Game Devs Only Use PhysX For the Money, Says AMD

arcticstoat writes "AMD has just aimed a shot at Nvidia's PhysX technology, saying that most game developers only implement GPU-accelerated PhysX for the money. AMD's Richard Huddy explained that 'Nvidia creates a marketing deal with a title, and then as part of that marketing deal, they have the right to go in and implement PhysX in the game.' However, he adds that 'the problem with that is obviously that the game developer doesn't actually want it. They're not doing it because they want it; they're doing it because they're paid to do it. So we have a rather artificial situation at the moment where you see PhysX in games, but it isn't because the game developer wants it in there.' AMD is pushing open standards such as OpenCL and DirectCompute as alternatives to PhysX, as these APIs can run on both AMD and Nvidia GPUs. AMD also announced today that it will be giving away free versions of Pixelux's DMM2 physics engine, which now includes Bullet Physics, to some game developers."

Comments

They wish they'd thought of it first

[EvolutionsPeak]

Sounds to me like AMD just wishes they'd thought of it first. There's no reason AMD couldn't offer similar deals.

Re:They wish they'd thought of it first

[hedwards]

But, should they? If a developer doesn't want to use PhysX, they shouldn't. If they're doing it purely for money, then chances are that it's damaging to the industry. Sure physics acceleration is cool for certain types of games, racing games and FPS, but the problem is that developers shouldn't be paid to use technology that isn 't helpful for creating quality games.

Especially if it causes games to be less enjoyable on other hardware platforms. I could see a real problem with this in terms of anti-trust actions.

Re:They wish they'd thought of it first

[Anonymous Coward]

PhysX adds nothing to the game play. It's just stupid clutter on the ground... At least in any games I've played that use it. As an Nvidia user, PhysX is no longer a reason to keep with the brand... at least now that I have used it.

That said, it's no surprise to me that game developers wouldn't support it without incentive

Re:They wish they'd thought of it first

[InsaneProcessor]

It reads like sour grapes to me. If there wasn't enough money in doing it (either promotional or customer desires), then they wouldn't be doing it.

Re:They wish they'd thought of it first

[Anonymous Coward]

Great, so we'll have two graphics card manufacturers bribing games developers with money to implement visual features that only work on their graphics cards. Although the cynical part of me thinks 'they might not be able to afford it', the more benevolent part thinks 'because the concept is absolutely not in the best interest of customers'. In an economic analysis, at the very least, the costs spent by NVIDIA and ATi will be absorbed through their graphics card prices (so consumers pay in any case), with a net productivity loss from the developers implementing graphics features twice that could have been based on a single model.

Re:They wish they'd thought of it first

[Fwipp]

Did you miss the part where OpenCL and DirectCompute run on both NVIDIA and AMD graphics cards, and that AMD is promoting an open industry standard instead of a proprietary vendor-specific API?

Because I know it was really obvious, and sort of the entire point of the article, but it really sounds like you did.

Re:They wish they'd thought of it first

[AndrewNeo]

There's nothing PhysX actually adds to the experience of a game that Havok (physics done in-CPU) doesn't. Sure, PhysX can handle more intense simulations, but I don't see how it could improve, say, Half-Life 2's implementation of Havok.

Re:They wish they'd thought of it first

[NeutronCowboy]

Look at the alternative: instead of adding useless physics to a game that doesn't need it, they could be adding advertisements. Advertising dollars are dollars nonetheless, and I very much prefer a quick splash screen of "powered by PhysX" and some mindless physics interactions than an in-game billboard (possibly even updated over the Internet, shudder).

Re:They wish they'd thought of it first

[DIplomatic]

I heard game devs are only incorporating the "third dimension" into games because of the money, not because they want to.

Re:They wish they'd thought of it first

[BatGnat]

Just because some companies use PhysX for pretty effects only, does not mean that someone else wont come along and use it for something cool that will add something to gameplay...

