Asus makes PhysX boards

[tEd]

http://techreport.com/onearticle.x/8335

Proabably most physics engines will be optimized for multithread so dual core cpu or PhysX or both?

[_xxx_]

I have to PPUke... :wink:

[Richard]

Villagers demand pricing and SKU tiers breakdown performance specs.

[Dr. Ffreeze]

I can't wait. I hope it hits critical mass. I love to see things blow up and everything go flying. =)

[digitalwanderer]

Villagers demand pricing and SKU tiers breakdown performance specs.
Yup, badly! There's even some rumbles of mutiny...

-rumble, rumble, rumble....mutiny, mutiny, mutiny....rumble, rumble, rumble....mutiny, mutiny, mutiny....rumble, rumble, rumble....mutiny, mutiny, mutiny-

[Dr. Ffreeze]

Dear AGEIA Technologies,

Please note that I will buy (2) Two PhysX cards to counteract digi's mutiny of (1) One in whole. =p

Thank You,

[ndoogoo]

:D

PCI or PCI-E?. The x86 needs a little help in hand, it has not go much faster over the last year or two.

It would also be nice if you had access to the chip and not just through the physics API.

[Kaotik]

:D

PCI or PCI-E?. The x86 needs a little help in hand, it has not go much faster over the last year or two.

It would also be nice if you had access to the chip and not just through the physics API.
I believe the cards shown while ago had both interfaces?
At least PCI, and IIRC PCI-e too

[nutball]

It would also be nice if you had access to the chip and not just through the physics API.

Word. I have a few things to throw at it that the API doesn't cope for (despite the PR hype).

[mashie]

:D

PCI or PCI-E?. The x86 needs a little help in hand, it has not go much faster over the last year or two.

It would also be nice if you had access to the chip and not just through the physics API.
The chip support PCI, PCIe 1x and PCIe 4x interfaces. It is unlikely anything but development boards will have more than one interface though.

[digitalwanderer]

Dear AGEIA Technologies,

Please note that I will buy (2) Two PhysX cards to counteract digi's mutiny of (1) One in whole. =p

Thank You,
Hey, I got four hungry rigs here I have to feed....not just Bubbles! :?

[suryad]

I thought this physics chip wont fly...though I personally think it is cool and it is a good thing I guess for the physics chip advocates to see a well known computer hardware manufacturer like ASUS to jump on the bandwagon. Coolness!!

[Geeforcer]

I still don't think this chip will do much good in the mid to near term, considering that GPU is far more of a limiting factor at this point.

[rwolf]

I thought CPU was the limiting factor.

[Jawed]

Well, after seeing the videos of what the next gen consoles "can do" (hey, preliminary impression, alright?), I reckon it'll be 3 years before PCs have the CPU power to match. So PCs are gonna need all the help they can, e.g. PPU :lol:

Jawed

[Ostsol]

I wonder if we'll see integrated solutions. . .

[Geeforcer]

I thought CPU was the limiting factor.

Not when decent resolution, AA, AF, HDR, etc come into play.

[Mulciber]

I thought CPU was the limiting factor.

Not when decent resolution, AA, AF, HDR, etc come into play.

but those have nothing to do with physics

[Geeforcer]

I thought CPU was the limiting factor.

Not when decent resolution, AA, AF, HDR, etc come into play.

but those have nothing to do with physics

No that would be CPU. But since the graphically intensive games are not CPU-limed in the first place, how much performance increase would PPU deliver in most cases in the current games? It seems to me that it would take games designed with PPU in mind to show any real benefits..

[digitalwanderer]

But since the graphically intensive games are not CPU-limed in the first place
Woah it down there, I thought quite a few games are CPU limited rather than graphically limited at the moment. :?

[Kaizer]

No that would be CPU. But since the graphically intensive games are not CPU-limed in the first place, how much performance increase would PPU deliver in most cases in the current games? It seems to me that it would take games designed with PPU in mind to show any real benefits..

I'd guess that separating out physics calculations onto a separate AIB would take atleast a couple of killer-apps anyway..? People seem to be good at creating markets these days, so if they have the funds and energy to involve/inspire the right developers, they might get this snowball running... :)

--
K

[ANova]

Meh, I would rather see some good innovations for the CPU rather then having to buy yet another board.

[Geeforcer]

But since the graphically intensive games are not CPU-limed in the first place
Woah it down there, I thought quite a few games are CPU limited rather than graphically limited at the moment. :?

I don't know, haven't been keeping up with the scene last few months. Can you give some examples?

[Saribro]

It seems to me that it would take games designed with PPU in mind to show any real benefits..

Nobody is going to write a fabulous physics engine if you know you're going to make the CPU crawl and leave your game unplayable.

[Wallslide]

I know that Counter-Strike:Source is very CPU limited. I have a 6800 Ultra, and was unable to get much of any FPS change even by lowering resolutions, so I just left the resolution at 1600x1200 (No AA or AF). When I upgraded my CPU from a p4 3.0ghz machine to an Athlon 3600+ though, I got a huge increase in performance (20fps+ ).

[PatrickL]

I have some doubts about that new add-in board. When i ask around me, less and less people bother to buy an add-in sound card already, i really doubt theses guys will buy that ppu card. Of course if it is proven to be a must have things may change.

[soylent]

I have some doubts about that new add-in board. When i ask around me, less and less people bother to buy an add-in sound card already, i really doubt theses guys will buy that ppu card. Of course if it is proven to be a must have things may change.

That's because there are integrated sound cards available. Even an integrated PPU should still be much better than just a processor.

[PatrickL]

I agree but i did not see yet any MB maker put one on their product.

[Frank]

But since the graphically intensive games are not CPU-limed in the first place
Woah it down there, I thought quite a few games are CPU limited rather than graphically limited at the moment. :?

It depends what you want to do. There is a large amount of all kinds of FP caclculations (vector, mostly) that the CPU has to run as long as you have no other unit that can do them. The GPU is not flexible enough. If you add a high-speed math (vector) unit, like CELL or a PPU, you can use that to off-load most of it, leaving the CPU for the managing and things like AI. There is much more batched brute force FP calculations to 3D games on the CPU than anything else.

