PPU's Dead

[Caffeinated]

At this point, I agree that the PPU will never become mainstream. I hope it does, as another option to customers, but at this point it looks doubtful.

Multi-core CPUs are not what will kill (has killed?) the PPU, however. It is GPU physics that will prevail. GPU physics are still something that everyone with a modern video card can (or will be able to in the future) take advantage of, and they outperform multi-core CPUs by a long shot.

It seems that a lot of people here are opposed to GPU physics and want CPU physics instead. I don't see why. You won't have to buy anything (other than having a DX10 graphics card, and by the time hardware physics becomes mainstream, you will), and the system will outperform one based on even a quad-core CPU.

Quad-core CPUs may be "good enough", as someone said at the beginning of this thread, but GPU physics are much better and almost as easy.

I also don't understand why some people say "we don't need hardware physics acceleration". WIth GPU physics, you don't have to buy anything, you don't need extra slots in your computer, and you get a better overall gaming experience. Who cares if we need it? Since it only makes things better, why not take it anyway?

That's a really good point. I guess it boils down to not *having* to buy a separate piece of hardware. If it can be done directly on my GPU, then I'm all for it. If it uses one of my processor cores, I'm all for it. When you ask me to spend $200 and clear some space in my crowded case for another piece of hardware, I'm not too excited about that.

 

[djBon2112]

That's a really good point. I guess it boils down to not *having* to buy a separate piece of hardware. If it can be done directly on my GPU, then I'm all for it. If it uses one of my processor cores, I'm all for it. When you ask me to spend $200 and clear some space in my crowded case for another piece of hardware, I'm not too excited about that.

Amen to that!

 

[theelectic]

With GPU physics you're back into the effect vs. gameplay physics debate again. If GPU physics can do gameplay and effect, great. If not, dedicated hardware will have a place.

 

[devman]

Imo it would be nice if the PPU just went the way of the 3D accelerator and the whole thing to folded in to a new generation of graphics cards.

 

[djBon2112]

^^ Why wouldn't it do gameplay physics though? If it can do one it can be made to do both.

 

[TeamStrykerCore]

Imo it would be nice if the PPU just went the way of the 3D accelerator and the whole thing to folded in to a new generation of graphics cards.

This is exactly what's going to happen.

It's like trying to have 5 optical drives

1 to write dvd's
1 to read dvd's
1 to write cd's
1 to read cd's
1 to do lightscribe

Are people going to buy those devices...

or the $37 device on newegg that does everything even thought it might not be the best at doing one function.

 

[theelectic]

^^ Why wouldn't it do gameplay physics though? If it can do one it can be made to do both.

Not really, I forget the exact reason but it's touched upon in this thread:

http://www.hardforum.com/showthread.php?t=1122440

 

[SuperGee]

i wouldnt call their first attempt lightweight.

I do. Why.
iNtel is long on 65 en do 300 ? Transistor solutiuons.
nV is on 90nm en do almost 700Mil transistor solution.
while PPU is on old 130nm just using 125Mil of transistors.
Emagine a 500 mil tranistor solution on 90nm pulling 60Watts. With 512MB
I think starters start down the road. 130nm it is. Maybe with nextgen the may use a bit modern dieshrink, 65 from TMC? while iNTel is moving to 45nm
GPU get there Physics power out off there shader. Lets say halve of the die would be use for them. Means a G80 have 350 maybe more transistors for GpGPU power.
And a lot more mem. R600 will have 1GB vs 128 max of the PPU. To store local physic data.

That's a really good point. I guess it boils down to not *having* to buy a separate piece of hardware. If it can be done directly on my GPU, then I'm all for it. If it uses one of my processor cores, I'm all for it. When you ask me to spend $200 and clear some space in my crowded case for another piece of hardware, I'm not too excited about that.

You don't have to buy. But others will. Most PC using people won't spent much on there game rig. Your not alone in that. But that's not relevant to ageia. It's the enhaustiast market that buys hardware they need/ could use or want for playing there games as it was meant to play in optimal settings. And in case of big sucses it could become mainstream but doesn't have to. The could let that over to CPU. Or shared GPU use. The budged solution to Phsyics.

This Audience market share is small compared to the masses. But if it still is like 5miljon to 500 miljon it stil a profitable group. So if 1 out of 10 see something it it. The ball roles for ageia.
That why the KillerNic is only aimed at the hardcore gamers market. It could be profitable.
Thats the bussness model. Whats your target market.
Surtenly not the cheap people. They are the once using Dual and later quadcore for low Physics settings. And they settle for that. That is choice. Or think that one GPU can do it. Surtenly it shares resources so it a budged solution. Means low Physics.
While people who choose to spent money on dedicated HArdware for Phyisx/HavokFX
Can play at way hihgher Physics settings. Well Hardware vendors also make produkts for the high-end hardcore gamers market and hardware enthausiast. Wich will spent money on things they can use.
While the value and usefull ness by the lack off games is currently low. That just a starters problem. Very normal. More games to come so it will change.

