. So it will have a PPU from the start. Later the needs comes for a Extra G-card for HAvok FX games. That would be possible mid/end 2007.

Till there comes a time where almost every game will use a standard PhysX API DirectXPhysX. Then it will be the Performance or support for that standard API wich make the difference. But that situation is still far away.

Now and in the near future, it is what Physics API those dev uses for your favo game.



And you are forgetting about a little thing called C (CUDA) far far easier to program for,not to mention a little thing you might have heard of called Direct X (ie: 10) :D I bet 20 bucks
that in 2 years this 'PPU' is a long lost memory,and Physics in games on the PC,will be
done almost entirely on 'GPU's'.Everyone uses Direct X,every programmer worth his salt
knows C inside and out.... Very soon everyone will have a built in 'PPU' in there DX10 hardware,especially as lower end DX 10 parts hit the market.Need I say more ?? Get your head out of the sand.
 
SuperGee said:
I think you forget the API dependancy games will be.

For hardware accelerated Physics you need the hardware for the API being used.

A g80 Quantum does nothing with it's quantum or its more its stream procs for Physics,if the game uses PhysX as API. While a PPU will do nothing with a game using Havok FX.
The API will use whatever CPU is used instead.
So it's a matter if any or even more PhysX game supporting PPU will get populair or not.

If you even like a lot of different games. It could be
Quad core, HavokFX, PhysX games, you need all solutions to be able to play every game in optimum as it was ment to be played. That will take a QC, PPU and a second G80/R600. Because the enthausiast wont deliver in on rendering power. And wenn DX10 games come out they will make good use of the G80/R600. It will not be overkill anymore.
So sharing rendering and Physics comes with a cost. With DX10 titles that needs a haevy DX10 Card to render optimal.


So I'am planning a Vista rig for begin 2007 wenn DX10 games become avaible.
4GB mem
A QC
R600
X1950 for Physics.
PPU

I could buy that Physics G-card later because first there will be a growing need for PPU. So it will have a PPU from the start. Later the needs comes for a Extra G-card for HAvok FX games. That would be possible mid/end 2007.

Till there comes a time where almost every game will use a standard PhysX API DirectXPhysX. Then it will be the Performance or support for that standard API wich make the difference. But that situation is still far away.

Now and in the near future, it is what Physics API those dev uses for your favo game.

Huh? I am not really sure what the hell you are trying to say.

But, eventually one standard for physics processing will take over. It could be done via additional CPU cores, or it might be done onboard GPUs and even the PhysX card could take over as the defacto standard. We don't know at this point.

You will not need to have all three types of solutions in your PC. Eventually I can see Microsoft adding a standard physics API as part of Direct X. Such a move should clear this problem up.
 
I'm pretty sure it'll become some sort of onboard solution at some point. Kind of what happened to onboard nic and sound. First, there was cards you had to buy for network and sound, then everyone had internet and wanted sound, so the mobo manufacturers got a penny solution and made it an onboard item. When was the last time you've seen a motherboard without onboard sound and nic? Exactly.
 
SuperGee said:
Ageia is more like 3DFX and some of them are EX 3DFX.

Rejected by everyone? In the early day the only way to reject Rambus was by rejecting iNtel. So to go for AMD. As iNtel forced Rambus on you. The only one able to reject rambus directly was iNtel wich it did in the long run.
Rambus was not so populair because of it's patent royalty's and royalties hunting in the bussness. Worked bad for marketing for rambus. Reminds constumers that the pay extra cash to make a firm richer and get not much return for it.
Its like extra taxes.

RDRAM is a better solution it offers easier channeling then DDRram.
We are stick at 2 channel. while RDRAM could be easy extended to 4 or 8 channel.
Before that iNTel dropped the pushing for Rambus.
And Rambus did give The early P4 the band wide it needed But at a high cost.
And the reason why it is so high partly because off royalties to rambus, people don't like at all.

Ageia PPU PhysX is like 3DFX Voodoo Glide.

Why do you call a GPU a PPU. So a G80 is a PPU?

First, I said, EVENTUALLY. Not right away. Eventually. I have RDRAM, and gradually it's become impossible to get it. Secondly, with RD600 and 680i, they both have a third PCI-E x16 port for video card physics in the future. (I think NVIDIA calls it Quantem something.)
 