You only have to watch out for copying data too much to and from, as that is a relatively slow thing and would leave the CPU with nothing else to do in the mean time. Blocks of code that fit into the cache are good and fast, handling large amounts of data that don't fit starves the bandwidth pretty fast.

[trinibwoy]

I agree but i did not see yet any MB maker put one on their product.

Which is exactly why an add-in PPU would be more attractive that an add-in sound card. Also, it's probably much easier to evaluate the impact of the PPU than the sound card especially since integrated solutions are of sufficiently high quality already.

[IgnorancePersonified]

I wonder if we'll see integrated solutions. . .

I'm thinking more and more this type of unit will replace the DSP in the south bridge of nforce products as an example. Perfect candidate.
I don't think the sound card market is a good analogy cause it seems to be different to other sectors of the pc industry. Stunted and deformed.

[Geo]

If this thing is a success, one of the big boys will buy it up and integrate it.

[Sage]

If this thing is a success, one of the big boys will buy it up and integrate it.

Creative will buy it up and then we will soon be saying...

it seems to be different to other sectors of the pc industry. Stunted and deformed.


...just like sound cards

[tEd]

There's a picture of the physX card at GDHardware's E3 coverage

http://www.gdhardware.com/events/e3/2005/day1/001.htm

[nutball]

Great, yet another noisy fan to try to eliminate from my case! :evil:

[CI]

There's a picture of the physX card at GDHardware's E3 coverage

http://www.gdhardware.com/events/e3/2005/day1/001.htm

What's with the connectors on both edges? PCI and PCIe 1x?

[tEd]

There's a picture of the physX card at GDHardware's E3 coverage

http://www.gdhardware.com/events/e3/2005/day1/001.htm

What's with the connectors on both edges? PCI and PCIe 1x?

One is AGP the other is PCIe 1x

[Thowllly]

What's with the connectors on both edges? PCI and PCIe 1x?

One is AGP the other is PCIe 1x
Nope, CI was right, one is PCI (keyed for both 5v and 3.3v), the other PCIe (and AFAIK it could be 2x or 4x too). Search for some pictures of pci and agp if you don't believe me :P

[tEd]

What's with the connectors on both edges? PCI and PCIe 1x?

One is AGP the other is PCIe 1x
Nope, CI was right, one is PCI (keyed for both 5v and 3.3v), the other PCIe (and AFAIK it could be 2x or 4x too). Search for some pictures of pci and agp if you don't believe me :P

pci it is then , i didn't really checked. Looked AGPish to me at first sight but if i think about it why would you want a physx card to have a agp interface anyway.

[CI]

Interesting if that's the reference design. I was wondering how much bandwidth is needed for the kind of processing needed. Wouldn't there be any inefficiencies by having the NB, SB chips and respective buses in between the CPU/RAM and the PPU card on PCI(e)?

[PatrickL]

Xbit lab is reporting that theses cards will be close to 300 $. For that price i bet most people should upgrade their cpu than buy that. I very much doubt we see anytime soon a lot of games that need a top cpu a top gpu AND a PPU to run perfectly well :?:

[Mulciber]

i'll get one just for bragging rights

[ndoogoo]

Xbit lab is reporting that theses cards will be close to 300 $. For that price i bet most people should upgrade their cpu than buy that. I very much doubt we see anytime soon a lot of games that need a top cpu a top gpu AND a PPU to run perfectly well :?:

$300 :shock: I was expecting around the £50 mark. Must come with tons of memory for that price. Shame it will never reach the mass market at that price.

Cheaper just buying a X-Box 360. More bang for your buck.

[scificube]

Xbit lab is reporting that theses cards will be close to 300 $. For that price i bet most people should upgrade their cpu than buy that. I very much doubt we see anytime soon a lot of games that need a top cpu a top gpu AND a PPU to run perfectly well :?:

Where are they getting those numbers from?

Are those the intro or high end parts?

Do they account for this being the first entry products into the market that need to recover the costs of R&D etc.?

Anyway...

As I see it the PPU is about the only way PC's are going to be able to handle next gen console ports.

It will be years before AMD and Intel introduce chips that are on the level of the Cell or the X360's CPU. Barring these CPU's should hit the PC market or some similar offering from IBM (cost!?!...a lot!) there is no way next gen physics will retranslated back to the PC realm. Given how the PS3's Cell has got the X360's CPU licked the console could use a PPU to even the score or actually best it...unfortunately there simply aren't enough memory resources, bandwidth etc for all that without some significant reworking of the whole system. I wonder if they can and will beef up the CPU more before launch...but then Sony can do that too...

Sorry got OT for a bit. IMO if you want the same level of physics and AI to come over with console ports then a PPU is necessary to handle the physics load and free up our dual-cores to handle the AI. Our GPU's should be able to handle the raw graphics but even still the tessellation etc abilities the Cell has will mean that stuff won't make the crossover in any reality I feel.

OT does anyone feel that the PS3 with it's Collada and Linux love would give OpenGl and Linux a significant push in the PC world? I mean the PS3 is going to be successful and if ports are possible and a common thing of course. I don't mean to say that XNA won't keep directx and Windows in a dominant position but perhaps now the competition just got a more stronger than people realize...competition is a good thing.

[PatrickL]

Well just read (http://www.xbitlabs.com/news/multimedia/display/20050520034045.html) here you will know as much I know :)

[nutball]

It will be years before AMD and Intel introduce chips that are on the level of the Cell or the X360's CPU.

Why so? What gives you that idea? Cell maybe, but the XBox 360 CPU(s) don't sound all that far from what's already on the market for x86 (three-core rather than dual-core, and SMT which is of application-dependent benefit, and 3.2GHz which isn't exactly special as P4 has already shown).

Turn it around, it could be years before console programmers work out how to actually use the power of the CPU(s) in the consoles.

I'm interested to know how a pile of hype about PS3/XBox360 implies that the PC is for-sure dead in the water.