While a PPU game need less settings options, so can do it on a High setting. It need les options then a havok FX game beause there are a lot of flavors G-card and GPU's.
A Gamer with a dedicated G-card depending on model could put on medium setting for a X1800Xl. R600 on Ultra high. While Quadcore is low and Dual is ultralow.

And what wil Dev do. Well the basis will be CPU. Mainstream will be
Dual core. So what a midrange Dualcore can handle Devs set that as minimum target for Physics option.
Level 1
Quadcore is level 2 to 3.
Shared GPU is level 2 to 4
Dedicated GPU there is so wide range it would be from 2 to 10 max.
PPU would be level 6 to 7

So to play optimal a PhysX game you need a PPU.
To play a Havok FX Game in supper high settings you need a second dedicated for Physics R600 or G80/81 To pull the phsyics slide way up to max or Ultra high setting.
And all Physics checkboxes on.
Like Cell factor have checkboxes for cloths and fluids.

So it boils down to it just your choice to spend but you are just one who chooses. There are Miljards of people. And Dev's give you choice in a option setting to play there games you want to cheap or with all the power it support.

think of Oblivbion a game for directX 9 wich could use the power off DX10 hardware.

Games compete to. Some will want to stand out of the crowed and support new and powerfull tech like Quadcore PPU and R600 or G80 for Physics.

 

[ghost6303]

I do. Why.
iNtel is long on 65 en do 300 ? Transistor solutiuons.
nV is on 90nm en do almost 700Mil transistor solution.
while PPU is on old 130nm just using 125Mil of transistors.
Emagine a 500 mil tranistor solution on 90nm pulling 60Watts. With 512MB
I think starters start down the road. 130nm it is. Maybe with nextgen the may use a bit modern dieshrink, 65 from TMC? while iNTel is moving to 45nm
GPU get there Physics power out off there shader. Lets say halve of the die would be use for them. Means a G80 have 350 maybe more transistors for GpGPU power.
And a lot more mem. R600 will have 1GB vs 128 max of the PPU. To store local physic data.

the manufacturing process and transistor counts have nothing to do with anything. PPUs arent supposed to be a CPU. they are designed for a function, the 130nm process fits what they are designed for and thats what was used. could it be more efficent? yes. will the die size decrease in the future? probably. this doesnt mean they dont perform right now. intel has been around for decades, its taken that long to get to the 60nm process they are using today. for a pioneer of new technology, a 130nm process is a feat to say the least if you know how its made. dont judge a book by how many pages it has (or how small the pages are).

also when you look at the pricetags:
iNtel is long on 65 en do 300 ? Transistor solutiuons. $150-800
"nV is on 90nm en do almost 700Mil transistor solution." $200-650
"while PPU is on old 130nm just using 125Mil of transistors." $100-200

they are priced accordingly, not that im saying it warrents paying $179 for a PPU that has no use due to lack of software support, but the numbers are not out of skew for the technology.

think of Oblivbion a game for directX 9 wich could use the power off DX10 hardware.

ummm...what? any game could use DX10 i guess- but 99% of the hardware on the market isnt ready to support it, so they made the right choice and went with a platform 100% of their customers could play on vs. a platform 1 in 500 people could use.

 

[djBon2112]

think of Oblivbion a game for directX 9 wich could use the power off DX10 hardware.

I agree with the above poster, they made the right desicion. Now, whether they release a DX10 patch when it becomes more prevelant (a la Flight Sim X), remains to be seen.

 

[SuperGee]

the manufacturing process and transistor counts have nothing to do with anything. PPUs arent supposed to be a CPU. they are designed for a function, the 130nm process fits what they are designed for and thats what was used. could it be more efficent? yes. will the die size decrease in the future? probably. this doesnt mean they dont perform right now. intel has been around for decades, its taken that long to get to the 60nm process they are using today. for a pioneer of new technology, a 130nm process is a feat to say the least if you know how its made. dont judge a book by how many pages it has (or how small the pages are).

also when you look at the pricetags:
iNtel is long on 65 en do 300 ? Transistor solutiuons. $150-800
"nV is on 90nm en do almost 700Mil transistor solution." $200-650
"while PPU is on old 130nm just using 125Mil of transistors." $100-200

they are priced accordingly, not that im saying it warrents paying $179 for a PPU that has no use due to lack of software support, but the numbers are not out of skew for the technology.

You don't get the point. A 65nm 400mil PPU cost almost the same as a 125Mil on 130nm. because on one waver they get the same amount off chips.
But the difference is in computing power.
That Lightweight while it could have more crunchin power. For the same $

ummm...what? any game could use DX10 i guess- but 99% of the hardware on the market isnt ready to support it, so they made the right choice and went with a platform 100% of their customers could play on vs. a platform 1 in 500 people could use.

