Will hype kill the PPU.

[Chernobyl1]

Yes, I agree. But there are many people here saying that it WON'T be needed in the future, it SHOULDN'T be needed in the future, and it will ALWAYS be useless. That's not true.

I believe Physics will be needed in the future, it should be implemented in the future and it has been and always will be useful.

This is exactly what I mean

c'mon, I tried hard to get back on topic, lets not ruin this thread.
It has ended getting a lot of information exposed.

 

[HOCP4ME]

Okay, I agree with that. Hardware-accelerated physics will come somehow. But most people are bashing the PPU in favor of other solutions, when in reality the PPU is a great product with great potential.

As for will hype kill the PPU (the topic), I don't think there really is much hype. It's become a two-sided situation. Some people think the PPU will die, and some think it will become the physics solution of choice. There isn't anyone saying "hey look at this uber-cool piece of hardware that will revolutionize gaming" anymore like there was six months ago.

 

[Chernobyl1]

I'm cool with that

 

[dotK]

Please elborate on the highligthed?

Terra...

Some things that can cause the PPU to be a bit less than desirable:
#1 - No games are currently available that make the PPU a worthwhile investment.
#2 - Multiplayer support in most games will probably be limited to just effects physics. UT2K7, for example, is one of the potential (and most likely the first) killer apps and is going to be mainly a MP game which means it won't take full advantage of the PPU. MMOs will never see anything beyond effects physics.
#3 - A lot of the stuff can be done with a CPU with similar performance (see Cell Factor)
#4 - Liquids on the PPU look awful. Liquids are one of the more "advanced" things people were looking forward to from a dedicated physics solution and Ageia has really dropped the ball.
#5 - Ageia is going to have competition from two major companies, NVIDIA and ATI. Both companies also have proven track records and are going to prove to be strong competition. I would be a bit hesitant to spend money on a PPU now when 2 juggernauts in the computer industry are going to be releasing their own solutions. They have the money to encourage developers to see it their way.
#6 - "Wait a few more months" got old months ago. They released the PPU too soon. The Sega Saturn was released too soon and like the PPU, had mediocre software support. The console launch was awful and very few people bought it. Sony (in this case ATI/NVIDIA) had plenty of time to hype up the PS (GPU physics) and ended up killing the Saturn.
#7 - The GPU physics solution offers you the option of using an old video card for physics that would normally just sit around. With Ageia's solution, that kind of flexibility is not present and you have to buy a new card.
#8 - No PCI-express PPUs means some people (who are also likely to be the first adopters) simply can't use the cards.
#9 - I don't really recall all the exact details, but older games will run older versions of the PhysX engine so it's possible that driver updates won't fix many issues in older engine versions that are likely to be present. Drivers need to have every version of the PhysX engine in them.
#10 - Limitations on how individual physics engines can be programmed, not completely programmable PPU.
#11 - Scope of what the PPU can do might be smaller than initially hoped for. According to that article at Tom's, the PPU won't be able to have buildings collapse under their own weight when damaged (unless maybe if they're designed out of "bricks," which wouldn't be very realisitic at all). How much truth there is to this has yet to be seen as there are NO such games out for it to prove it to be true or false.
#12 - $300 is too much money. Most people don't even spend $300 on a new video card so it's crazy to think a lot of people are going to spend $300 on something that makes crap fall down or particles fly around.
#13 - The whole physics "revolution" that we're on the brink of now might not amount to anything in the long run. The extra physics might not add much to gameplay or might greatly increase development time significantly that most developers will opt out of elaborate physics. It's all speculation at this point, so it's entirely possible that PPUs might go the way of the dinosaurs.
#14 - Multicore CPUs are going to become increasingly popular. While a dedicated solution may be better, the better solution is not the one that always wins (hi2u VHS).

 

[Terra]

Some things that can cause the PPU to be a bit less than desirable:
#1 - No games are currently available that make the PPU a worthwhile investment.

Different opinions.

#2 - Multiplayer support in most games will probably be limited to just effects physics. UT2K7, for example, is one of the potential (and most likely the first) killer apps and is going to be mainly a MP game which means it won't take full advantage of the PPU. MMOs will never see anything beyond effects physics.

CellFactor - Revolutions will feature gameplay physics via LAN and WAN...