[scificube]

Well just read (http://www.xbitlabs.com/news/multimedia/display/20050520034045.html) here you will know as much I know :)

Thanks!

edit:

...hmm...that's kinda steep...$250 to $300...that's all and well fine for me, but for the average joe that price seems rather prohibitive...

...I also wonder if things are in this price range, how in the world could you begin to talk about integration into the mobos....

...I hope more is going on behind the scenes and that these parts are the price of admission where prices should fall quickly...

end edit:

[scificube]

It will be years before AMD and Intel introduce chips that are on the level of the Cell or the X360's CPU.

Why so? What gives you that idea? Cell maybe, but the XBox 360 CPU(s) don't sound all that far from what's already on the market for x86 (three-core rather than dual-core, and SMT which is of application-dependent benefit, and 3.2GHz which isn't exactly special as P4 has already shown).

Turn it around, it could be years before console programmers work out how to actually use the power of the CPU(s) in the consoles.

I'm interested to know how a pile of hype about PS3/XBox360 implies that the PC is for-sure dead in the water.

What you say about know how to take advantage of multi-core processors is true...but it's true of all mulit-core processors. The problem being though one multi-core != another and Cell just has raw power and though less so the X360's CPU does as well.

You asked why I said what I said so I'll put out how I see things.

PowerPC processors are typically clocked much lower than what the Cell and X360 CPU are and still outperform or perform on par with x86 cores such as Intels. Point being a PowerPC core running at the same clock speed as an Intel x86 core would mean the PowerPC core is significantly more powerful in this matchup. Proof to point - we have a more direct example in the AMD x86 core. If AMD placed a 3GHz+ core in it's next FX and Intel simply stayed in the 3GHz range then Intel's CPUs would get creamed...badly I feel.

For now let's consider AMD's 2GHz offerings and Intels 3GHz offerings to be synonymous as they perform more or less on the same level with these respective clock speeds. The Rub here is that a core from one of these next gen consoles will more powerful due to raw speed alone. (note what I said about PowerPC clockspeed...it is much like how you would compare AMD's speeds to Intel's)

For the a good while into the future x86 isn't going to be sped up but instead going to leverage multi-core in the PC market for power. That is to say their multi-cores will be in the ranges I have already mentioned until probably moving to 65nm or some other next logical step.

This means that core for core there is a significant performance disadvantage for the x86 parts to the abnormally high clocked PowerPC cores in the Cell and X360 CPU.

The next layer is that these cores are more robust for gaming applications core for core than their x86 or even the PowerPC brethren. The VMX units in the X360's CPU are powerful and hella fast...there won't be an answer for them in the current form of SIMD instructions in the x86. I am being superficial here but the point I want to make is that these processors can be made to be "gaming" parts that perform general and non-specific tasks where as their PC counter-parts must be more generalistic and drag along legacy capabilities that slow them down or make them bigger and hotter or waste die space they could be using to make them more robust in similar areas.

Then there's the issue of simple numbers. Two cores vs. 3+ cores in console CPU's for a very good while I feel.

So this is how I see it.

PC CPU's are at a speed disadvantage (one that is not merely by numbers but by a measurable metric), optimization for gaming, and physical core count per chip.

PC CPU's will be at a significant disadvantage relative to being gaming beasts for a good while IMO. Remember...I am thinking in terms of gaming performance and I do note that it will take time to learn how the utilize these chips fully...the problem is that this learning disadvantage goes both ways leaving raw power as the point of inflection.

edit:

I never said PCs are dead in the water or PC gaming was dead in the water and I never will. I said PC's got some hard times coming and a PPU IMO would help to calm the storm.

If I thought PC gaming was dead (as is getting owned and nothing could be done about it) I wouldn't be here talking about it. :)

end edit:

[pjbliverpool]

The X360 cores arn't directly comparable to PC cores and they arn't the same as G5's. According to anadtech they are much simplified and would be significantly slower than a PC core.

This makes pefect sense given how much they cost to manufacture. Market forces are the same for all companies and Sony and IBM can't just sidestep them for consoles.

Think about it. Intel charges a huge premium for its EE edition almost exclusively for gamers. If these IMB cores were so great then for what they must cost to produce Intel could probably easily afford to whack 3 of them onto the P4 for the same cost as its extra cache.

[Dr. Ffreeze]

Wow,

$250 to $300 was more than I was hoping for. I thought that they would have some different speeds. Before I will buy one, a game that I own has to support it. The problem is I don't buy $50 games very often. I just got HL2 and Far Cry but that is only because it came free with my motherboard cpu combo. City of Villains supports it and I am now playing City of Heroes, but I am not sure the villain appeals to me as much as the hero.

What would be nice is if AGEIA has enough capital to pay Valve, ID, and maybe Crytek to go back and put in a patch that supports the PhysX card. That would get me closer to buying it, the more games in my game library both new and old that support it. I don't think that would happen, but it would be nice.

[scificube]

The X360 cores arn't directly comparable to PC cores and they arn't the same as G5's. According to anadtech they are much simplified and would be significantly slower than a PC core.

This makes pefect sense given how much they cost to manufacture. Market forces are the same for all companies and Sony and IBM can't just sidestep them for consoles.

Think about it. Intel charges a huge premium for its EE edition almost exclusively for gamers. If these IMB cores were so great then for what they must cost to produce Intel could probably easily afford to whack 3 of them onto the P4 for the same cost as its extra cache.

I did say these CPUs were optimized for gaming and thus I had already taken into account they would be simpler parts. They need not carry around functionalities be they legacy or not that will not be put to use on the gaming platform they are meant to be used on.

I did not suggest these cores were comparable to G5 cores. I was speaking to the point made about raw clock speed but I did mention these cores were not like cores on the PC platform...or for that matter on the Mac. It actually part of why when it comes to gaming why I think PCs are at a disadvantage.