Point is some Dev whil push the limit. Some will use Physics settings only a PPU or R600 or G81 can handle. Some exception will go beyond that.

which means you can play it optimal till the hardware is out for it.

Like Play Oblivibion on a 2560x1600 screen you need a 8800 or X1950XT CF solution
Without AA and options max out.

While most devs aim at a mainstream target market.

There are Dev's that push the hardware and Dev's that aim at easy volume sales.
Keep the game requierment low enough to reach a large audience.

But the Games that push the hardware are the once that stand out.
But you have to buy hardware for them to play them in recommended or optimal settings.
A choice to be made.

 

[griff30]

It's AVP3 or the PPU is the next CueCat.....

Actually I would like to know if you guys can tell me what would keep AMD or Intel or even a start up from placing a PPU on die like the FPU or ALU we already have.

Also I would like to know what you peeps think of the new 80 (!!) core CPU Intel is showing. Obviously each core is general purpose but 80 !! JEBUS!

We wont see 80 cores anyday soon but maybe 20 in a couple years. (hope)

 

[SuperGee]

Actually I would like to know if you guys can tell me what would keep AMD or Intel or even a start up from placing a PPU on die like the FPU or ALU we already have.

* First PPU is Patent protected using native API PhysX or former Novodex.
So to see PPU on die anny where. Ageia must join a CPU maker or make its own CPU solution to combine with PPU on the same die. Just like 3DFX slap there own 2D part to there 3D produkts.
Or AMD was planning to support third party CoProc's. In to a Open Socket. you could stick a PPU with AM4 Socket in to suchs socket next to a AMD CPU in to just onother such socket. HT universal open chip interconnect tech.

* Second AMD has ATI Wich use shaders for Physics and Intel is already in the low budged GFX industry leader and are planning to storm the high-end market to. I think iNTel and AMD are going to use a dirative of shader tech with full general purpouse Co procsesing units wich are good at Physics and AI and other Parralizing computing stuff. You get something like a X86 EMT64/AMD64 Cell like something.
Because Shader from SM3.0 and up can do Phsyics also well. They don't need Ageia Patented tech and off course PhysX and HavokFX could be replaced by DirectXPhysX but that far into the future. It will be CoProc unit supporting or comform the DirecXPhysX spects. Wich supports Shaders and Whatever hardware solution is around by then. Including PPU. Where Dev's can choose between PhysX, HavokFX or DirectPhysX. API don't disapare overnight it need some migration and maturing.

* Third is it relevant for the next 2 to 3 years.
If I want to build a rig within the next year or so for 3 years of duty. Do I care what whatever company existing in 2011 whil have on the market? That relevant for the big upgrade or future new PC wich would replace what you buy now or within the next year.

Well I do think that at a time they put dozen off cores on one CPU and for different purpouses and mixes. But that far far away.

I focus on the near future. And the nextgen games to come. Not the NEXT next next gen games. PPU will be history just like a G80.

The PPU don't compete with what ever comes out in the far future.
It competes with ATI x1 and X2 series and nV 7 and 8 series. The now and very near future.

 

[griff30]

SuperGee,
Thank you for your reply!
And I meant this in the most complimentary way because English is probably not your first or even second tongue, so thank you for your taking the time to reply.

 

[SuperGee]

"It's AVP3 or the PPU is the next CueCat....."
Yah, have no clue what this means?

 

[TeamStrykerCore]

I focus on the near future. And the nextgen games to come. Not the NEXT next next gen games. PPU will be history just like a G80.

The PPU don't compete with what ever comes out in the far future.
It competes with ATI x1 and X2 series and nV 7 and 8 series. The now and very near future.


That my friends is exactly what's going to happen. Also when Micro$oft get's a hold of the physics market and develops their own API / builds it into direct x forget any competition. Remember openGL folks?

Thank jeebus for America and it's capitalism *rolls eyes*.

 

[SuperGee]

Oh yes MS would do that but not in the near future, more the far.
The near future are first PhsyX and then some HavokFX titles and growing in numbers within few years. PhysX started in 2006. HavokFx wil start to compete in 2007. And in End 2007/ 2008 wil be the batle.
DirecX PhysX comes with a update or new version of DirectX. Miscrosoft just started with it possible in end 2006, they had job's open for it.
While PhysX is derived from novodex. And HavokFX from Havok.
Ms start from the ground up. Or buy some firm out There are so manny Physics API.

The positive with DirectX it's hardware independant. With DirectX performance will be the key to succes becuase choice would be more based on that.
Now it is what dev's choose to support. PhysX or HavokFX. Each solution needs its specific hardware. PPU and Gp-G-Cards. So depends on wich games gamers gonna like.
With DirectX if it the main standard. Gamers can choose what performs best for it's money and games.

