Why does HardOCP hate the ageia chip

[iansilv]

I can't figure this out- it seems liek every day there is an editorial comment or something that rips the ppu chip- why? I mean, the thing jsut came out, and I think the idea ha s aheck of a lot of potential- whynot give ita chance for what it is?

It's a brand new product that may reallyimprove our gaming experiences- although I do need a pci-e version!

 

[pxc]

Are you disagreeing with any points they made? If so, why not offer a counter-argument instead of just whining?

 

[LoneWolf]

PCI-e won't help or hurt the PhysX card; the bandwidth on the card is what is important, not the bandwidth on the slot at this stage of the game. So, a PCI-e PhysX card would net you no more performance than a PCI one, and would probably cost more.

People here don't hate the Ageia PhysX; rather, they're questioning it with some justified constructive criticism.

Part of the current backlash against the PhysX card by its detractors is the price of $299 for a card with very little support, in addition to having just paid $200-500 for a vid card capable of playing today's games. Combine the price of those two objects and you've got an entry-level PC.

Part of the backlash is indeed that there is no "killer app" requiring the card. Wheen the 3dfx Voodoo came out, the killer app was the GLQuake mod, FREE if you already owned Quake. The difference in apperance was so remarkable for a game that so many people had to have, that it spawned a revolution, and other developers were able to code Glide wrappers that patched their games for a large improvement in graphics. Even then, I wouldn't have bought a Voodoo, but an online clearinghouse put up several hundred of them at $99 a pop, between a third and half of what they retailed for everywhere else.

There are quite a few people who could and would probably pay $99-129 for a PhysX card right now with no killer app; there's a price-point where enthusiasts are willing to risk, and then write hardware off if it doesn't pan out. But at $299, gamers want instant gratification, and rightfully so, as that's a fair chunk of change.

P.S. Potential != Results. Just ask a number of high school teachers. When the PhysX has real results, maybe people will change. Until then, I'll wait for it to live up to its potential.

 

[Ockie]

Well said lonewolf.

 

[weebob]

as above.

things may turn around, 3dfx managed it, I had a voodoo 1 when there was nothing really available to show you what it can do and was really pissed off because of this.

 

[BoredTiger]

Part of that impatience you referred to resulted in me actually purchasing one of these hehhehe

I'll let you all know if it's any good in real-world application (Ghost Recon AW). I've read a few reviews about how it slowed down framerates, etc etc....but I'll be the judge of that

I guess my main curiousity point is...will it hamper multiplayer games when online. If I'm fragging online will the card hamper my framerates, ping, etc.

I made the jump on purchasing the card after viewing the game demos at E3. The demonstration was always impressive, but actually being able to play Cell Factor and tool around in the game brought on a different sense of the card. So...I'm taking a risk here...if it sucks, I'll return it

I agree that the price point is a bit ridiculous with the support it has available...and that is probably one of the main reasons why hardware enthusiasts are not too enthused with the PhysX card introduction.

Keeping my fingers crossed though

 

[NulloModo]

Well, I will come out and say that I feel it is a worthless product. My reasons are..

1. It is another thing you will have to buy to keep your PC running the latest 3D applications. Video card prices are already ridiculous, adding another $300 for a Physics processor is highway robbery. PC gaming is too expensive as it is, we don't need to add to the cost with a product no one was really asking for in the first place.

2. The effect it could have is going to be minimal at best. Who really cares if your game has realistic physics or not? Games are fantasy, not reality, I don't want a car in a driving game to handle like the real car, I want it to handle like an arcade game approximation of the real car so I will be able to do crazy things with it.

3. With dual core CPUs becoming the norm, and quad core on the horizon, there is nothing a PhysX card does that couldn't be coded into the engine and handled by the CPU.

 

[ChingChang]

PCI-e won't help or hurt the PhysX card; the bandwidth on the card is what is important, not the bandwidth on the slot at this stage of the game. So, a PCI-e PhysX card would net you no more performance than a PCI one, and would probably cost more.

I think a lot of people are wanting a PCI-E version not because they want it to run faster, but because they don't have any PCI slots available.