Intel charges out the bootie for EE chips because they can. It's not because they're the best out there. That would be the Athlon FX series. To reiterate if FX's were clocked to the speeds EE's were then Intel chips would not be the lesser option they are now but really a non-option for the die hard gamer that would buy an FX or EE in the first place. (ignoring price of course)

I really don't understand what you mean with the idea of Intel slapping IBM cores on the CPU or what I percieve the implication is...if IBM is so hot why hasn't Intel snatched up their goodies and started using it. For one the x86 and PowerPC ISAs are different, they IP is that of competitors in the same market and really I don't think they could ever get an efficient working part of any sort. I also see no good and widespread need for such a part.

If you mean why hasn't Intel just changed it's architecture to something like the PowerPC's then this won't happen because they would be breaking their promise of retaining software compatability indefenitely with the CPU line OR they would have to ensure their CPU still had x86 functionalities OR could emulate...

...were you just yanking my chain?

Grrrr!

[ZoinKs!]

I'm interested to know how a pile of hype about PS3/XBox360 implies that the PC is for-sure dead in the water.
This happens every time new consoles come out.

It's true that the pc game market isn't the overwhelming force it used to be, but it's also not quite as dire as most numbers show. For one thing, the reported sales figures don't include big items such as the massive piles of cash that go into mmo subscriptions or downloaded content such as Steam.

On topic, $300 for a physics card is out of my budget for the time being. It'd be better to have a less expensive but still decent option rather then a completely awesome but expensive board.

As it stands now, the board is facing a "chicken or the eggs" issue.

[Dr. Ffreeze]

New Info:

Not so good too... =(




We sat down the company and witnessed two demos running live on first-run, A0 silicon. The final silicon that will ship in boards is A1—AGEIA made a simple metal spin, but made no major changes to the silicon design. It'll be produced on a standard .13 micron CMOS process at TSMC.
Good

Initially, it will only come only in a PCI card, with PCIe cards expected further in the future.
Bad, if I am going to spend $250 I don't want t PCI card. I want a PCIe. I do not see myself upgrading a PhysX card as often as I do a video card or sound card, unless they give me new must have features.

A large rocky hillside had about 4,200 boulders dropped at the top, which all bounced, tumbled, and interacted in a realistic (and speedy) fashion. AGEIA claimed that a dual-core CPU can handle maybe 800-1,000 in a demo like this, but was quick to note that 4,200 boulders was nowhere near the capability of their chip. There's a driver issue right now where a lot of the timings need to be worked out between the massively parallel math units in the chip. Within a couple of months, the company will have a new driver which will enable them to raise the boulder count to 32,000. They're confident they can reach that number, but even if they can only get halfway there, 16,000 to 20,000 boulders is a lot better than a CPU can do.
ACK! So right now it doesn't work 100%. It is beta but it just scares me.

At launch, the PPU cards will only handle rigid-body and particle-based fluid dynamics physics calculations. A driver update early in 2006 will enable both soft-body and hair/clothing acceleration as well.
ACK ACK! Pay now, we will enable the PureVideo... er Other Cool Stuff(tm) later. This is very bad in my book!
www.extremetech.com (http://www.extremetech.com/article2/0,1558,1817922,00.asp)

I want them to be successful, I want cooler physics in my games, but this is not good news. I know they still have time to fix things, but when I want something badly I get a bit skeptical even though I try to remain positive. =)


Good opinion at Arstechnica
After reading about these demos, I stand by my conclusions from my original post: "it's difficult to envision a game that is essentially the same but comes in two different versions: one that has a standard software physics engine, and one that has a hardware-accelerated physics engine that's such a quantum leap in gaming experience that you absolutely have to run out and buy an add-in card." I guess we'll see soon enough if developers can make this happen.

[scificube]

Definitely not good news!

I say don't release if you're not ready...you're only going to get one shot to wow the world with this...that or hire a bunch of code monkeys pronto!

How do they expect someone to invest $250 plus if they can show what they have is ready for the big lights? Aegia seemed to be alot of the right moves, but if this is true they'll certainly prove they're capable of making some bad moves to.

As a likely early adopter I expect some problems but not to the tune of 50% performance and 50% features loss due to driver issue...that's a bit too much to swallow.

[wireframe]

New Info:

Not so good too... =(

We sat down the company and witnessed two demos running live on first-run, A0 silicon. The final silicon that will ship in boards is A1—AGEIA made a simple metal spin, but made no major changes to the silicon design. It'll be produced on a standard .13 micron CMOS process at TSMC.
Good

Initially, it will only come only in a PCI card, with PCIe cards expected further in the future.
Bad, if I am going to spend $250 I don't want t PCI card. I want a PCIe. I do not see myself upgrading a PhysX card as often as I do a video card or sound card, unless they give me new must have features.

A large rocky hillside had about 4,200 boulders dropped at the top, which all bounced, tumbled, and interacted in a realistic (and speedy) fashion. AGEIA claimed that a dual-core CPU can handle maybe 800-1,000 in a demo like this, but was quick to note that 4,200 boulders was nowhere near the capability of their chip. There's a driver issue right now where a lot of the timings need to be worked out between the massively parallel math units in the chip. Within a couple of months, the company will have a new driver which will enable them to raise the boulder count to 32,000. They're confident they can reach that number, but even if they can only get halfway there, 16,000 to 20,000 boulders is a lot better than a CPU can do.
ACK! So right now it doesn't work 100%. It is beta but it just scares me.

At launch, the PPU cards will only handle rigid-body and particle-based fluid dynamics physics calculations. A driver update early in 2006 will enable both soft-body and hair/clothing acceleration as well.
ACK ACK! Pay now, we will enable the PureVideo... er Other Cool Stuff(tm) later. This is very bad in my book!
www.extremetech.com (http://www.extremetech.com/article2/0,1558,1817922,00.asp)

I want them to be successful, I want cooler physics in my games, but this is not good news. I know they still have time to fix things, but when I want something badly I get a bit skeptical even though I try to remain positive. =)

Good opinion at Arstechnica
After reading about these demos, I stand by my conclusions from my original post: "it's difficult to envision a game that is essentially the same but comes in two different versions: one that has a standard software physics engine, and one that has a hardware-accelerated physics engine that's such a quantum leap in gaming experience that you absolutely have to run out and buy an add-in card." I guess we'll see soon enough if developers can make this happen.