PPU2 could have a DirecX driver to.

Just Like PhysX API, DirecPhysX SDK will be free.

It will hurt Havok more. DirectX competes With Havok. While Ageia could focus on Hardware make that there core busness.

 

[griff30]

"It's AVP3 or the PPU is the next CueCat....."
Yah, have no clue what this means?

Sory, I'm saying that Aliens Vs Predator 3 (or AVP3 for short) could be the PPUs "killer App" and that without it, the PPU will become the CueCat. CueCat was an overhyped barcode scanner that resembled a cat and connected to your keyboard, you scanned a barcode and it redirected your browser to that products website, with ads paying for the use of course.
AVP3 could be the killer app as it could have all the interest of FPS game types and RPG with thousands of objects interacting on screen in a MASSIVELY online fps/rpg GAME.
I would go as far as to say half the readers here would buy a copy to play and a PPU to go with it if AVP3 was well written.

 

[RamonGTP]

Good riddance I say. A dedicated PPU card is a total waste of space and money IMO. With the combination of CPU and GPU based physics processing that is now available (although not quite implimented) there is absolutly no good reason for a dedicated solution. With the big players like ATI, nVidia and Havoc turning their backs on a dedicated PPU card, I don't EVER see this thing getting off the ground. The vast majority of the time I've seen the PPU in action it was dissappointing, to the point of I would rather not even run one if it was given to me for free. Very little difference in the gameplay with a very significant performance hit.

 

[SuperGee]

Wrong. First GPU go the Hardware accelerated extra card route to.
You can use a extra G-card for Phsyics in some nextgen HavokFx games.
Wich is just the same way as PPU needing a slot.
Optional at a performance cost you can share one G-card for Phsyics and rendering. Wich come at a cost. Bit more Phsyics but less render power.
To compete or use full Phsyics settings agians PhysX. With HavokFX you need a dedicated G-card to.

And yet agian PhysX isn't about FPS. Altho reviewers only bench tings to FPS.
PhysX and HavokFX is about more Physics then a CPU can handle. Enhancing games more then is possible without it.

And agian Ageia and Havok are just starting its new. It takes time till enough games are around where a few stand out and give there choosen solution more or enough value to gamers to buy them.

 

[RamonGTP]

WIth GPU's being as powerfull as they are, not to mention SLI and Crossfire setups, I'll take the small rendering hit anyday. I like the flexibility as well as the better support GPU based physics has going for it.

 

[theelectic]

I like the flexibility as well as the better support GPU based physics has going for it.

That cracks me up. WHAT GPU based physics support? Show me one game that you can play right NOW that supports GPU physics. I'm not a PhysX !!!!!! but at the very least they have actual working games to support their hardware, no matter how crappy they are.

 

[RamonGTP]

If you had read my previous post, you would see that I already mentioned that it hasn't been implimented yet... Dispite that, it still has wider support than the PPU does. ATI, nVidia and Havoc are all behind GPU based physics... And PhysX has an unkown company by the name of Ageia. Does that clarify my defination of support for you?

So you can go ahead and spend your money on a dedicated PPU with crappy support, crappy results and generally crappy games that you can play now... I'll simply wait it out until GPU physics processing becomes a reality, which won't be long. Meanwhile, while I'm waiting, i'll still enjoy better performance from my games than you, minus some extra shrapnel bits from explosions that last about half a second.

 

[SuperGee]

If you had read my previous post, you would see that I already mentioned that it hasn't been implimented yet... Dispite that, it still has wider support than the PPU does. ATI, nVidia and Havoc are all behind GPU based physics... And PhysX has an unkown company by the name of Ageia. Does that clarify my defination of support for you?

So nothing is better then something? How can you judge something that is just so short out now, on its own new created market, as bad. While it still must mature in Games numbers, wich Devs have to learn to use it the right way. That take time to. The better once alway come later.
Havok FX have nothing. And it's common sense that the large Pack off general games that come out are crap. Like more of the same old or bad copy of some idea clone in a dozen. It's about the games that stand out. Like 3 to 5 out of 25. So the chance that something good comes out of 5 games is small, Ageia stance now. Get bigger with 10 get more better if all 120+ are out. And the chance of a Killer app between them grows with numbers. While HavokFX have to grow in numbers from zero.
Same count for HavokFX. Will the first one be that killer game. I don't think so. Could it be a tad better then graw that chance is a bit bigger. But have to deal with the same quirks Dev industry have to deal with.

So you can go ahead and spend your money on a dedicated PPU with crappy support,

PhysX its free support is large Lot's of games in produktion. Doesn't mean that those games are out yet it take some time.
It aviability of games out in retail in the gamers market will grow. So it's worth, value and usefullness will grow.

crappy results

Yah so its the start period. That can be a tough time wich go by eventualy. But it getting better in time. Wenn some better games are out Graw will be forgotten. Like I don't know the first Glide game for the fresh Voodoo Original.
If we look at GPU now it zero noting nada. If you judge your favorite agianst Ageia to the same crierium. NOW. GPU and Havok FX is even worse. Its Noting. But you judge Both solution differently.

and generally crappy games that you can play now...

