Regarding the PPU delay...

[Guden Oden]

Did aegia say why?

I must have missed the original announcement - well, I assume there was one - because the first thing I heard about it was reading some anandtech article or such where it was mentioned it being pushed ahead until Q2 '06. Then I saw some other confirmation of the anandtech blurb, but I don't remember where that was. So apparantly it's been delayed, except I still don't know the cause of it.

Did they say it's because there's no games yet using their physics library (which would be a rather big reason I suppose, considering the planned retail price if nothing else), or is it some other root cause?

 

[YeuEmMaiMai]

Did aegia say why?

I haven't even heard about the delay.........

 

[_xxx_]

Wasn't it because of a failure in one of the metal interconnect layers?... Oh wait, wrong thread!

 

[DudeMiester]

Did they say it's because there's no games yet using their physics library (which would be a rather big reason I suppose, considering the planned retail price if nothing else), or is it some other root cause?

It was because there were no games.

 

[Cartoon Corpse]

dagnabbit.

 

[wireframe]

Let's hope they weren't counting on Duke Nukem Forever as the introductory title. That could take some time.

 

[karlotta]

It was because there were no games.

Well UBI had Ghostrecon3, but that was delayed till Febuary......hmmmm.

 

[Guden Oden]

At least there's been a bunch of announcements of game devs using their API, so with a bit of luck that should lead to a good selection of titles, and if the first batch has at least one hit game successfully using massive physical simulations, deforming level geometry, water simulation etc, then that will spur a torrent of new interest in the tech.

...And hopefully after the delay, they'll release both PCI and PCIe flavors on launch day. I simply don't have any free PCI slots in my box, and even if I did I wouldn't want to use a technically inferior interface anyway. PCI is good for like, USB/firewire root hubs and modems and TV cards and shit like that...

 

[wireframe]

...And hopefully after the delay, they'll release both PCI and PCIe flavors on launch day. I simply don't have any free PCI slots in my box, and even if I did I wouldn't want to use a technically inferior interface anyway. PCI is good for like, USB/firewire root hubs and modems and TV cards and shit like that...

I'm looking at a bunch of mainboards and I think I am seeing a major problem as far as slots are concerned. If you put SLI/CrossFire into tthe equation you begin to see that there are almost no slots left and those that are are a choice between PCI and PCIe, depending on the model (this is not taking any data limitations into account (shared slots and so), just physical limitations). I think we need a push forward to have everything PCIe and that means sound and these PPUs. We can already forget about a NIC or any other fun little cards we may want to pop in there. Without LOMs it would be a squeeze. Even onboard audio is looking like a necessity soon.

Of course, if you take SLI out of the equation then you free up tons of space, especially if you get rid of dual slot SLI and go to single slot configuration. One can say that SLI has almost completely monopolized the mainboard, eating up all its space.

 

[DOGMA1138]

Thread hyjacking!
any idea on how much $$$ the damn thing will cost?

 

[Guden Oden]

any idea on how much $$$ the damn thing will cost?

In the neighborhood of 250-300 bux I believe Aegia has said. I think though, that they're throwing out this rough figure to judge reactions, and the reactions so far have been pretty negative. That much cash, on top of one or even two graphics cards, well, that's a pretty thick wad of money just for playing games.

 

[kyleb]

For $300 the thing better do a hell of a lot more than just add fancy physics to games. For that kind of cash I could easly pawn off my 3000+ winchester and score a 4200+ X2.

 

[SlmDnk]

There has been no official announcement from AGEIA about the delay.

But there's a hint to a possible delay, though: http://www.gamecloud.com/article.php?article_id=2227

Manju Hegde - As we’ve said in the past, our hardware shipping date relies heavily on the development of compelling content. We are in the process of evaluating PhysX processor optimized games and are assessing our launch plans at this time.

Also, there should be a pretty big PhysX media flood soon: "We are in the process of evaluating content right now, so news will be coming soon.".

 

[jvd]

perhaps if there is a delay its for the best. Mabye a few games on the market before its release that can take advantage of it would be good stuff.

Anyway at 100-200$ i would most likely buy one if there were some good games that supported it at more than 200$ though and your pushing it

 

[kyleb]

I agree. Up to $200 would be reasonable but I can't see it being worth much more than a fancy soundcard.

 

[wireframe]

Have you guys given it much thought if you are only willing to pay $200 and compare it to a sound card? Think about it. Until the X-Fi, sound cards were a handful of transistors. Even the X-Fi's transistor budget is a bit questionable: do we really need all that jazz to have sound that is 10% better in theory?

It just sounds very unlikely that such a card would carry a low price tag like $200, even if AGEIA subsidized the costs with sales of games using the technology. Where are these guys gonna make money?

Although I think video cards are getting a bit on the expensive side, I am also amazed that you can get something like the 7800 GTX 512MB for less than $800. If you think about it a little there are CPUs people are buying that cost more than that. CPU's with transistor counts not even coming close to half of the G70. Add 512 MB of super fast memory, a PCB, and all the related circuitry and it is a bloody marvel than they cost as little as they do. That or Intel and AMD seriously need to reconsider their CPU pricing.