My goodness. These guys need a new marketing department. I suppose it's nice when companies come forward and state the good with the bad, but this is just plain stupid. Why even talk about these things and worry the consumer?

Oh well, it will be interesting to see how this device will fare if the first generation flops. Anyone think there will be enough support to keep pushing something like this unless there is immediate uptake?

As much as I like the idea of dedicated hardware and this concpet of a PPU is intriguing, I really don't see this working unless they have some killer app at launch and a pricepoint that makes gamers scramble for the device. Maybe they are rushing this a bit too much. Perhaps they would be better served to wait, work on a second revision, let PCIe mature, and make sure they have a launch title to bundle, or at least supply a demo of.

[Richard]

I read it's going to be $249-299, which is a little too much in my book. I also hope they get a LOT higher than 4200 objects, as they promise. Right now, I think I'll just skip the first PPU generation altogether; I doubt there will be many games making use of it/using it to its full potential for some time anyway; if only PS3/xbox360 also had a dedicated PPU...

As for PCI, I don't have a strong opinion either way. PCI is still present in PCI-e boards and I suspect it will continue to be that way for a while.

[Geo]

Everybody see the pic over at [H]? I have to wonder if that fan is for functional purposes or marketing ones. . .

Umm, somebody 'splain why PCI-only would be a problem? Are we getting many (uh, any?) mobos that have more pcie slots than those for the vid card(s)? AGP-user here, and not looking to upgrade anytime soon, so perhaps I just missed it. . .

[rwolf]

Sounds like the xbox360 gpu will do lots of things that Agiea will do.

[scificube]

Sounds like the xbox360 gpu will do lots of things that Agiea will do.

Yeah that's the point. You don't want the CPU to have to do it...you want it to be free to do other things like AI etc.

...err...you did mean CPU right?

[Geeforcer]

Good point about the killer app (lack there of). Voodoo Graphics had GL Quake and MachWarrior 2. What does AGEIA have?

[scificube]

Good point about the killer app (lack there of). Voodoo Graphics had GL Quake and MachWarrior 2. What does AGEIA have?

NADA! Unless UT2007 makes use of the PPU...at least there's a chance there and that's a pretty big game...

I know they support Aegia's Novodex physics engine...I can't remember if they said they would support the PPU itself allthough I've read at more than one site that Tim Sweeney supports the idea and likes the part...I just can't remember any official statement about supporting the PPU :(

[Geeforcer]

In a miracle of good timing, Gamespot just put up an interview with Mark Rein that talks about that.
http://www.gamespot.com/news/2005/05/20/news_6126181.html


GS: Epic announced a couple of months ago that it is using NovodeX physics in Unreal Engine 3. Ageia, the company behind the NovodeX API, is preparing to release a physics chip for the PC. How will UE3-based PC games take advantage of the Ageia hardware?

MR: Things that you write for the NovodeX API automatically become hardware-accelerated, automatically get faster [with a physics processor]. What's more significant isn't really that. What you get is instead of having 600 rocks rolling down a hill you have to do 6,000 rocks rolling down a hill. So you'll be able to do so much more, do much deeper simulations than what we could possibly do without it. But people who don't have it won't feel completely ripped off--they just won't get the full experience.

GS: Will there be cases where you'll detect physics-accelerated hardware and offer a game mode that includes gameplay that depends more on physics?

MR: We won't do it where it will mess up gameplay. It's a multiplayer game where you can't handicap people that don't have a piece of hardware. You have to scale for them. But in terms of the amazing visual effects and all those things going on, it'll be that much better on the chip.

GS: The PlayStation 3 and Xbox 360 will both have extremely powerful multicore processors capable of doing a lot of physics calculations. Is that going to make it difficult for developers to port console games over to the PC?

MR: The PlayStation 3 Cell architecture is very similar to the hardware design of the Ageia chip, so the PC will be able to get superaccelerated also. The Ageia NovodeX API, when they bring it over to the PlayStation 3, will be very fast, very powerful--similar to the Ageia [chip]. But the cool thing about that is if someone's lead moneymaking platform is the PlayStation 3, they're going physics up the wazoo because they have so much power. That makes it a very economical choice for PC users than to just pop a [physics] chip into the PC and then have the full physics effects that will be available on the PlayStation 3 or the Xbox 360.

Humm, sounds slightly more encouraging, but again, unless there is a big difference between the two versions it will not push hardware. It is possible that some games jointly developed for PC and PS3 will carry over some of the physics from that can be taken advantage of on the PC using PPU.

[Ailuros]

What you get is instead of having 600 rocks rolling down a hill you have to do 6,000 rocks rolling down a hill.

I shouldn't probably single out just one sentence (and especially a simple example), but I can't get rid of the feeling that this whole affair might turn out more a quantity than a quality issue after all.

*plays "Papa was a Rolling Stone" and tunes up the volume* :roll:

[scificube]

What you get is instead of having 600 rocks rolling down a hill you have to do 6,000 rocks rolling down a hill.

I shouldn't probably single out just one sentence (and especially a simple example), but I can't get rid of the feeling that this whole affair might turn out more a quantity than a quality issue after all.

*plays "Papa was a Rolling Stone" and tunes up the volume* :roll:

That's true but really this is the only way this can start out.

If they start putting the physics into the gameplay at this point people are going to get pissed. It has to be about aesthetic value until the PPU is a standard part in either being integrated into mobos or PC vendors include it like they do cheap sound cards.

When it's expected a gamer will have one then physics usage can be of more substance in gameplay...at least that's what I keep telling myself...repeatedly :twisted:

Anyway, tis good new to me...methinks UT2007 be a fine choice for killer app :)

[ndoogoo]

I know a perfect game for this PPU. X-Com :D

[Shapeshifter]

Good point about the killer app (lack there of). Voodoo Graphics had GL Quake and MachWarrior 2. What does AGEIA have?

unreal 3.