Re:They wish they'd thought of it first

[liquiddark]

I don't think PhysX is doing anything to slow that particular vector of suffering. Those studios that would sacrifice goodwill for additional funds are almost certainly just waiting on the appropriate framework.

Re:They wish they'd thought of it first

[jpmorgan]

Suggesting that OpenCL and DirectCompute are alternatives to PhysX is analgous to saying that OpenGL is an alternative to Unreal Engine.

The basic reality here is that four years ago NVIDIA decided invest a lot of money in making GPUs more general purpose, to apply them to more problems than just 3D rendering. ATI didn't care and just focused on making the fastest 3D card possible. Today there are alternatives to NVIDIA's technology, most notably OpenCL... but it's worth remembering that OpenCL is very strongly derived from CUDA. In fact, most of the OpenCL spec looks like they ripped it out of the CUDA spec and changed the function calls from cudaSomething to clSomething.

So yes, open standards are good. But in this case it does smell strongly of sour grapes. ATI made several bad business decisions and have been left playing catchup.

Re:They wish they'd thought of it first

[Vanderhoth]

I very much prefer a quick splash screen of "powered by PhysX" and some mindless physics interactions than an in-game billboard (possibly even updated over the Internet, shudder).

Isn't flashing "Powered by PhysX" at the beginning of the game more annoying then driving past a billboard or seeing a commercial on a TV as you run past it? I know I get annoyed when I sit down to play a game and have to sit through five minutes of logos "From studio x, Powered by Chip-set-something, in association with company z". I'm not saying advertising the other way is necessary, useful or not annoying to a lesser extent.

Re:They wish they'd thought of it first

[jpmorgan]

Actually, cuSomething to clSomething. OpenCL is basically a copy of the the CUDA driver API, which prefixes its functions with cu. The runtime API (a higher level API) is what uses the cuda prefix.

Re:They wish they'd thought of it first

[Anonymous Coward]

I did not miss it, my response was specifically to OP's "There's no reason AMD couldn't offer similar deals."
I assumed he referred to AMD adapting the strategy of NVIDIA, and I gave two reasons not to (they can't afford it, and, it would be bad for customers in the long run)

Re:They wish they'd thought of it first

[DragonWriter]

But, should they? If a developer doesn't want to use PhysX, they shouldn't. If they're doing it purely for money, then chances are that it's damaging to the industry. Sure physics acceleration is cool for certain types of games, racing games and FPS, but the problem is that developers shouldn't be paid to use technology that isn't helpful for creating quality games.

The payment could just mitigate the risk associated with bearing the extra cost of adding PhysX to a game when not all of the market can utilize it and there is limited experience with it in the developer community. That doesn't mean its bad for the industry, or bad for the quality of the game.

Especially if it causes games to be less enjoyable on other hardware platforms. I could see a real problem with this in terms of anti-trust actions.

Really? Can you point to any provision of anti-trust law that this would violate?

Re:They wish they'd thought of it first

[UnknownSoldier]

Spoken like someone who has never had to

a) animate + draw 2D sprites doing:

- neutral pose
- run pose
- neutral with weapon
- neutral with shield
- running with weapon
- running with shield

for ALL THE FRAMES. I haven't even m mentioned the other million permutations of the avatar+enemies in states such as poisoned, etc,

OR

b) programmed a graphics engine that has had to light said 2d sprites.

3D "won" because of it scaled up content creation. i.e. The convenience of animating, texturing, lighting and shading blows away the sheer amount of worked needed in 2D content creation. It never was about the money, or quality. It was about betting on a technology that would eventually be "good enough." 2D still looks better because it is easier to make something look good from one angle with fixed lights. /rant off...

Re:They wish they'd thought of it first

[Z34107]

It's a difference in scale over Havok. I haven't had much time to play video games lately, but I saw a particularly nifty shot from Arkham Asylum. Shoot a bookshelf without PhysX and it falls over. Shoot it with PhysX and suddenly every individual page from every book flies through the air, each tracing its own path down from the sky.

So, you can do physics in Havok. But not on that scale.