More games to come out more chance to something good.

I'll simply wait it out until GPU physics processing becomes a reality, which won't be long.

That wierd. No waiting for PPU but for GPU and Havok you sure take the time.

Meanwhile, while I'm waiting, i'll still enjoy better performance from my games than you, minus some extra shrapnel bits from explosions that last about half a second.

No no no.
PhysX isn't about FPS. You need enough FPS not 10000 frames per second 30 to 60 is enough. So before you think about slapping a PPU or extra G-card in it for Phsyics you need already a strong Gameing rig to get the best out of it. A System with already decent FPS. So PPU is a addon on a decent configed gamerig wich already have decent FPS in games. And to enhance your game experience you put a PPU with it.
Loosing or gaining a bit FPS. But win on gameplay experience.
And wenn HavokFx finaliy have something good out wich is a longer wait. Could be end 2007. Well next to the PPU there is a second CF slot I put a Dedicated G-card in it.
Because this G-card wont'help with FPS, just like PPU, but generate a heavy Physics load the G-card(s) for rendering have to deal with. You loose FPS to. But get a whole different experience. So If your used to CF you need Triple CF 2+1. In a time there are many PhysX games and only a few HavokFX games.

If you go the shared Physcs GPU route you get a larger FPS hit. Because you loos GFX power wich is use for Physics wich give a large load on partly G power you have.

Also PhsyX support is high. because it's free wich attracs a lot of Dev's. HavokFX cost extra on top off Havok. wich cost even more.
While GPU are favorite by it's flexibility and aviability by the gamers. PhysX could be the Dev's favorite. That money can be spend elsware.

The only Plus for you is That HAvok cost money HavokFX even more. So only the Dev houses with a good Puiblisher backup and budged will go for that. So the bigger boy a Devs. So they would have more time and money to make something good ou of it.
So the Crap to avarage to good to killerapp ratio could be differrent for the HavokFX solution.

 

[Ockie]

I find it funny how the only people defending the PPU at this point are the Zealots who has taken the plunge already for their decorative pieces of hardware

 

[SuperGee]

I find it funny how the only people defending the PPU at this point are the Zealots who has taken the plunge already for their decorative pieces of hardware

A zealot is more someone playing on the man with no argumentation.

So a more personal reply on the decorative point.
Well I bougth two to preview them myself with only GRAW and Celfactor demo out.
Yes Value and usefullnees was low but is climbing slowly.
Because I want to experience the potentional first hand. Some special kind off early adopter. That my choice.

Now in time I have already played BOS out, it apaers I like it more then Graw. While Graw is accualy more my thing, but the difficulty kills the fun. And looking for the Addon and the second game of BOS wich are out alraedy.
And in the near future I got to have Celfactor revolution to testdrive and UT2007.

With DX10 titles coming up mid 2007 I build a Vista DX10 gamingrig wenn there is a good Dx10 game out. And put for PhysX games aslo PPU in it. And that would be my third.

To Play Haevy Gameplay Physics over Home LAN. With three Players.

In that time there are more PhysX game out. possibly CF:R will be there.

Lokin forward to that.

 

[RamonGTP]

SuperGee, I respect your opinion but I respectfully disagree.... Yes, there are games that are going to come out that support the Physx card, but I can guarantee you that those same games are going to have a patch that will also support whatever other types of physics technologies hit the market and actually become successful, and as it stands now, it looks like GPU based is what is going to succeed.

In this case, and as far as I'm concerned, yes... Nothing is better than something if "something" is going to cost me not only money, but significant performance as well, and the only thing I get in return are some extra shards of shrapnel. I'd rather pocket my $200, get my 15+fps back and later spend that $200 on something that actually makes a difference worth spending for. Also, the Physx card isn't THAT new either.

The way I see it, if Physx was even half as successful as you want it to be, or if it provided half of what it promised, this thread would not even exist.

 

[SuperGee]

SuperGee, I respect your opinion but I respectfully disagree....

Thats cool with me.

Yes, there are games that are going to come out that support the Physx card, but I can guarantee you that those same games are going to have a patch that will also support whatever other types of physics technologies hit the market and actually become successful, and as it stands now, it looks like GPU based is what is going to succeed.

I don't think many Dev teams and there publisher will spend extra cash on Havok and another on HavokFX just for a Patch. They would do the move on there next Project. If Ageia fail with PhysX hard. only a few woul take that route.