#3 - A lot of the stuff can be done with a CPU with similar performance (see Cell Factor)

And a lot cannot...cloth, fludis ect.(see CellFactor)

#4 - Liquids on the PPU look awful. Liquids are one of the more "advanced" things people were looking forward to from a dedicated physics solution and Ageia has really dropped the ball.

Look?
That must be a GPU/graphical design problem.
The PhysX renders NO graphics...
It acts like a real-time fluids...wich is what the PhysX is about.

#5 - Ageia is going to have competition from two major companies, NVIDIA and ATI. Both companies also have proven track records and are going to prove to be strong competition. I would be a bit hesitant to spend money on a PPU now when 2 juggernauts in the computer industry are going to be releasing their own solutions. They have the money to encourage developers to see it their way.

And they have a major absences of demoes ect.
They say they can...but we have to wait "9-12 months"...
Or even developers rallying behind them...

#6 - "Wait a few more months" got old months ago. They released the PPU too soon. The Sega Saturn was released too soon and like the PPU, had mediocre software support. The console launch was awful and very few people bought it. Sony (in this case ATI/NVIDIA) had plenty of time to hype up the PS (GPU physics) and ended up killing the Saturn.

To soon?
No hardware, no game support, simple as that.
What got old motnhs ago is the GPU-physcis PR...all words nothing to show.

#7 - The GPU physics solution offers you the option of using an old video card for physics that would normally just sit around. With Ageia's solution, that kind of flexibility is not present and you have to buy a new card.

And you have the preformance data for that GPU-Physcis?
I would love to see it...so would the rest of the world...

#8 - No PCI-express PPUs means some people (who are also likely to be the first adopters) simply can't use the cards.

PCI-E PhysX is comming, but I have no PCI-E, so I am quite happy PCI, perhaps people should get a mobo that siuts their needs instead?

#9 - I don't really recall all the exact details, but older games will run older versions of the PhysX engine so it's possible that driver updates won't fix many issues in older engine versions that are likely to be present. Drivers need to have every version of the PhysX engine in them.

The current driver is at 5.26MB.
Compare that to ATI or NVIDIA's driver and I (yet again) fail to see your point?

#10 - Limitations on how individual physics engines can be programmed, not completely programmable PPU.

Elaborate please?

#11 - Scope of what the PPU can do might be smaller than initially hoped for. According to that article at Tom's, the PPU won't be able to have buildings collapse under their own weight when damaged (unless maybe if they're designed out of "bricks," which wouldn't be very realisitic at all). How much truth there is to this has yet to be seen as there are NO such games out for it to prove it to be true or false.

Good thing CellFactor - Revolutions will hve fully destructable architechture then isn't it?
An Tom's can kiss my ass, all they have showned is that they know nothing about game-physics...wich is what PhysX is for...not science-class...r as a render in science-experiments...

#12 - $300 is too much money. Most people don't even spend $300 on a new video card so it's crazy to think a lot of people are going to spend $300 on something that makes crap fall down or particles fly around.

You think that is all the PhysX API is about?
And do you say 1 PhysX card = bad, while 2 graphics cards = good?
I sell my old hardware(like GPU's) to lessen the cost of upgrades, so untill we get some numbers on the table, and not just PR, GPU-physics has no interest to me...

#13 - The whole physics "revolution" that we're on the brink of now might not amount to anything in the long run. The extra physics might not add much to gameplay or might greatly increase development time significantly that most developers will opt out of elaborate physics. It's all speculation at this point, so it's entirely possible that PPUs might go the way of the dinosaurs.

And it is possible that the PPU are here to stay for ever, Your are kidding yurself if you think dedicated hardware physcis is going away anytime soon...
But physics is getting more and more used.
The Doom eninge has more phycis than the previously engine from ID.
HL2 engine has more physics
More = bigger requirements of the hardware, and I like to offload my CPU(with GPU, soundcard, Nic PPU ect.) so I can enjoy my games to the max.

#14 - Multicore CPUs are going to become increasingly popular. While a dedicated solution may be better, the better solution is not the one that always wins (hi2u VHS).

Comparing Betamax and VHS to PPU vs CPU is soooooooooo bad.
The CPU dosn't even come CLOSE to the PPU.
Not even quad-core CPU's...

Terra - Sure you have read up about PPU's at all?

 

[Arcygenical]

Terra, I don't think the statement "Sure you have read up about PPU's at all?" was necessary. I mean, he does bring a few good points to the table, ones that... if nothing else, proves he HAS read up on the PPU.

Alot more than lots of others here on this forum.