I'd suspect that it's not being used for anything other than "ground clutter" is because you can't design your game around PhysX - not everyone has an NVIDIA card. So, PhysX has to be optional and can't change gameplay - which pretty much relegates it to ground clutter.

Re:They wish they'd thought of it first

[Dishevel]

Actually I prefer the billboards. Imagine. Now a game dose not become monetarily successful based solely on number of units sold. But with in game advertising (sold on a per impression basis.) Then a good portion of the revenue for the game would be how often and for how long people play the game. Which means the developers have an incentive to make me play the game over and over. Free add ons and DLC. all for the price of a billboard in the game I can look at or not. If the advertising becomes too intrusive then it will take away my enjoyment of the game making me play it less and making the ads less effective and giving the game devs less money.

I think it will be great fro gaming and gamers.

Re:They wish they'd thought of it first

[icebraining]

Games like GTA (at least San Andreas) already have billboards, would it be so bad if they had real companies instead of fakes? I wouldn't care.

In fact, I modded some of those to look like real ones, using The Sopranos wallpapers :)

Re:They wish they'd thought of it first

[Anonymous Coward]

Yeah, but in truth, Nvidia didn't either. They bought Ageia to get this, but it was a smart buy.

FYI, at first you had to pay developers to use 3D at first too.

Oh and I'll give people another piece to chew on: AMD (and ATI before them) has the worst developer relations group. They utterly suck at anything useful other than throwing money themselves. How many millions of dollars did ATI spend again to gain a couple FPS (less than 10) in Source engine again?

Hypocrites shouldn't throw stones. Then again, it's never stopped Nvidia either...

Re:They wish they'd thought of it first

[chronosan]

I heard game devs are only incorporating Video Graphics Array output into games because of the money, not because they want to.

Re:They wish they'd thought of it first

[Totenglocke]

Especially if it causes games to be less enjoyable on other hardware platforms. I could see a real problem with this in terms of anti-trust actions.

Really? Can you point to any provision of anti-trust law that this would violate?

Exactly - that's like saying that there's an anti-trust suit just because Modern Warfare 2 looks better on PS3 than on the 360 or because it looks better on a system running a 295 GTX than on a system running a 9800 GT.

Re:They wish they'd thought of it first

[afidel]

In my experience if you try to push PhysX that hard on a midlevel GPU or lower it trashes the FPS to unplayable. In fact I disabled GPU based PhysX and had the driver fallback to CPU rendering because in most games I have spare cores but the GPU is already being pushed to the max. With a quad core CPU costing almost nothing extra over a dualcore (on desktops at least) and decent GPU's starting at $100 and quickly going up from there I think that's probably the norm for the vast majority of gamers.

Re:They wish they'd thought of it first

[LoRdTAW]

Mod parent up! Why did the summary compare two API's (OpenCL and Direct Compute) to a physics engine. Also, wasn't there an article a little while back that announced Nvidia is porting physx to OpenCL?

Any why is AMD promoting Direct Compute as an open standard? Direct Compute is Microsoft's GP-GPU API (like CUDA and OpenCL) but is part of DirectX which does not run on MAC or Linux etc. AMD should promote OpenCL over Direct Compute for game and application use.

Re:They wish they'd thought of it first

[Khyber]

"So, you can do physics in Havok. But not on that scale."

We didn't need physx acceleration to do that - we demoed that kind of fun stuff in pure software using 32-bit system vs 64-bit system several years ago - Far Cry, anybody?

Re:They wish they'd thought of it first

[hughJ]

Using a game like Arkham Asylum as a comparison between physx and havok is silly. You've got one physics path that's been specially developed to showcase the strengths of a particular brand of hardware, while the other is left as a compatibility fallback for everything else. That's not an apples to apples comparison. With that same reasoning one could say that ATI has worse anti aliasing than nvidia, simply because the game shipped without support for it for ATI.

Re:They wish they'd thought of it first

[644bd346996]

No, it's like saying that Skype can handle a bigger conference call on a Pentium 4 than on a Phenom II.