At least so Ive heard.

[pc999]

Anyone knows here I can get a video of their demos?

Thanks

[Jawed]

Don't bother, they're shit.

Jawed

[digitalwanderer]

Anyone knows here I can get a video of their demos?

Thanks
Which ones, the Ageia ones?

Was "NovodeX Rocket_1_1" theirs? 'Cause it's my fave demo in recent memory.

[pc999]

Anyone knows here I can get a video of their demos?

Thanks
Which ones, the Ageia ones?

Was "NovodeX Rocket_1_1" theirs? 'Cause it's my fave demo in recent memory.

Anyone. :lol:
Thanks.

Don't bother, they're shit.

Jawed

Sorry but I have curiosity.
Thanks for the reply anyway.

[mashie]

This (http://www.mashie.org/hardforum/physx.wmv) 200k video is quite nice.

[Jawed]

Hey, I'm feeling generous:

http://www.gdhardware.com/events/e3/2005/day1/001.htm

Don't say I didn't warn you.

Jawed

[pc999]

Thanks, and jawed you are right, they're shit.

PS3 had done a better job to show really good physics.

[soylent]

I thought that demo was quite good, it's not the graphics you should be looking at after all.

Relatively arbitrary looking damage on those crates instead of breaking in 2-3 different ways again and again(e.g. HL2). Actual breaking, not just being 'gibbed' and replaced with a hand full of splinters. Relatively few ugly bugs(only saw one ugly clipping issue). Plane breaking into bits wasn't very interesting. To be able to do that last huge crash into the boxes with no performance penalty to the CPU in an actual game would be awesome.

The demo with the house being smashed was just bloody awesome.

Argh, not another 'cute' misspelling. Physix?

[Fodder]

I said PC's got some hard times coming and a PPU IMO would help to calm the storm.
PCs are having a hard time specifically because of complexity, target platform variation and cost of entry, and you think raising all of those will help?

[IgnorancePersonified]

The picture posted on xblit looks like the one card has both pci and pcie connectors..... This could constitute the only part of my current system other than the floppy drive that makes it into any upgraded system.

[jpr27]

If this add in card is around $200-$300 you can count me out on that. Especially if the tech goes the way of the gpu and you have to upgrade every few generations. I know physics and its requirements but even though I have the $$ to spend for say bragging rights, I would rather go spend the $$ on SLI or upgrading to a dual core CPU. :twisted:

Now if the cost was around $100 I could probably be persuaded but with recent claims of driver issues (only 50% working) I would wait another generation until it worked "as advertised". :roll:

[IgnorancePersonified]

Yeh but if everyone waited till everythig worked "as advertsised" the whole computer industry would have stopped at calculators.

[nutball]

Yeh but if everyone waited till everythig worked "as advertsised" the whole computer industry would have stopped at calculators.

That says something significant about the people who buy computers don't ya think?!

[Moose]

Wow, add up the cost of a next gen Vid card or two, new mobo, a processor or two, a physics card, a huge power supply to power everything and you could easily buy a Xbox 360 and a PS3.

I'm not going to say that PC gaming is dead but it is closer than ever.

With new tech it will catch up to what the new consoles will be able to do, but at what cost???

Add to that, that more and more games are being developed for the console first and then ported to the PC (most rather crappily).

The fat lady isn't singing yet, but she's climbing the stairs to the stage. :(

[Entropy]

Wow, add up the cost of a next gen Vid card or two, new mobo, a processor or two, a physics card, a huge power supply to power everything and you could easily buy a Xbox 360 and a PS3.

I'm not going to say that PC gaming is dead but it is closer than ever.

With new tech it will catch up to what the new consoles will be able to do, but at what cost???

Add to that, that more and more games are being developed for the console first and then ported to the PC (most rather crappily).

The fat lady isn't singing yet, but she's climbing the stairs to the stage. :(

One problem is that games programmers for PCs have a high performance target. Thus they miss out on capitalizing on the greatest strength of the PC as a games platform, its installed base. How big would SIMS have been if it had targeted a top-of-the-range PC?

By escalating system demands, PC developers are locking themselves into an ever tighter niche. Still, that niche may be sustainable. But I do think there is an opportunity for those who buck the trend, and instead target something like the latest Intel integrated graphics.

[scificube]

I said PC's got some hard times coming and a PPU IMO would help to calm the storm.
PCs are having a hard time specifically because of complexity, target platform variation and cost of entry, and you think raising all of those will help?

That's a dire way of putting things but yes...if it sells more games and moves more HW. No...if it doesn't. I believe it will so I like the PPU. I also don't look at the complexity issue alone within itself. While there will be greater complexity there should also come greater flexibly and freedom. Here again I feel the good out weighs the bad.

Target platform variation is a persistent problem that will be here with or without the PPU. I feel as long as physics doesn't seep into gameplay during the interim the PPU should populate or die in the PC realm that it an issue that can be dealt with...just like supporting HDR for specific GPUs or high end parts, it should be another aesthetic layer over normal physics until that bound can be broken. I am for devs pushing the HW I buy be it the CPU, GPU or in this case the PPU but of course I don't want it to be at the expense of being to game at all for those less fortunate than I lest I one day be one of them...again.

Cost of entry...this will be high if you're a serious PC gamer as stock parts just won't due most of the time even if your shooting for the mid-range. In any case though...if not paying for a PPU one would need to pay for a uber CPU that I feel is not coming for a while in order for the physics to translate over from consoles ports. It's either this or don't do physics to this level as there is no free lunch.

Hope that helps you better understand how I feel about it :wink:

[IgnorancePersonified]

I wonder how long ms and sony are going to subsidise console hardware costs?

[DudeMiester]

Well seeing as IBM intends to sell CELL as a workstation CPU, and considering the breadth of CELL software that PS3 is going to produce, I think desktops with CELL are certainly reasonable. In this case you can throw your PPU out the window, and just throw in another CELL. As CPUs become more flexible and optimisable, perhaps one day we'll even see the GPU go bye-bye in the favour of additional CPUs. Basically, I'm sure the PC will catch up and stay in the lead but rather through a radical advancement of the PC itself, not just throwing in more components.