 

[R1ckCa1n]

And it is possible that the PPU are here to stay for ever, Your are kidding yurself if you think dedicated hardware physcis is going away anytime soon...
But physics is getting more and more used.
The Doom eninge has more phycis than the previously engine from ID.
HL2 engine has more physics
More = bigger requirements of the hardware, and I like to offload my CPU(with GPU, soundcard, Nic PPU ect.) so I can enjoy my games to the max.

Don't you run a 7800Golden Sample or something like that with a slow CPU? Unless you are running SLI (which is clear you don't) there is no way you are running "MAX SETTINGS" unless your stuck at 1024x768 and below. If so, I wouldn't call that enjoying a game in 2006.

....maybe that is why you think everyone is GPU limited

 

[Terra]

Don't you run a 7800Golden Sample or something like that with a slow CPU? Unless you are running SLI (which is clear you don't) there is no way you are running "MAX SETTINGS" unless your stuck at 1024x768 and below. If so, I wouldn't call that enjoying a game in 2006.

....maybe that is why you think everyone is GPU limited

Linky
And then replace this image:

I found PAT(Asus Hyperpath) in my BIOS after I ran that test, so it should only improve...
Stock settings, so your post was again utterly wrong...

And I repeat:
Have ALL the different CPU vs GPU([H]ard, Anandtech, Xbitlabs ect.) articles never crossed your eyes?

Terra - Any more wrong things you would like to say?

 

[dotK]

Different opinions.

Of course some people will enjoy playing tech demos of falling blocks to justify their $300 purchase. Most people, however, want a bit more substance.

CellFactor - Revolutions will feature gameplay physics via LAN and WAN...

Becuase everyone LANs all the time... Great, so you have a game you can play at a LAN party, where it will probably become a sideshow at best while most people play the games they normally play online and are familiar with. The other 362 days of the year you won't be playing those physics intense games. I think everyone is significantly more interested in what can be done over the internet.

And a lot cannot...cloth, fludis ect.(see CellFactor)

In the end, they might be able to be simulated by the CPU as the code for CPU based cloth and fluids was not in place. I simply stated that a lot can be done by CPU. I never said that it would be able to do everything as well as a dedicated solution.

Look?
That must be a GPU/graphical design problem.
The PhysX renders NO graphics...
It acts like a real-time fluids...wich is what the PhysX is about.

I am not speaking of the graphical merits of the fluids but rather their behavior. Fluids in the real world, or at least the world I am posting from, do not form small globs and roll like bowling balls.

And they have a major absences of demoes ect.
They say they can...but we have to wait "9-12 months"...
Or even developers rallying behind them...

Strangely enough, that sounds JUST like the PPU. "Wait a few more months..."

To soon?
No hardware, no game support, simple as that.
What got old motnhs ago is the GPU-physcis PR...all words nothing to show.

And what of all the PPU PR? Where are all these game demos I can play that show off the power of the PPU and show how it will take gaming to the next level. Like the GPU physics demos, they don't exist. The best that's available is BoxFactor.

And you have the preformance data for that GPU-Physcis?
I would love to see it...so would the rest of the world...

And that has what to do with my point? Absolutely nothing. The point is that GPU based physics offers you options that the PPU never will. You're never going to be able to take a PPU and use it as a video card. No matter how you try to spin it, having more options is always a good thing.

PCI-E PhysX is comming, but I have no PCI-E, so I am quite happy PCI, perhaps people should get a mobo that siuts their needs instead?

Or Ageia should accomodate the customer and not the other way around? That's how it works in most industries.

The current driver is at 5.26MB.
Compare that to ATI or NVIDIA's driver and I (yet again) fail to see your point?

The point has nothing to do with the size of the driver. What it means is that the driver may be updated to the latest and greatest version featuring <whatever> but games that were based on older driver releases won't see any benefit from it. Only games built on the newest driver version will take advantage of that version. Just about every video card, on the other hand, experiences performance boosts from newer driver releases.

Elaborate please?

You don't have full control over how the game uses the PPU. Games will use the PhysX engine to interact with the hardware and as such will be limited by it. Of course, this could be changed in the future and more control could be given to developers.

Good thing CellFactor - Revolutions will hve fully destructable architechture then isn't it?
An Tom's can kiss my ass, all they have showned is that they know nothing about game-physics...wich is what PhysX is for...not science-class...r as a render in science-experiments...