 

[InorganicMatter]

3. With dual core CPUs becoming the norm, and quad core on the horizon, there is nothing a PhysX card does that couldn't be coded into the engine and handled by the CPU.

*sigh* We've been through this a dozen times. CPUs are single-threaded, and PhysX chips are highly optimized for multithreaded computations (like boxes colliding). A CPU could never even dream of processing Physics in the way the PhysX could.

 

[magoo]

There is no "hate". The product is obviously capable. Almost everybody, [H] included sees the potential in this card and its engine or API or SDK or whatever you call it.

Right now the question raised by one site was launching without a "killer" game to show-case it, others reflect that opinion in similar fashion sighting a lack of applications at present.

I just dont have alot of room left in my case, it would be nice if this was PCI-E x 1 capable so that when the right time comes I could squeeze it in.

It just may not be the right time to buy one.....I dont think anybody hates the card or the concept.

 

[MrNasty]

Well, I will come out and say that I feel it is a worthless product. My reasons are..

1. It is another thing you will have to buy to keep your PC running the latest 3D applications. Video card prices are already ridiculous, adding another $300 for a Physics processor is highway robbery. PC gaming is too expensive as it is, we don't need to add to the cost with a product no one was really asking for in the first place.

You don't have to buy it - if you're afraid it will become successful enough for that you've just admitted it's a worthy technology.

2. The effect it could have is going to be minimal at best. Who really cares if your game has realistic physics or not? Games are fantasy, not reality, I don't want a car in a driving game to handle like the real car, I want it to handle like an arcade game approximation of the real car so I will be able to do crazy things with it.

Fine, but some people like realism - there are enough games out there to cater for both, non?

3. With dual core CPUs becoming the norm, and quad core on the horizon, there is nothing a PhysX card does that couldn't be coded into the engine and handled by the CPU.

This has been debunked for ages:

Q: why don't we do graphics on the CPU?

A: because they're too damn parallel.

The same thing applies here - to get any number of objects interacting with speed, you need dedicated, short, floating point pipelines. To get the most basic fluid sims you NEED a PPU. Or a GPU even.


Look: I'm really tired of people dumping on Ageia because they can't see an immediate benefit and so draw that line out to forever and come up with "no reason to get one, ever".

If you like what you see, get one. If you don't, don't. But why spend all that time crapping on something that isn't to your taste?

You don't see me saying SUVs shouldn't be allowed - there's no reason for driving one, get a normal car. Why pay for 4x4? You don't need it on tarmac and no-one goes camping anymore.

Now go outside and count the range rovers that pass over 5 minutes.

You see? Just because you don't need it or rarely use it doesn't mean that it isn't nice having it or knowing that you can use it if you need it.

UT2k7 uses the PPU - I bought one cos I love UT and I like the idea of this technology.
I also enjoy fluids modelling, hence this card interests me professionally as well.

When you come to buy a DX10 Graphics card remember this - it won't be me pushing you into getting 2 so you can have effects physics. It'll be nVidia.

 

[magoo]

Well, I will come out and say that I feel it is a worthless product. My reasons are..

1. It is another thing you will have to buy to keep your PC running the latest 3D applications. Video card prices are already ridiculous, adding another $300 for a Physics processor is highway robbery. PC gaming is too expensive as it is, we don't need to add to the cost with a product no one was really asking for in the first place.

2. The effect it could have is going to be minimal at best. Who really cares if your game has realistic physics or not? Games are fantasy, not reality, I don't want a car in a driving game to handle like the real car, I want it to handle like an arcade game approximation of the real car so I will be able to do crazy things with it.

3. With dual core CPUs becoming the norm, and quad core on the horizon, there is nothing a PhysX card does that couldn't be coded into the engine and handled by the CPU.

You dont want to argue with these guys.....they'll put your statements in those little boxes.

 

[CrimandEvil]

3. With dual core CPUs becoming the norm, and quad core on the horizon, there is nothing a PhysX card does that couldn't be coded into the engine and handled by the CPU.