I agree that PPU's are a fad. Dual core is now the standard. Quad core is already here. It will not be long before CPUs have specialized cores dedicated to things like physics. Ageia would be wise to close up shop and sell their tech to Intel and/or AMD.
 
Devnull said:
I agree that PPU's are a fad. Dual core is now the standard. Quad core is already here. It will not be long before CPUs have specialized cores dedicated to things like physics. Ageia would be wise to close up shop and sell their tech to Intel and/or AMD.
The critic on ageia is there aren't games to realy show off PhysX.
This is a problem for now in current time could change in 6 month.

So it would be not long means you are very patient and wait like 3 to 4 years for Copro to surface on CPU die's and reach mainstream. where waitin on AMD K8L still.
If 6 month is such problem for short time sightness off the general public what would be a year or three. That would be far far away in the future.

QC is now just one Extreem edition of $1000.
Later there comes a $800 and then later then that a $550 part.
Still not meanstream that would be around end 2007 to 2008.

The first competitor to show up agians ageia PPU in not so long time is HAvok FX using current GPU's or of the very near future so you can count G81 and R600 in..
That will show up in 6 to 9 month. and also that needs a killer app.
Altho there solution is world favorite. They can have some delay with there killer app.
Where ageia is in harder need of a killer app soon.

Wenn Phsyics dedicated cores comes on CPU. We are years further where most PCgamers wich want decent hardware have reach a second upgrade from now.

wich means such ondie Copro would compete with
G90 or g100
PPU2/3
And run next next next gen games

Where dedicated hardware for Physics is a couple of times more powerfull then the Copro on that CPU. it's the poor man's solution for hardware accelerated Physics.

Like AMD plans for mudulair CPU to put also a bargain bin GPU's Copro and specialised units on CPU "Die" There still be a need for dedicated midrange and High-end hardware.

Like you have now in the near future the choice of putting 2 G80 in a rig for HavokFX.
Or a midrange soulution that competes with PPU a single G80 do both rendering and Physics
While from now you can put a PPU in it for a midrange pice.

The Hardware situation over 3 years or more don't have any influence for what I would need for the next 2 years.

Whatever you have or need now for the next year will be out dated in 3 years.
Wenn it is so far I wil make a DX11/12 Rig with what is best in that period of this industry.
Now I gonna make a Vista DX10 rig in begin 2007. To be ready for the next 2 years.

And that could use a PPU to.
 
Why not a socket on the mobo for a cpu dedicated to physics? Even an existing socket, like 939, 478 or 775, etc, with a cpu solely for physics calcs. Maybe a dedicated ram slot for the ppu cpu.
Would this work, or not ?
 
oqvist said:
Blimey is to release a racing simulation. A real racing simulation that is using the Ferrari license. They are going to use the Ageia chip for it´s physics calculations.

If you don´t know who Blimey is it´s the developers of the GTR titles together with Simbin. They are working on their own game with their own game engine now.

AGEIA for me has the most use in racing and flight simulators which is extremely demanding on the CPU for all the physics work.

Seeing how hard game developers have ot optimize for dual cores even if it´s easier with the AGEIA it may kick off :)
If they do I'll buy a PPU, that sounds awesome
 
Dan_D said:
I wouldn't think they'd be different, just one would perform better than the other. Obviously non-PPU systems would have a harder time doing the same things a PPU system would.

Actually, if the game was specially designed for a PPU, then the physics calculations done would have to make buying a PPU worth it. It will have to be in a league that would make it impractical to do on a CPU.

Another PC going against it, would have to have the physics calculations significantly reduced in order for the software to handle it. That would make a rather large impact on the cars handling.

How would like it if the Ferrari felt like a Diablo.
 
I think the ppu is here to stay whether ageia is the torch holder or not. The fact they are giving away the engine for free will put it in a alot of games vs the havok engine. While s they may never make a profit themselves, I guarantee either amd or nvidia would buy them up in a heartbeat if they were headed for bankruptcy. Amd's slides for the fusion cpu already speak of possibility adding physics cores to it.
 
Back
Top