Did you read the article as it was written or as you wanted it to be written? There are some very valid points in it that hold true for gaming. We'll see how the fully destructable architecture works out, because like so many things it's nothing but a promise at this point.

You think that is all the PhysX API is about?
And do you say 1 PhysX card = bad, while 2 graphics cards = good?
I sell my old hardware(like GPU's) to lessen the cost of upgrades, so untill we get some numbers on the table, and not just PR, GPU-physics has no interest to me...

At the end of the day, that's what physics is about. Physics in games will be about making stuff fall down and get bumped around. If it's not about this, then what is it about?
Most people I know hang on to hardware for a relatively long time to the point where selling it would yield very little. It would be nice to be able to use a nearly worthless video card for something instead of selling it for peanuts.

And it is possible that the PPU are here to stay for ever, Your are kidding yurself if you think dedicated hardware physcis is going away anytime soon...
But physics is getting more and more used.
The Doom eninge has more phycis than the previously engine from ID.
HL2 engine has more physics
More = bigger requirements of the hardware, and I like to offload my CPU(with GPU, soundcard, Nic PPU ect.) so I can enjoy my games to the max.

I never said that the possibility for PPUs to stay here forever was impossibile. Rather, I stated that there is a chance that they will dissapear. Crysis seems to be able to do a fairly good job of using the CPU for physics. If PPUs and GPUs for physics acceleration never really catch on, game developers will stop supporting it.

Comparing Betamax and VHS to PPU vs CPU is soooooooooo bad.
The CPU dosn't even come CLOSE to the PPU.
Not even quad-core CPU's...

It was just an example of a clearly better product getting beaten out by something inferior. Perhaps you would find GPU (VHS) vs. PPU (Betamax) less soooooooooo bad.

Terra - Sure you have read up about PPU's at all?

Yes, I have.

 

[TheGeekFreek]

Just for the sake of being on topic..

I have a Ageia PhysX I won at Quakecon.. and Hype is going to kill this card unless someone puts out something worth using it.. cause right now, there really isnt shxt.. a few games, most of which only get VERY minimal use of the card.. Cell Factor is a promising look at what the card does, but the Tech Demo was just flat out AWFUL.. Was lagging on MP against 4 bots on new ALIENWARE computers, Bots were retarded, LAN play would only go for 5 min and then it would crash, many just random crashes, long load times.. The list goes on.. Personally, I just want CF and UT2K7 and ill be happy..

-Evil Juggalo

 

[Terra]

Of course some people will enjoy playing tech demos of falling blocks to justify their $300 purchase. Most people, however, want a bit more substance.

If you think CellFactor is all I game, then you are mistaken.
Error #1

Becuase everyone LANs all the time... Great, so you have a game you can play at a LAN party, where it will probably become a sideshow at best while most people play the games they normally play online and are familiar with. The other 362 days of the year you won't be playing those physics intense games. I think everyone is significantly more interested in what can be done over the internet.

I seldom LAN or WAN.
And I told you CF does feature WAN?
Error #2

In the end, they might be able to be simulated by the CPU as the code for CPU based cloth and fluids was not in place. I simply stated that a lot can be done by CPU. I never said that it would be able to do everything as well as a dedicated solution.

The Cloth code is there for CPU
Error #3

I am not speaking of the graphical merits of the fluids but rather their behavior. Fluids in the real world, or at least the world I am posting from, do not form small globs and roll like bowling balls.

Wich fluids are we alking about?
Links please...

Strangely enough, that sounds JUST like the PPU. "Wait a few more months..."

The PhysX is out and used in games
Error #4

And what of all the PPU PR? Where are all these game demos I can play that show off the power of the PPU and show how it will take gaming to the next level. Like the GPU physics demos, they don't exist. The best that's available is BoxFactor.

The following games are either OUT or comming this year, with PhysX support:
Ghost Recon: Advanced Warfighter
Rise of Nations: Rise of Legends
Bet on Soldier: Blood Sport
Cell Factor
City of Villains
Unreal Tournament 2007
Gunship Apocalypse
Sacred II
Loki
Dogtag
Fallen Earth
Crazy Machines 2
Arena Online
Diabolique
Warhammer MMORPG
Eye of the Storm
KARMA
Vanguard: Saga of Heroes
Alpha Prime

And that has what to do with my point? Absolutely nothing. The point is that GPU based physics offers you options that the PPU never will. You're never going to be able to take a PPU and use it as a video card. No matter how you try to spin it, having more options is always a good thing.