QFT

We saw it running in Direct X 9 mode, and Crytek suggested that DX10 would not bring much more to the table other than increased performance for cards that support it. Obviously, there is a lot of physics going on in the game - calculating all that foliage isn't easy - and we asked Crytek how they were handling it all. They confirmed that there are no plans to integrate Aegia PhysX support, and that all the physics routines were written by themselves. Most of the processing is done on the processor - making use of multiple cores where they're available - and some processing is even offloaded to the GPU where it can, although to quote the Crytek rep, "We're going to be keeping the GPU pretty busy most of the time just with graphics, though."

http://www.bit-tech.net/news/2006/05/11/crysis_playtest/

 

[weebob]

lol

shoot me now

 

[MrNasty]

You dont want to argue with these guys.....they'll put your statements in those little boxes.

Like this? You didn't deserve the ribbing you got, but you did pull the "lalala I'm not listening" card - a lot of people don't like that. That combined with how vaguely you defined "interact" the first couple of posts got some peoples' goats.

If you have to speak, the least you can do is listen back

I do

 

[Rich Tate]

The only thing I have said personally is that I want to see more big name support before I am willing to jump in and give a full seal of approval.

I love the idea and the technology, but for 300 bucks, I want to see some real working products, and something big too.

 

[cyks]

Well, I will come out and say that I feel it is a worthless product.

Worthless? Gross, overly emotional, exaggeration.

a Physics processor is highway robbery.

Robbery? Gross, overly emotional exaggeration number two.

2. The effect it could have is going to be minimal at best.

Number three. minimal at best? rofl, do you even know what physics is?

Who really cares if your game has realistic physics or not?

Four. I do.

Games are fantasy, not reality, I don't want a car in a driving game to handle like the real car

Do you honestly think adding a PPU to a computer eliminates the boundaries between fantasy and reality?

I want it to handle like an arcade game approximation of the real car so I will be able to do crazy things with it.

Do you even know what physics is?

3. With dual core CPUs becoming the norm, and quad core on the horizon, there is nothing a PhysX card does that couldn't be coded into the engine and handled by the CPU.

No one disputes what a CPU can or cannot do. The main point that you have o so cleverly avoided is how fast the CPU can do it. Whilst quad core CPUs may be years down the road, who's to say that a PPU will not be equally as enhanced? Surely not you...

 

[nonlnear]

Let me preface my comments by saying that I think that the PhysX card is a bad value right now. While useful for techno-curiosity, there is little compelling gameplay benefit - yet. That said, comments like the one I'm about to dissect frustrate me.

Well, I will come out and say that I feel it is a worthless product. My reasons are..

1. It is another thing you will have to buy to keep your PC running the latest 3D applications. Video card prices are already ridiculous, adding another $300 for a Physics processor is highway robbery. PC gaming is too expensive as it is, we don't need to add to the cost with a product no one was really asking for in the first place.

This is basically the standard "I don't want to spend money" answer. That's all well and good. Then don't spend your money. I'm guessing you don't have a dual 7900GTX or X1900XTX rig. I don't either, but some people do. They are probably not going to mind rolling an extra $270 into their rig.

2. The effect it could have is going to be minimal at best.

This is just plain ignorant. The first crop of demos and games ahve been a big yawn. It will be a little while before physics enabled gameplay can be said to be truly revolutionary. But it's going to happen. Buildings collapsing realistically - while you're in them; underground scenarios with physically modeled tremors and shaft collapses. These are the kind of things you can do when you can solve LARGE vector systems in real time. The reason there haven't been demos with this kind of implementation is that they require completely new game engines - rebuilt from the ground up to use physics modeling at the most fundamental level. Ageia is not in the business of writing engines. They have an API, some tools, and support, but they aren't making the games. We all have to wait for the engines to be written. But they will come.

Who really cares if your game has realistic physics or not?

Clearly not you. That's why you're probably not going to buy one.

Games are fantasy, not reality, I don't want a car in a driving game to handle like the real car, I want it to handle like an arcade game approximation of the real car so I will be able to do crazy things with it.