[IgnorancePersonified]

Sure.... 'just' throwing in an extra cell unit sounds simialr to either placing it on some sort of 'bus' or intregrating either more cell units into the main die or having it somehow integrated onto the mobo or what constitutes the 'south bridge' of a console platform.....
Does that mean they gunna throw in extra cell units regularly = are there going to be different spec consoles out there removing the main advantage to pc's?? How about the xbox - they gunna throw a cell unit in there or go with a ppu style unit for xbox2.5?

Since my pc is not a closed form factor - rather than buy a mobo with it intergrated and the associated hardware with it, or since they are not intergrated into a cpu die or 'south bridge' chip right now I would like the option to buy one and connect it up to my existing pci bus and then future pciex1 connector - that really does sound cool. Seeing how the avalanche of AGP -> PCIE early adopters and Joe Bloggs overwhelmed a company like ATI...I'm thinking "old" platforms will be around for awhiles. really - the most interesting thing to me in games would be to get one that has good support, run it on an old pci only system with ppu, a new dual core system, a new single core system etc etc - swap the parts around and write a good article on it. With - without ppu.

[Dave B(TotalVR)]

I wonder if we'll see integrated solutions. . .

IMO that's the only way this will be successful.

[Dave B(TotalVR)]

This (http://www.mashie.org/hardforum/physx.wmv) 200k video is quite nice.

I wanna see a similar comparison with a multiplayer character being Gibbed :twisted: :twisted: :twisted:

[scificube]

I'd could go with a Cell over a PPU if they become desktop parts. It wouldn't be specialized for physics but would be more flexible and have benefits elsewhere over current desktop CPU's. It'd take it if I got the chance and could afford it. Cost would be about the only major issue when deciding which way to go as I expect a PPU to be a whole lot cheaper on the average and wouldn't require me to toss my other system parts in order to use it. (definitely going to need a new mobo with a hella fast bus to support it, and very fast system memory. That's a new CPU, mobo, memory, and heck a new vid-card as to not bottleneck the system...so you see my point.)

It's a hard call. I wish I could say both but I don't have that kind of money. I'll be buying an Xbox360, a PS3, GAMES, and "A" PC part for good ports or games that require similar horsepower most likely as that's a lot money right there. A PPU most likely but if the Cell makes it to the desktop, is affordable and I can get some sucker to buy my old rig...I'd go that way too, but then my new Cell rig wouldn't be making an appearance for a good long while.

[Nemesis77]

Bad, if I am going to spend $250 I don't want t PCI card. I want a PCIe.

Why? What if PCI-E gives you no beenfits at all? there are ALOT more PCI-slots out there than there are PCIe-slots! If PCI is more than enough, the only reason for wanting PCIe would be to take part in a virtual "my penis is bigger than yours"-competition.

ACK! So right now it doesn't work 100%. It is beta but it just scares me.

Considering that the damn thing is not even available right now, I fail to see the problem. Since NVIDIA and ATI have driver-problems with SHIPPING products, I wouldn't get too worried if some unreleased piece of hardware has driver-problems as well.

ACK ACK! Pay now, we will enable the PureVideo... er Other Cool Stuff(tm) later. This is very bad in my book!

Why not wait untill the product is available at retail before using the "Jump to conclusions"-mat? So their beta-hardware with beta-drivers do not have all features enabled. Color me surprised!

You are NOT "paying them now". The product is not yet available. Why won't you wait untill you can actually buy the damn product and THEN decide whether it sucks or not.

I want them to be successful, I want cooler physics in my games, but this is not good news.

Hey, I heard that 3Dfx had problems with first versions of the Voodoo Graphics, and their early drivers (that were used internally) did not work that well. Clearly, Voodoo Graphics sucked! And I heard NVIDIA's NV1 was dead in the water as well!

[Unknown Soldier]

Hey, I'm feeling generous:

http://www.gdhardware.com/events/e3/2005/day1/001.htm

Don't say I didn't warn you.

Jawed

I downloaded the demo's from mashies post and this site . .and am not that impressed with it overall. I thought HL2's physics looked way better. It's still in beta though so should get better.

US

[TheBubba]

Sorry I'm a long time lurker. Thought I'd throw in my half cent. I agree with a lot of the sentiment here.. These won't fly. The Inquirer (I know they aren't the most reliable) said these would go for 250-300 bucks!

Inquirer Link (http://www.theinquirer.net/?article=23468)

I'm sorry but that is insane (IMHO). Maybe they'll sell a few to the guy who has to have everything but that isn't going to amount to much. They won't be be adopted in any sort of volume past the 4 or 5 guys previously mentioned.

Like people have been suggesting, integration is the only way these things will ever be adopted. Not as a $250+ card. The hype will die and so will the idea of a seperate PPU board... I think I read it in the stars, planetary alignment or some such.... :lol: If this were to ever become a requirement it would certainly make people think about jumping ship to the console world... But that will never happen. (*sigh of relief*)

[mashie]

Like people have been suggesting, integration is the only way these things will ever be adopted. Not as a $250+ card. The hype will die and so will the idea of a seperate PPU board... I think I read it in the stars, planetary alignment or some such.... :lol: If this were to ever become a requirement it would certainly make people think about jumping ship to the console world... But that will never happen. (*sigh of relief*)
I have a feeling it will be quite the opposite, when people start to play games on x360 and ps3 with a new maginitude of physics they will start to demand it in PC games as well.

Time will tell, not the stars ;)

[Dr. Ffreeze]

Nemesis77,

Why? What if PCI-E gives you no beenfits at all? there are ALOT more PCI-slots out there than there are PCIe-slots! If PCI is more than enough, the only reason for wanting PCIe would be to take part in a virtual "my penis is bigger than yours"-competition.

Where did I say I wanted a PCI-E for extra performance? PCI-E does give you a benifit regardless of performance. I don't want to buy any new PCI cards so I can protect my investment for motherboards I buy in the future. I buy upgrades vary rarely, so I will want mine to last a good number of years, say 5 or so. I don't upgrade all the time like many gamers so it really matters to me if I am going to drop $300 on a card.