The only thing I am hoping for in PPU development is that it is done in slow and clealy defined steps. I don't want to dish out for a new PPU every time a new game is released. I don't even want to swap it as often as every year like with video cards. I'd rather pay $500 for it and know it will be good for two years +. I actually hate the incremental BS in the GPU business. It's not the money, it's not knowing what you will get. At certain times you have this "high powered card" and you are wondering when games will use all that power, then, in one fell swoop, it goes from being idle to being crushed. If it is this type of development we will see with PPUs I will seriously have to reconsider my gaming habits and start thinking Playstation and Xbox instead of all these gadgets that never come together to form the beast of a processing station that they propose.

 

[jvd]

It just sounds very unlikely that such a card would carry a low price tag like $200, even if AGEIA subsidized the costs with sales of games using the technology. Where are these guys gonna make money?

Once ppus find thier place they will make thier money.


They need a large installed base to get more games. Then once there are lots of games they can make boards with more memory or even multi ppus on a board and charge more .

 

[wireframe]

Once ppus find thier place they will make thier money.

They need a large installed base to get more games. Then once there are lots of games they can make boards with more memory or even multi ppus on a board and charge more .

I don't think that is the crux. Havok has been doing business for years without dedicated hardware support. Novodex is just another Havok, right? So, if Havok is focusing on GPU support that means they will have to make money without manufacturing GPUs. I figure the point of Novodex is to take something like Havok to a higher level, hence the PPU hardware.

Are you thinking that AGEIA is focusing on hardware because that is the only way they thought they could break Havok's back?

PS. I am not aware of any games that use Novodex, as a Havok substitute, but if they already have a pool of previous customers, please clue me in.

 

[jvd]

I don't think that is the crux. Havok has been doing business for years without dedicated hardware support. Novodex is just another Havok, right? So, if Havok is focusing on GPU support that means they will have to make money without manufacturing GPUs. I figure the point of Novodex is to take something like Havok to a higher level, hence the PPU hardware.

the software isn't free , that is a cost the devs have to pay . That is how havok is going to make money back.

I bleieve that novodex is going to use the money from software to offest some losses from the hardware sale.

With the idea being when there is enough support they can sell more expensive add in cards later as the middleware and hardware becomes popular .

Are you thinking that AGEIA is focusing on hardware because that is the only way they thought they could break Havok's back?

I think making the hardware is a way to make sure there is proper hardware that can run a certian lvl of thier middleware with no problems .

The gpu solution is going to lead to games having to scale over many diffrent types of hardware .


Instead of a ppu card your going to have gpus from ati and nvidia and aside from those diffrences your going to have to worry about the diffrences in the lines as well .

Where as novodex can dictate how many ppus come out at diffrent power lvls and ram lvls they can't dictate how many diffrent grpahics cards come out .

 

[wireframe]

the software isn't free , that is a cost the devs have to pay . That is how havok is going to make money back.

That was my point. That Havok is not going to be making money through hardware sales (unless they get some payments for "Havok Certified GPUs").

I bleieve that novodex is going to use the money from software to offest some losses from the hardware sale.

This is what I said in the original post. That was my meaning with "subsidizing" the hardware.

With the idea being when there is enough support they can sell more expensive add in cards later as the middleware and hardware becomes popular.

But why wouldn't they be expensive from the start? Don't you see the circular problem here? They are not Havok. They don't have a long list of customers (that I know of) that are paying for the software. They are entering the market from both sides: software and hardware. This hardware has to come from somewhere and it is going to cost money. If Intel, a company with a long history of producing chips and owning their own fabs, charges $25 for chipsets (more now) that are mass produced and this is a business they are regulating with their CPU sales, don't you think an up-and-comer will have reasearch costs, development cost, manufacturing costs, and distribution costs to cover? And with what? Income from their historically loyal developer fan base?

This is why I am saying it must be a dramatic increase in performance and the hardware still cannot be cheap. They must have made some deal before going forward and I doubt that deal was made on a "well, let's just try it" and a handshake.

The gpu solution is going to lead to games having to scale over many diffrent types of hardware .

Instead of a ppu card your going to have gpus from ati and nvidia and aside from those diffrences your going to have to worry about the diffrences in the lines as well .

Where as novodex can dictate how many ppus come out at diffrent power lvls and ram lvls they can't dictate how many diffrent grpahics cards come out .

This has very little to do with pricing of AEGIA hardware, I think. The Havok/GPGPU threat will only determine how far down in the pricing they will have to go or if they utterly fail. This doesn't change the fact that they will have a lot of costs to cover.

Think about Voodoo 1 when it was launched. It is a relative piece of junk compared to what AEGIA is proposing. Remember the price? Why not add some to that to compensate for inflation and we should begin to see that $200 sounds a bit far fetched.

I am willing to be pleasantly surprised by being wrong on this.

BTW, I think both AGEIA/Novodex/Physx and Havok will also support SMP processors transparently. The question is how this will be regulated in the games because it seems that only with AEGIA is there a possibility to know with absolute certainty what the hardware is capable of.

If you put together how I see this, my viewpoint becomes clear:
Against the odds, I am optimistic that this could work. It's going to be expensive and not every game will support AEGIA (actually, very few I would think). So, it better be damned fast. And here we see the circular behavior again...it keeps on having to be faster and faster or it has nothing to offer...so it is going to be expensive.