That's what you think games are for. There are a lot of people who enjoy accurate simulations. Those people will be thrilled with what physics can bring to the table.

3. With dual core CPUs becoming the norm, and quad core on the horizon, there is nothing a PhysX card does that couldn't be coded into the engine and handled by the CPU.

Once again, that is completely false. I could refer you to the FAQ, but the FAQ really doesn't get into the details. Modeling the physics of basically any PDE system in real time (or not) involves incrementally solving very large matrix equations.

When you look at how best to implement the algorithms involved, you find that the best hardware setup looks a lot more like a GPU than a CPU. Lots of memory bandwidth feeding lots of parallel cores. Caching is most beneficial for code - not so much for data. The data sets are just too large. And they are always changing. Look at that from a chip designer's point of view, and it's obvious that doing physics calculations well will require something that looks a lot like a GPU. You need a fast, wide memory bus, and lots of cores with short pipelines. (Well, what is really critical is that the pipeline's length not be too large compared to the time required for a complete memory read/write cycle.) No matter how many cores you tack on to an x86 CPU, it's always going to be orders of magnitude slower than a dedicated PPU of comparable feature size/clock speed. The architecture is just all wrong.

If anything kills the dedicated PPU, it will be th GPU. But that will require some changes as well. And even if GPUs are modified to the point that they can act as dedicated PPUs, it's debatable whether they will ever kill the PPU. But that's fodder for another thread.

 

[MrNasty]

QFT

http://www.bit-tech.net/news/2006/05/11/crysis_playtest/

OMG - This just in, game doesn't use Ageia PPU, read all aboooouuuutit

Good point though - Crysis is very complex and doesn't use it. I'm looking forward to it, gonna be an awesome game

Hardly conclusive that there is no benefit in using one though!

 

[CrimandEvil]

OMG - This just in, game doesn't use Ageia PPU, read all aboooouuuutit

Yeah, whatever.

Hardly conclusive that there is no benefit in using one though!

Yeah which is probably why I didn't say that.

All I did was post a link to a recent news item where a highly anticipated games is doing it's physics on the the extra core of a CPU with no need for an Ageia PPU.

IMO as it stands now the Ageia PPU doesn't look so hot but thats because there isn't anything out there that actually supports it, it's all grafted on hardly making any use of the card at all.

 

[MrNasty]

Yeah, whatever.

Yeah which is probably why I didn't say anything close to that.

Whatever

Wasn't being arsey, just your tone and QFT sounded like that.

 

[magoo]

Like this? You didn't deserve the ribbing you got, but you did pull the "lalala I'm not listening" card - a lot of people don't like that. That combined with how vaguely you defined "interact" the first couple of posts got some peoples' goats.

If you have to speak, the least you can do is listen back

I do

Hey, thanks man. I really didnt think I deserved it either. My goat got a little roasted,too. I will take the advice to heart. (man,I'm still getting kicked for that interact deal......the pain is unbearable.... )
Anyway, thanks.....youre not so nasty..... PPUs all around. ( I hate those little boxes )

 

[Brent_Justice]

Potential != Results.

well said

 

[GoldenTiger]

well said

Yes, though I think it should be more heavily noted that the PPU hasn't had its chance to show its potential yet .

EDIT: Let me clarify this statement. What I mean is, people act as though it already had its shot and failed when they rip into the idea of a PPU, whereas it hasn't had even one good chance to strut its stuff. If the stuff ends up being inadequate, sure, by all means act like you are (not you, Brent, I mean the people on this forum), but for now, stop acting as though it's already failed .

 

[fenton06]

from the little bits pieces around the web that I have read, UT2007 is supposed to have a new engine and support the Ageia card. I personally think that when UT2007 comes out, it will give the PPU a chance to show of whats its got, and then proceed to kick all kinds of ass. But until there are real world tests/benchmarks/results, I will not buy one until I need one.

 

[BoredTiger]

UT2007 uses the PPU?