In this case, and as far as I'm concerned, yes... Nothing is better than something

Well I don't agree. What I will agree with is that The Now Nothing camp HFX looks good and strong and stand strong. 2 Big players+1.
But sure that nothing will grow in something small and will grow into a very heavy Competive solution By then it's 2008. Thats the time scale.
While the Physics Something, still not to brag about. Is Growing to Leading it for some time. Your mistake is. That Alanwake have not to compete agians Graw. More against UT2007 and Cellfactor revolution and then some. You don't need DX10 or PhsyX hardware now it value and price become more attractive in time wenn good games are out using it well. Within a year or so. Wenn it is Okt 2007 the situation is different. HavokFX games are around and a lot more physX games.

if "something" is going to cost me not only money,

The Nothing HavokFX/GPU camp its hardware cost money to depending how much Physics power you want or lack of it the sharing solution.

but significant performance as well,

Performance?
The PPU goal is not reduce FPS. It's to enhance games not FPS but leave FPS out. It has minor influence on FPS. Dev's Have a minimum FPS target.
Or have you still those Alpha beta drivers very dated early preview FPS scores in mind.
FPS changes a bit but not drasticly. IT could be even better. Depending on the PhysX enhancment a devs has in mind. By now Ageia is a couple of WHQL drivers further.
The early largere FPS hits where a driver isue.

and the only thing I get in return are some extra shards of shrapnel. I'd rather pocket my $200,

This sounds like Graw. Theres also CellFactor and coming up Gameplay PhysX games.
First there was only graw and a gameplay PhysX demo.
And some People will do like effectPhsyics enhancments. Not every body. Its personal taiste and priority. I would go more for the gameplay PhysX
What will havokFX deliver? Mostly effect Physics like scrapnel?

get my 15+fps back and later spend that $200 on something that actually makes a difference worth spending for.

15? where does that come from. Cellfactor plays well. It's more +/- 5FPs depending wich game.

Also, the Physx card isn't THAT new either.

In Game design development cyclus it still is new. Not brand new. Weeks arent at scale even months. It's Years.

The way I see it, if Physx was even half as successful as you want it to be, or if it provided half of what it promised, this thread would not even exist.

I don''t agree.
Ageia promised a PhysX accelerate wich accelerates Physics with dedicated hardware. So it does. To show of it needs Games. Games that make good use of it. Not just use it.
Graw is bad example. it just the first one. So yes it introduction to the market could be better. But there are more Games coming out. And some could be way better. It's More Game developers using PhsX for PPU are bussy with produktion of those games with could deliver on promise. Like Tim sweeny EPIC UT2007. Artificial/Emersion Cellfactor. But they working still on that promise. Actualy the game dev's. It's largley out off ageia hand. But more in Epic and Reality engine game and others.

While at short notice there is some heavy competition anounced and roling out wich are favorite. Running behind, but HavokFX and GPU solution will cost Gamers also money. Only you can choose how much you gonna Blow cash on that. Also HavokFX must fullfill this promise.
Both promisies are still open.

There are 3 options.
1 ) The Favorite is nV Havok ATI side would win, Well they have strong backup and fan support.
2 ) The Firestarter Ageia the underdog. still bussy having a hard time but could win.wenn the Killer app arives in time.
3 ) Both solution compete for some years for now.

In 2008 we might know if PPU or it nextgen in still around.

It depends on the games.

 

[RamonGTP]

Then I suppose we'll simply wait until 2008 and see what happens... Until then, we can simply agree to disagree.

 

[SuperGee]

Good Idea, well waiting sucks but nothing we can do about that.
Better they spend some good time to make a game instead rush them.
I wonder what PhysX and HavokFX brings to us gamers.

 

[maleficarus]

"I find it funny how the only people defending the PPU at this point are the Zealots who has taken the plunge already for their decorative pieces of hardware"

I don't have one and I think the PPU is a huge step for more realisim in games. What I find even more funny is the fact that before this technology can really take off guys like you trash it. I dunno about you but nothing has been exciting the last 7 or so years when it comes to PC hardware. Only thing ATi and Nvidia have to offer is faster core and memory speeds and some fancy FSAA names.

 

[Raidium]

After reading all the post I think that the real answer here is that it all depends on who provides developers with the tools that make taking advantage of extra processing power easier. Multi CPU and Multi Core has been aroudn for a while but games still dont take advantage of it. Now that things are heading the way of cell artictcture (multi core, duo core etc) you dont really NEED a physics processor. What you need is the ability to assign some of those resources to doing the physics tasks. That takes a development kit.

SO if Ag gets the act together and provides great tools to game devs quickly and can help them justify the added resources that it takes to produce a game with Ag technology then it will become sucessful.

If Nvidia or ATI comes out with a development kit that alows the developer to take advantage of their core structure to do the same then you will be looking at either an nvidia physics unit, or you will be doing it on your video card.

Then again if the tools for taking advantage of multi core procs become aval to developers and it doesnt take much time then we will do it without adding a card in the system.