We know NOTHING of the full feature set or the preformance for GPU-physcis.
Error #5
And with only 0.77% af gamers using multi GPU's a lot will need to buy new cards and new mobo to get GPU physics.

Or Ageia should accomodate the customer and not the other way around? That's how it works in most industries.

You mean like NVIDA allows SLI on Intel Mobo's?
Error #6

The point has nothing to do with the size of the driver. What it means is that the driver may be updated to the latest and greatest version featuring <whatever> but games that were based on older driver releases won't see any benefit from it. Only games built on the newest driver version will take advantage of that version. Just about every video card, on the other hand, experiences performance boosts from newer driver releases.

You mean like SM.20 games can't take advantage of SM3.0?

You don't have full control over how the game uses the PPU. Games will use the PhysX engine to interact with the hardware and as such will be limited by it. Of course, this could be changed in the future and more control could be given to developers.

What part of the PhysX API dosn't the develpoers have control over?
I'll look in the SDK for you then?

Did you read the article as it was written or as you wanted it to be written? There are some very valid points in it that hold true for gaming. We'll see how the fully destructable architecture works out, because like so many things it's nothing but a promise at this point.

A promise from the devlopers wich words I take over yours any day.
I don't ask a janitor about atronomy, I don't ask a laywer about biology and I don't ask a sciencetsit about game-physics(I would ask a developer), that why Tom's is retarded...

At the end of the day, that's what physics is about. Physics in games will be about making stuff fall down and get bumped around. If it's not about this, then what is it about?

Collions (as you mention)
Trigger events
AI sensor
Fluids
Smoke
Cloth
Destraucatable architechture
Friction
The list is actaully quite long...
Error #7

Most people I know hang on to hardware for a relatively long time to the point where selling it would yield very little. It would be nice to be able to use a nearly worthless video card for something instead of selling it for peanuts.

I upgrade every new generation, and sell my old hardware..loook in the Fr Sale subfrum...I am far from alone.
When I got my 7800Gs, I sold my 6800GT for 33% of what I gave for it and around 40% of what my new card coted...I can se that is a wrthless price

Error #8

I never said that the possibility for PPUs to stay here forever was impossibile. Rather, I stated that there is a chance that they will dissapear. Crysis seems to be able to do a fairly good job of using the CPU for physics. If PPUs and GPUs for physics acceleration never really catch on, game developers will stop supporting it.

The Physcis in Crysis is limited cmpred to CF, look at the videos again.
Tree's fall over...anbecome harder than cement to shot...
The hardware spcs are also fun and nothing is know about the preformance.
Remeber FarCry...it slammed all hardware at the....crysis will be no different
Error #9

It was just an example of a clearly better product getting beaten out by something inferior. Perhaps you would find GPU (VHS) vs. PPU (Betamax) less soooooooooo bad.

Not really, as neither you nor I have any clue to the preformance as ATI/NVIDIA/Havok has delivered ZIP.
You don't see developers flocking to suooprt GPU-Physics...
They (+60) on the other hand suppport the PhysX...
And a bad analogy is still a bad analogy.

Yes, I have.

On Tom's?
#Error 10

Terra - 10 strikes and you're out

 

[Terra]

Just for the sake of being on topic..

I have a Ageia PhysX I won at Quakecon.. and Hype is going to kill this card unless someone puts out something worth using it.. cause right now, there really isnt shxt.. a few games, most of which only get VERY minimal use of the card.. Cell Factor is a promising look at what the card does, but the Tech Demo was just flat out AWFUL.. Was lagging on MP against 4 bots on new ALIENWARE computers, Bots were retarded, LAN play would only go for 5 min and then it would crash, many just random crashes, long load times.. The list goes on.. Personally, I just want CF and UT2K7 and ill be happy..

-Evil Juggalo

Are you talking about CellFactr R36?
Becuse I don't have the issues you are talking about?
But nice to hear that 2 games will make you jhappy...
Some poeple sound like unless it's 100% support in all games, they wil never be happy

Terra...

 

[SuperGee]

Its the software that bottlenecks the graphics cards, after all, you want it to really
Your graphics card is always going to be pushed to the max no matter how fast it is.
Perhaps you meant that the PhysX API/Hardware isnt developed enough to interface properly with the PC environment?