I didn't notice that during the UT2007 demo at E3...I'll take a better look tomorrow morning. Gonna swing by and stop by the PS3 booth (missed it last time, crowds were a bit too big for my tastes), before work heheh

I didn't realize this...but that should make things better for the PPU if a major gaming title like UT is utilizing the technology...that can ONLY help the product.

 

[GoldenTiger]

UT2007 uses the PPU?

I didn't notice that during the UT2007 demo at E3...I'll take a better look tomorrow morning. Gonna swing by and stop by the PS3 booth (missed it last time, crowds were a bit too big for my tastes), before work heheh

I didn't realize this...but that should make things better for the PPU if a major gaming title like UT is utilizing the technology...that can ONLY help the product.

Yep, Unreal Engine 3.0 has built-in native support for the PPU, and Unreal Tournament (UT2007) 2007 is purported to have a bunch of PPU-only levels .

 

[Ockie]

UT2007 uses the PPU?

I didn't notice that during the UT2007 demo at E3...I'll take a better look tomorrow morning. Gonna swing by and stop by the PS3 booth (missed it last time, crowds were a bit too big for my tastes), before work heheh

I didn't realize this...but that should make things better for the PPU if a major gaming title like UT is utilizing the technology...that can ONLY help the product.

http://www.nvnews.net/vbulletin/showthread.php?t=70056

 

[LoneWolf]

well said

Thanks!

...you wouldn't happen to be looking for an editor in the Upper Midwest US, would you?

 

[LoneWolf]

Yes, though I think it should be more heavily noted that the PPU hasn't had its chance to show its potential yet .

EDIT: Let me clarify this statement. What I mean is, people act as though it already had its shot and failed when they rip into the idea of a PPU, whereas it hasn't had even one good chance to strut its stuff. If the stuff ends up being inadequate, sure, by all means act like you are (not you, Brent, I mean the people on this forum), but for now, stop acting as though it's already failed .

I don't think the product has failed. I just don't think it's worthy of purchase...yet.

Remember the Diamond Edge 3D? Based on the first nVidia NV1 chip, it was envisualized as the new wave, having 2D/3D graphics and even an audio processor.

The card tanked; it was hard to code for, and there just weren't enough games for it, plus it was a bear to work with for end users (I had one customer rip his out in disgust back in the day). Eventually Microsoft completely killed it with DirectX, whose rendering methods differed from that of the proprietary NV1. That didn't spell the end of nVidia nor of 3D graphics in games either; it was just the first step.

The Ageia PhysX card is "the first step". It may succeed or fail; we don't know yet, that depends on what games designed to use it are really like, and what long-term pricing turns out to be. But whether it succeeds or fails, it's opened the door to the idea of enhancing the physics of video games via hardware and now that the door is open, it's unlikely we'll close it again.

 

[bildad]

Apparently UT2007s physx implentation according to the interview is nothing more than what we saw in GRAW, effects physics with no gameplay impact. No 'killer app' here...move along.

 

[phide]

What surprises me is that Ageia seems hellbent on reasonable margins. Little of the $299 MSRP of an Ageia card is comprised of components (based on my current assumptions). We're essentially going to be paying out so they can recoup R&D expenses, and of course a portion of their profits.

It seems the ideal way to get this sort of product in the market is to saturate it. Why not break even, ride on financiers for a while and flood the market with the hardware so there's a momentum for software implementation? Without a good portion of consumers with the capabilities, there will be no developer support past useless visual simulations. Present the PPU cards to the market like Sony or Microsoft present their consoles, and the end result will be market dominance.

 

[hardwarephreak]

Yes, though I think it should be more heavily noted that the PPU hasn't had its chance to show its potential yet .

EDIT: Let me clarify this statement. What I mean is, people act as though it already had its shot and failed when they rip into the idea of a PPU, whereas it hasn't had even one good chance to strut its stuff. If the stuff ends up being inadequate, sure, by all means act like you are (not you, Brent, I mean the people on this forum), but for now, stop acting as though it's already failed .

I think what most people are stating though is THIS is its' chance to strut its' stuff. The product is out, and with all that potential going to waste...I thought UT2007 would be the game that would do it for them, but now all I see in interviews is that its just for the eye candy, not for physics that IMPACTS game play. I don't really think people are so naive to not understand the potential for this product, nor its' abilities, however, I do believe the problem lies in priceerformance(use) ratio.