What you have to understand is that development time is expensive. It cost extra time to develop for any of the above solutions. I am going to develop my game for the widest audience possible. Unless there is some incent to go ageia why? Is the draw of a game that has physics offloading going to sell me 5million more units? There may be a handfull of games that are going to pit their sucess on how amazing a technology demonstration they are (ala Quake\ DOOM). For the rest of the titles that keep the big game companies alive forget it.

I wouldnt buy their stock though.

 

[Ockie]

"I find it funny how the only people defending the PPU at this point are the Zealots who has taken the plunge already for their decorative pieces of hardware"

I don't have one and I think the PPU is a huge step for more realisim in games. What I find even more funny is the fact that before this technology can really take off guys like you trash it. I dunno about you but nothing has been exciting the last 7 or so years when it comes to PC hardware. Only thing ATi and Nvidia have to offer is faster core and memory speeds and some fancy FSAA names.

You do know nvidia and ati is working on their own PPU? ATI actually showcased their triple play a while ago.

We are saying that the agiea solution to PPU is pretty much done, there are other methods. No one is saying that PPU is worthless, however ageias approach is.

 

[Ockie]

After reading all the post I think that the real answer here is that it all depends on who provides developers with the tools that make taking advantage of extra processing power easier. Multi CPU and Multi Core has been aroudn for a while but games still dont take advantage of it. Now that things are heading the way of cell artictcture (multi core, duo core etc) you dont really NEED a physics processor. What you need is the ability to assign some of those resources to doing the physics tasks. That takes a development kit.

SO if Ag gets the act together and provides great tools to game devs quickly and can help them justify the added resources that it takes to produce a game with Ag technology then it will become sucessful.

If Nvidia or ATI comes out with a development kit that alows the developer to take advantage of their core structure to do the same then you will be looking at either an nvidia physics unit, or you will be doing it on your video card.

Then again if the tools for taking advantage of multi core procs become aval to developers and it doesnt take much time then we will do it without adding a card in the system.

What you have to understand is that development time is expensive. It cost extra time to develop for any of the above solutions. I am going to develop my game for the widest audience possible. Unless there is some incent to go ageia why? Is the draw of a game that has physics offloading going to sell me 5million more units? There may be a handfull of games that are going to pit their sucess on how amazing a technology demonstration they are (ala Quake\ DOOM). For the rest of the titles that keep the big game companies alive forget it.

I wouldnt buy their stock though.

Agreed 100%.

 

[Rich Tate]

<snip>

Great post. I also agree with you.

 

[maleficarus]

"We are saying that the agiea solution to PPU is pretty much done, there are other methods. No one is saying that PPU is worthless, however ageias approach is."

I disagree. Waiting for Nvidia or ATi to release something without jacking another $100+ on there cards is not going to happen anytime soon. Having a dedicated PPU add-on card is the way to go imo as it gives users the choice to spend the cash vs. being forced to if it came with your VGA card in some form. I can honestly see it now a MID range mainstream card retailing for $399.99 and there high end models for $699.00 sigh...

Or go for Ageia solution for $150.00 or whatever it is now that comes with its own dedicated memory.

 

[SuperGee]

After reading all the post I think that the real answer here is that it all depends on who provides developers with the tools that make taking advantage of extra processing power easier. Multi CPU and Multi Core has been aroudn for a while but games still dont take advantage of it.

Most Dev team going for a licence game engine. These Game engine makers will take the time To go to the edge of technology. Offering Game Developers a strong basis to use new tech and the latest hardware. Like unreal engine 3. The engine takes some time to make. But after that there comes a wave of games.
SMP starts with the Gameengine. Valve Epic and more support it.
Knowing that 5 Game engines support PhysX and with it PPU.
Well know game would be UT2007 Unreal3 using UE3.0
Well know PPU demo Cellfactor cT and Revolution fullgame using
Reality Engine.

So its PPU suport with high level tools is no Problem. Game engines and Middle ware is on it place. And HavokFX and PhysX already supporting Multicores on Phsyics level. Now the rest of the game.

Now that things are heading the way of cell artictcture (multi core, duo core etc) you dont really NEED a physics processor. What you need is the ability to assign some of those resources to doing the physics tasks. That takes a development kit.

You mixing different isues. Well Lest take this Cell thing. The AMD version would be Fusion. But that is the far future. While IBM Cell could do most general graphics task a GPU does to. But PS3 will use a RSX non the less. Why a Dedicated GPU is better and more powerfull at that Task. AMD Say Fusion wont replace dedicated hardware. While there is still need in the market for more powerfull dedicated solutions.
So next to Fusion there will be still G-card. While Fusion have one or more GPU cores Wich are lite once, so in total still a low budged to low midrange solution.

A PPU3 would be in that far away time be the Mid to high-end hardawre accelerated solution. And for GPU a G100 or R900. Those SPE, the X86 versions in a Fusion CPU would be the budged solution.