Well it's now know that PPU isn't for more FPS but for Much more PhysX. Because that use the computational power where for the dedicated hardware is.
But only Ageia PPU is reviewed.
Ageia solution is a Triangle CPU PPU GPU but more CPU PPU CPU GPU. For the Gameplay PhysX loopback. PPU takes some time to compute.
Havok FX is also like That CPU pxGPU GPU triangle. Also a extra Physics feed to the pxGPU wich put its result to the GPU. CPU has a higher load. But the pxGPU must compute to, so taking alos time away from the total frame time. For seting up the next frame. This is the same for both solution's
Less time for rendering. Could be impact on FPS to. At least a slightly dip can be expected. And just like Ageia havok must optimize Havok FX also in time. Its so new.
How ever no HavokFX game in the near future so no hand son result of the impact of havokFX on FPS. But if it take away some FPS you get mucho effectphysx eyecandy for back. Not a bad bargin.

Faster graphics cards are coming and if you have the cash, its available now in Crossfire or SLI. Is SLI or Crossfire not fast enough?
Other ways of processing Physics are coming too so I'm wondering if PhysX will survive?

After the wining on that $300 Price tag yes SLI CF make sense for rendering $$$$

A difficulty with the PPU is thats its being placed in a system where it needs extra CPU and GPU power to operate. This is unfortunate as something will have to give or be upgraded as has been shown.

I do expect the same for pxGPU havokFX solution. FX is a very high setting for HavokFX games. On top of a heavy loaded game already. The CPU must feed the GPU inderectly the render load with the effectPhysics load to the pxGPU first. Wich when done feeds the GPU with a extended render load. This extra load is more burden on the CPU pxGPU take time away needs time to compute and the GPU get a much heavier but PX eyecandy rich render load.
Wich will have some impact on FPS.

If it could really take workload off the CPU or GPU that would be great but the addition of PhysX today does indeed place extra load on everything.

GPU Physics is not about offloading CPU while competing with PPU.
PhysX for CPU is mostly gameplay wich is very common because every rig has a CPU and the most basic used Physics are often gameplay Physics. Physics Havok FX can not offload. Only effect PhysX when used.
For competing with PPU it misses the gameplay physics option and need a heavy load to compete with PPU but on effects Physics only.
Buying a PPU to just offload the CPU make no sense.
Off loading CPU with one GPU for Rendering and some minor Physics is a option, but don't bring much to the table.

Physics processing isnt "here to stay", it has always been here and integrated into gaming. This hasnt stopped because PhysX has appeared, so games without a PPU can and will have very good physics.

Well that not the point. Physics in games is even older then 3D games but with each CPU gen geting more powerfull the more PhysX can be done. Like in steps of 2X to 5X times. More PCU cycle to spare for Physics.
But its not about Physx but about Hardware dedicated accelerated Physics. Wich means Physics make a large step forward.100X And with ageia and now also HavokFX and aTI and nV jumping on it. It's comming but who will stay on top for now, the short future the longrun is unknown. Espect that Some games in the near future will have a lot more physX mostly in effects way.

Now AMD have bought ATI and are researching Physics processing, we may see hybrid CPU's and GPU's with a PPU onboard. I'd certainly expect something.

A PPU or Shadere array on a socket AM2/3 nice. Will this bring a new API. Be PPU compatible or a Copro consisting out of SM4,0 shaders a IO PU to manage it. CeLL PPU like. Wich has such general purpouse PU on it to manage the Specialises PU.

Nice Idea. But will get mrketing ready long After Havok FX games hit the market. Like a striped down GPU == PUSPU (Phyxiics Unified shader PU)
Imangine with 90nm the can put 64 US on one die. strip the G out of GPU more shaders. Plus a IO controler + mem controler maybe 128 UShaders.

With MS's Physics implementation coming up, we may be able to use all of these and a PhysX card at the same time for max effect.
Who knows what NVidia and Intel can bring to the table as well!

That would be coul no API licens just MS utelizing hwtaever detected. And seting up the way you want it.

 

[dotK]

Blah blah blah blah, blah blah blah, blah

If I'm at 10 strikes and out, you're at 1,000 strikes and out of this solar system with how much crap you've posted. "Where?" you're going to say - use the search feature. Now that we've gotten this childish thing out of the way, it boils down to this:

There have been many promises made as to what the PPU can do. A lot of these claims have come from gamers themselves and are not so much facts as hopes, but could be true. It is currently unknown exactly how the PPU will impact gaming because no games have been released that take full advantage of it. Current games that are out and support have very limited support, as is the case with GRAW.