Lets say I go out and buy this new sports car. Has all the cool features, and this monster engine that makes the car go really fast. Plus, it cost a pretty penny too. But now that I have it, I am only allowed to really rev the engine (demos) or drive around the block (current game PPU usage). All of which I can do with my old car. So the point in me spending my money on that sports car is what? Sure it can really go if I gave it the chance, but I can't due to no real place to go put the pedal to the medal.

Maybe that's a crappy analogy, but I think it is a valid point. Sure the power is there, and there is potential, a lot of potential...but don't point to the glowing particles and go "see, I told you it was worth it" becuase that doesn't do it for me when I know the thing is capable of soooo much more.

I am going to be picking up a copy of UT2007 the day it comes out, and until I read that Sweeney interview I was going to pickup a PPU. Now I think I will wait. There are other things that I can spend $300 on that will enhance my game play.

 

[MrNasty]

I think what most people are stating though is THIS is its' chance to strut its' stuff. The product is out, and with all that potential going to waste...I thought UT2007 would be the game that would do it for them, but now all I see in interviews is that its just for the eye candy, not for physics that IMPACTS game play. I don't really think people are so naive to not understand the potential for this product, nor its' abilities, however, I do believe the problem lies in priceerformance(use) ratio.

That's a fair analogy - I don't think anyone disagrees with you that the range of games out right now is disappointing with the possible exception of CF.

I'm going to join the ranks of unhappy campers who are upset that UT2k7 won't be using PhysX for much - but it is a chance to see how PhysX takes the load off the CPU as many users were hoping to be the case.

Also don't forget that as it's been built from the ground up using PhysX and is a UT game there is one salvation:

Mods

If anyone is innovative enough to push the envelope on physics gameplay it is UT's modding community - THAT should be something to look forward to.

 

[dotK]

They don't "hate it." They said it like it is. At the moment, the PPU is completely useless. It's barely supported by one game and in that game it does not even increase the immersion or gameplay quality (it can be argued that it even lowers it).

 

[BoredTiger]

Ok, I have to disagree that it is useless

I love it. I'm really enjoying my physX card and it's a helluva lot of fun to play GRAW and CELLFACTOR(demo) with it.

It's definitely adding a whole new dimension to the gaming.

I don't know if I'm benefitting from the dual core, but everything runs fine

 

[darkeen]

*sigh* We've been through this a dozen times. CPUs are single-threaded, and PhysX chips are highly optimized for multithreaded computations (like boxes colliding). A CPU could never even dream of processing Physics in the way the PhysX could.

As much as I don't want to defend nullo he has a point. Yes the PPU can do more and do them faster. The solution is use the CPU to track objects that relate to the way the game world acts. No eye candy. Bam now your down to a very small number of objects that need tracking. Just use your extra video card power to fill in the eye candy if you want it.

The real problems this card has it that there is no entry level item and no other applications make use of it. The video card industry pays for the development of out high end cards with the sales of low end cards. The ppu is high end only.

 

[Terra]

Well, I will come out and say that I feel it is a worthless product. My reasons are..

3. With dual core CPUs becoming the norm, and quad core on the horizon, there is nothing a PhysX card does that couldn't be coded into the engine and handled by the CPU.

A Quad Core Conroe would still loose out in physics calculations compared to AGEIA's PhysX...

Terra - Would you also run quad-core and no GFX?

 

[Xipher]

Big difference between hating and being critical.

 

[Night_Hawk-19]

Especially if you use 2 or 3 cores for processing and 1 or 2 for physics. Not just Intel coming with quad cores but AMD as well. I find PSU Ageia too expensive especially in Canada where retailers want 300-400 bucks for a addon card.

You can almost buy a high end graphics card or a good cpu for that price. If prices drop to 200-300 in Canada I will consider it but not enough games available to support having psu taking up a pci slot. I think a better route would have been to integrate the PSU on the video card or on the motherboard.