It a choice Gamers have to make spend it or not. Most will not. If enough do, the hardcore Gamers enthausiast. Gamers with large budged. You got a market for it. while Hardware accelerated support in games could be a common thing. And DirectX PhysX supporting all Hardware solutions avaible. PPU3 and G100 and Fusion etc.

So if Ag gets the act together and provides great tools to game devs quickly and can help them justify the added resources that it takes to produce a game with Ag technology then it will become sucessful.

Tools are there for a longer time while HavokFX is still fresh. The games are coming it takes time to make a game and also game engine. But they are in produktion for some time, yearsnow, this year some will come out.

If Nvidia or ATI comes out with a development kit that alows the developer to take advantage of their core structure to do the same then you will be looking at either an nvidia physics unit, or you will be doing it on your video card.

They use shaders. Wich becoming more general purpouse. Dev's arent in the game for doing two different paths nV and ATI one. So Most dev's will go for the HavokfX solution wich support both. One day in the far future DirecX will take that over.

Then again if the tools for taking advantage of multi core procs become aval to developers and it doesnt take much time then we will do it without adding a card in the system.

What this. PhysX supports already Multicore and PPU. So Tools arent the problem.
You see Multicore as equal to PPU but it's not. And then there is GPU.
They play at a different level Of Phsyics power overrings.
And that all abstracted with a API PhysX or Havok FX choosing between the two hardware solutions. Tools arent the problem.
Multi core for the game platform is still High-end in price so a small audience. will become Mainstream in 2008.

What you have to understand is that development time is expensive. It cost extra time to develop for any of the above solutions.

The big boys have some extra cash resources and expertise and ambition to make great game for some more time and some flexibility to stand out and make a AAA title. To stand out new tech used the right way can put far away off the competiton. Devs compete is a way. Using CPU instead of PPU or GPU would mean stayin behind. Some dev's don't have that choice because the are small players. Free PhysX helps.
However Using PhysX PPU right a small gamedev can stand out next to the big boys if doing it right. a balanced game focusing on Physics to enhance gameplay.

I am going to develop my game for the widest audience possible.

Every game have a target mainstream. More is welcome but also Optional. It can make your game stand out. And give it a bigger sucses. By using new Tech.
Some game Have such high profile and requierments that they push hardware sales.
altho for some projects its a risk assesment. Do a step back while other stake the opertunity.

Unless there is some incent to go ageia why? Is the draw of a game that has physics offloading going to sell me 5million more units?

No but look at Cellfactor. Do they offload or enhance the game more then a CPU can handle. That what attracts people. Surtenly not offloading.

There may be a handfull of games that are going to pit their sucess on how amazing a technology demonstration they are (ala Quake\ DOOM). For the rest of the titles that keep the big game companies alive forget it.
I wouldnt buy their stock though.

Well it's depends if the do it right. Good Gameplay mechanics using new tech. Enrich games with well thougth enhanced Physic use, intead make it a Tech demo expanded to a full game with the focus on the tech.
Using New tech can be a risk but also a opertunity to sucses. It depends what the game designer makes out off it.

 

[Ockie]

I think we got a new Terra on these forums... perhaps it's the same person?

 

[TeamStrykerCore]

That cracks me up. WHAT GPU based physics support? Show me one game that you can play right NOW that supports GPU physics. I'm not a PhysX !!!!!! but at the very least they have actual working games to support their hardware, no matter how crappy they are.

Welcome to 2007 games do calculate physics in them currently without addon hardware.

 

[mlambert890]

The fact that DX10 already has some physics functionality and the fact that a Direct Physics API is coming, and the fact that DX10 GPUs are fully programmable and, as indicated by the 8800GTX, have a lot of power and, finally, the fact that dual core procs are becoming a commodity and quad core will be soon all seems to paint a dim picture for a dedicated PPU.

This is really very much like what happened with audio and 3d rendering. The industry just doesnt tolerate a single vendor API solution for long. Also, physics processing is not an unknown, it is just handled by the CPU today, obviously. So with a dedicated PPU (today) you theoretically get a more impressive experience (VERY subjective from what I've seen), but it seems you DO see a framerate hit in at least some of the games that implemented PhysX support. Thats not good.

Personally, I couldnt fit a PPU in my setup if I wanted to, so the guy who raised the mobo issue has a good point. 8800GTX SLI and the X-Fi max out physical space on the eVGA 680i.

I think that before the dedicated PPU concept can gain any traction, the industry giants will have already brought a more workable solution to the table that is better for the devs (MS XNA support tends to be better than the support from a small proprietary company or from the open source community which is the real reason DX has been a success despite some people's refusal to believe it), uses existing hardware by either working in software and spawing threads that multi-core CPUs can manage or leveraging the programmable DX10 GPUs and is better for the end user since they really dont have to think about anything or buy yet another dedicated, expensive, part.