BTW, have you played Bet on Soldier? It sucks. It was a bargain bin game that nobody heard about or cared about before the developers added PhysX support for a reason. If you want to promote how "awesome" the PPU is by giving a short list of games supporting it I would suggest avoiding. Also, nobody cares how many games are on that list but rather how many good games are on it (and use more than just effects physics).

 

[Chernobyl1]

Well it's now know that PPU isn't for more FPS but for Much more PhysX. Because that use the computational power where for the dedicated hardware is.
But only Ageia PPU is reviewed.
Ageia solution is a Triangle CPU PPU GPU but more CPU PPU CPU GPU. For the Gameplay PhysX loopback. PPU takes some time to compute.
Havok FX is also like That CPU pxGPU GPU triangle. Also a extra Physics feed to the pxGPU wich put its result to the GPU. CPU has a higher load. But the pxGPU must compute to, so taking alos time away from the total frame time. For seting up the next frame. This is the same for both solution's
Less time for rendering. Could be impact on FPS to. At least a slightly dip can be expected. And just like Ageia havok must optimize Havok FX also in time. Its so new.
How ever no HavokFX game in the near future so no hand son result of the impact of havokFX on FPS. But if it take away some FPS you get mucho effectphysx eyecandy for back. Not a bad bargin.

sorry I'm missing the point you are making.

After the wining on that $300 Price tag yes SLI CF make sense for rendering $$$$

Terra stated that we dont have enough graphics processing power and that is the bottleneck in the PC. I agree, but it doesnt look good for PhysX.
If another graphics card is needed to run a PPU then the current generation PPU cannot perform its function properly unless you buy SLI/Crossfire and even then Terra isnt happy that is enough.
The cost is going to be more than the already expensive $300 and it doesnt do a lot yet.
I agree, that no matter which way Physics is performed, more graphics power will be needed. Ageia dont mention that on their advertising though so not only do you pay up to $300 for a card that cant be used for much yet, you also need SLI graphics using the fastest cards possible to get the best out of it.
The PhysX solution just isnt worth it at the moment.
If you buy one without the above knowledge you will be disappointed but unfortunately this information is missing from any sales pitch.
If part of physics processing is handled by a graphics card (ie there are extensions to the graphics card to handle Physics), the graphics card may not need to be much more powerful as Physics results can be applied faster and directly to where they are needed.

I do expect the same for pxGPU havokFX solution. FX is a very high setting for HavokFX games. On top of a heavy loaded game already. The CPU must feed the GPU inderectly the render load with the effectPhysics load to the pxGPU first. Wich when done feeds the GPU with a extended render load. This extra load is more burden on the CPU pxGPU take time away needs time to compute and the GPU get a much heavier but PX eyecandy rich render load.
Wich will have some impact on FPS.

I cant understand everything you have written but I'll do my best.
Are you giving reasons why PhysX isnt very good or are you saying something else?
By the time the Physics API is ready for CPU/GPU, you can be sure it will be useable.
PhysX was released before its ready unfortunately.

GPU Physics is not about offloading CPU while competing with PPU.
PhysX for CPU is mostly gameplay wich is very common because every rig has a CPU and the most basic used Physics are often gameplay Physics. Physics Havok FX can not offload. Only effect PhysX when used.
For competing with PPU it misses the gameplay physics option and need a heavy load to compete with PPU but on effects Physics only.
Buying a PPU to just offload the CPU make no sense.
Off loading CPU with one GPU for Rendering and some minor Physics is a option, but don't bring much to the table.

I didnt say thats what Physics was about, you made that assumption.
Physics for CPU is not mostly gameplay, where did you read that? It will differ with the requirements of the game/app.
Adding a PhysX card will put extra load on the CPU and GPU as mentioned already.
You said "Buying a PPU to just offload the CPU make no sense.".
As you can see I am fully aware of that.

How do you know that a CPU/GPU solution wont be good enough?
The hardware isnt here yet and the fact that both ATI and NVidia are looking at making a product suggests that it will be useful and sale-able.
Also to assume that the current performance of CPU's and GPU's for Physics is what will be sold as a Physics solution isnt correct. You can be sure there will be good Physics acceleration built into either the CPU or GPU or both.
My gripe is that the PhysX product for sale isnt complete and theres little to no use for it.
We have had owners of the card say the same thing here.