Considering that the damn thing is not even available right now, I fail to see the problem.
It does not work now, in beta. That means it might or might not work in the shipping version. See the possible problem?

Since NVIDIA and ATI have driver-problems with SHIPPING products, I wouldn't get too worried if some unreleased piece of hardware has driver-problems as well.
Well, I would guess that Nvidia had issues with beta hardware with PureVideo. Now that it is shipping, it still doesn't work in the 6800 line. That is a problem. If a beta piece of hardware does not work 100%, while understandable, it 'could' also not work in shipping parts. I said I was concerned, not that it was doomed to fail. If I had heard of PureVideo not working in the 6800 line while it was in beta form, I would have been concerned. NVidia was unable to fix PureVideo, maybe AGEIA will be unable to fix this driver issue. I think it bears watching.

Why not wait untill the product is available at retail before using the "Jump to conclusions"-mat? So their beta-hardware with beta-drivers do not have all features enabled. Color me surprised!

Did you not read the quote?
At launch, the PPU cards will only handle rigid-body and particle-based fluid dynamics physics calculations. A driver update early in 2006 will enable both soft-body and hair/clothing acceleration as well.
At launch the PPU cards will not have all features enabled until early 2006. When you buy the card in Q4, they will not have all the features enabled.

You are NOT "paying them now". The product is not yet available. Why won't you wait until you can actually buy the damn product and THEN decide whether it sucks or not.
Pay for the card in Q4 2005, get soft-body acceleration in Q1 2006. Who said the card sucks? Not I! There are areas of concern.

[DudeMiester]

Selling a card for a large amount of money, with incomplete drivers, and no intention of exposings it's true general purpose flexibility, is insane. If AGEIA thinks this is going to be a success they're out of their minds. Get the drivers fully working, fully expose the general purpose floating point processors, and then MAYBE I'll buy one. Otherwise, hell will have to freeze over first. Not that I don't think a CELL like co-processor (which is what the PPU is basically) isn't a bad idea, I just thing AGIEA is going about it very badly.

[Nemesis77]

Where did I say I wanted a PCI-E for extra performance? PCI-E does give you a benifit regardless of performance. I don't want to buy any new PCI cards so I can protect my investment for motherboards I buy in the future. I buy upgrades vary rarely, so I will want mine to last a good number of years, say 5 or so. I don't upgrade all the time like many gamers so it really matters to me if I am going to drop $300 on a card.

Well, if you have PCI-only devices, then you need to choose MoBo's that have PCI-slots. And besides, PCI is NOT going to disappear in a long time.

It does not work now, in beta. That means it might or might not work in the shipping version. See the possible problem?

I guess that's why they call it BETA? Instead of getting your panties in a bunch over a piece of hardware that is not yet even available, why not wait untill it's actually available? How well it works right now is next to irrelevant to us consumers, since we cannot buy this product yet. Developers might care, but I don't see a reason why we should care.

Well, I would guess that Nvidia had issues with beta hardware with PureVideo. Now that it is shipping, it still doesn't work in the 6800 line.

I'm talking about some games not working properly and such. Since they keep on having that kind of problems with shipping products, I wouldn't be too worried about some glitches with unreleased hardware. there is a reason why the hardware is unreleased.

If a beta piece of hardware does not work 100%, while understandable, it 'could' also not work in shipping parts.

Show me one piece of hardware that worked perfectly from the very beginning. ALL hardware has such problems. that is why they are not released before their time.

I said I was concerned, not that it was doomed to fail.

Well, if you are concerned because unreleased piece of hardware doesn't yet work perfectly, then you have alot to worry about. CPU's, vid-cards MoBo's, soundcards etc. etc. all face problems in the early versions. Are you worried about those as well? Why not? Because those problems are not mentoined publicly in some website?

Did you not read the quote?
At launch, the PPU cards will only handle rigid-body and particle-based fluid dynamics physics calculations. A driver update early in 2006 will enable both soft-body and hair/clothing acceleration as well.
At launch the PPU cards will not have all features enabled until early 2006. When you buy the card in Q4, they will not have all the features enabled.

You can think of it like that it DOES have all the features enabled. in 2006 it gets an update that gives you some bonus-features.

Pay for the card in Q4 2005, get soft-body acceleration in Q1 2006. Who said the card sucks? Not I! There are areas of concern.

Why not wait untill Q1 2006 before making the purchase? Why not buy it in Q405, use it like it is then, and then get extra features in Q106? Only problem here is that they mentioned that some funtionality will be enabled in 06. Had they not said a word about that, people would be happy as a clams right now. But now that they mentioned that they will enable more functionality later, people are screaming "OMG! It's not working!".

[TomW]

It is likely to be impossible to port some PS3 games directly to PC (without modifying the content) due to the physics they will employ, unless the PC is equipped with some kind of maths coprocessor, be it a PhysX or something else (or lots of normal x86 cores). I think the move ideally needs to come from AMD or Intel for it to be successful - they could incorporate 1 or more coprocessors into their multicore chips. It's happened before with the 387 coprocessor...

PC CPUs are only going to have the horsepower to match the physics ability of PS3's CELL in the fairly distant future. What do we do in the meantime? Dual core and even quad core can't handle many rigid bodies compared to PS3 and especially PhysX.

UE3 engine games may fall into tiers of how much power you need for the physics - at the bottom level, a single core PC, next tier will be dual or quad core PC and XBox 360, top tier will be PS3 or PhysX PC. Assuming that UE3 makes porting games easy (and I hope it does), you're still going to need the necessary physics horsepower to keep up with the original CPU the game was designed for.

Tom

[Moose]

PC CPUs are only going to have the horsepower to match the physics ability of PS3's CELL in the fairly distant future. What do we do in the meantime? Dual core and even quad core can't handle many rigid bodies compared to PS3 and especially PhysX.

Tom

buy a console???

That's my plan at this point.

:(