Well that not the point. Physics in games is even older then 3D games but with each CPU gen geting more powerfull the more PhysX can be done. Like in steps of 2X to 5X times. More PCU cycle to spare for Physics.
But its not about Physx but about Hardware dedicated accelerated Physics. Wich means Physics make a large step forward.100X And with ageia and now also HavokFX and aTI and nV jumping on it. It's comming but who will stay on top for now, the short future the longrun is unknown. Espect that Some games in the near future will have a lot more physX mostly in effects way.

I'm not sure what point you were expecting, but I made the point I wanted to make. It was addressed to Terra so maybe is out of context for you?

What you have written is confusing, I have read it a number of times and will do my best to answer.
Its true, Ageia have accelerated the growth of Physics for computers which is a good thing.
We know Physics is coming, I'm not sure why you told us that.
Whether it be hardware or software based is irrelevant, whatever is available for processing physics should be used.

How is this hardware solution different from a software solution anyway?
I will answer...

The PhysX card is really a restricted hardware/software solution because there isnt a way to program the PPU.
The PPU itself is programmable, so anything written for it is software. It is a co-processor for Physics.
Its the interface that we get which doesnt allow the PPU to be programmed, only the set API Ageia have made is presented to the programmer. This makes it look like a hardware solution but it is still added hardware with software that could be more functional than it is.

A locked down hardware solution is great if the algorithms are final and tested. If the algorithms are not final, they will be subject to change so this should be made as easy as possible. The Physics algorithms are undergoing optimisation for use in computer simulations and need to be able to change.
Developers have already expressed that they would like to be able to make their own algorithms/optimisations to the code running on the PPU.
Ageia have omitted a very important part of the interface to their product by not allowing the PPU to be programmed. Hopefully this will be changed later

Who will stay on top?
regarding Physics, it could be any of them.
Integrating Physics processing into an already successful product (CPU and/or GPU) is surely the easiest way to address the problem, the cheapest too.
Having the Physics processing close to the CPU and GPU will alleviate bottlenecks and allow the hardware to be used more efficiently.
Ageia can survive if they get a useable product to market in time.
Whats out now isnt much use.
It was hard to understand if you were making a point, I hope I have answered it.

A PPU or Shadere array on a socket AM2/3 nice. Will this bring a new API. Be PPU compatible or a Copro consisting out of SM4,0 shaders a IO PU to manage it. CeLL PPU like. Wich has such general purpouse PU on it to manage the Specialises PU.

There are quite a few API's for Physics in development. Do we need another new one?
If a physics co-processor is made, there is going to be a driver and API for it.
Is that what you were asking?

Nice Idea. But will get mrketing ready long After Havok FX games hit the market. Like a striped down GPU == PUSPU (Phyxiics Unified shader PU)
Imangine with 90nm the can put 64 US on one die. strip the G out of GPU more shaders. Plus a IO controler + mem controler maybe 128 UShaders.

I cant understand what you are saying.

That would be coul no API licens just MS utelizing hwtaever detected. And seting up the way you want it.

Yeah agreed

 

[HOCP4ME]

I think that AGEIA needs to sell a CF:R + PPU bundle for $150. That would get everyone to buy it if it was a good game, and then developers would know that many gamers had a PPU and would develop for it.

I agree that $300 is a lot of money, but I can see one good reason to buy a PPU right now. To make it useful someday. After all, the PPU has power that, if used be developers, can make games much more realistic and immersive. But, the two biggest reason why AGEIA might fail are:

1. They don't sell enough hardware.
2. No developer support because the hardware isn't mainstream.

If every gamer were to buy a $300 PPU right now, both of those problems would be gone, and we would see widespread adoption within one year. But that's not how it works, of course. Software support always has to come first before anyone buys anything. And $300 is pretty expensive, I do think it should be cheaper.

 

[nhusby]

The smartest answer to a question you dont know is: I dont know.

But we are not that smart... thats why we are here...

Will Ageia succeed? My wishfully thinking crystal ball suggests yes. But it has most definetly been wrong before.

I dont think hype will kill it... But I do think that dissapointment could. If too many people buy the card and dont recieve what they expected, then the words of bad experiences will spread. The unfortunate part of this is that its not ageia's fault. Ageia made more powerfull physics possible, if develops dont use it, or dont use it well, then the failure falls on ageia...

 

[SuperGee]

@nhusby
Oh yes I agree with that. The games deside who will win. But it take some time to reveal.