My take is that advanced physics engines can add significantly to games, but if Nvidia is using their position in the graphics market to push their proprietary physics engine, that's bad for consumers and of questionable legality.

Re:They wish they'd thought of it first

[DIplomatic]

I heard game devs are only incorporating the "third dimension" into games because of the money, not because they want to.

Spoken like someone who has never had to

a) animate + draw 2D sprites doing:

- neutral pose - run pose - neutral with weapon - neutral with shield - running with weapon - running with shield

for ALL THE FRAMES. I haven't even m mentioned the other million permutations of the avatar+enemies in states such as poisoned, etc,

OR

b) programmed a graphics engine that has had to light said 2d sprites.

3D "won" because of it scaled up content creation. i.e. The convenience of animating, texturing, lighting and shading blows away the sheer amount of worked needed in 2D content creation. It never was about the money, or quality. It was about betting on a technology that would eventually be "good enough." 2D still looks better because it is easier to make something look good from one angle with fixed lights. /rant off...

*slowly pokes head above desk* Is the rant over? I was attempting to make a witty remark about technology and such.... Anyway I like chronosan's post better.

I heard game devs are only incorporating Video Graphics Array output into games because of the money, not because they want to.

Re:They wish they'd thought of it first

[ivonic]

Since when are anti-trust actions based on law? I got the impression it was a case of suing as your competitors as often as possible to try and catch them out on some semantic technicality...

Re:They wish they'd thought of it first

[Z34107]

Very true - I was thinking of Valve's "Source" engine when I read "Havok." My brain had too little coffee for real-time posting simulations.

Re:They wish they'd thought of it first

[Nikker]

The developers are getting paid to develop, so how is this a bad thing for umm developers? How do we know it's not a move on the part of the game house to hold out and make Nvidia pay their developers to add a bit to the game? And how is adding more features to a game a bad thing? Won't someone think of the poor developers who are given more money to do their jobs?

Re:They wish they'd thought of it first

[chronosan]

w00t! validation!

Re:They wish they'd thought of it first

[Draek]

But they shouldn't have to, that's the thing. Microsoft and Intel have already gotten in trouble for offering similar deals to OEMs to favor their products (the latter also in detriment of AMD), so it's not unconceivable that NVidia could be hit with a similar problem as well, as they also hold a significant share of the market in question and this could very well be interpreted as a move designed to keep a hold of it by unfair means.

Re:They wish they'd thought of it first

[TheVelvetFlamebait]

Especially if it causes games to be less enjoyable on other hardware platforms. I could see a real problem with this in terms of anti-trust actions.

It won't be. For one, NVIDIA is not a monopoly, and neither are the game publishers, so there are alternatives in both graphics hardware and in games. For two, "they are not allowing the creative process to organically produce the game through voluntary deals" is not the same as "they are causing the games to be worse". You would first need a reason why their offer leaves the developer unable to make their own creative decisions. You would also need to find some specific examples of games that you can prove were degraded by this offer, and have specific information about what the game would have been if NVIDIA hadn't made their offer.

IANAL, but I'm fairly confident that to successively finger NVIDIA in court, you would at least need to address the above issues. There are probably plenty of other requirements to fulfil.

Re:They wish they'd thought of it first

[Totenglocke]

First, no one is forcing companies to do anything. Most companies DON'T use PhysX. Secondly, PhysX is only used in GAMES, hardly a "necessity" (and yes, I'm a gamer). Thirdly, if you think this is "of questionable legality", then I'm assuming that you think Apple should be sued to hell and back for locking the iPhone / iPod to iTunes, right? Or the fact that I can't use my AMD processor with an Intel chipset, that should be grounds for suing too, right?

You're also forgetting that no game the uses PhysX FORCES you to use it - it's an option that you turn on, just like AA or AF. If you want to use that option on a game, then you need an nvidia card. Just like how if you want to turn on other options to improve graphics, you need a better quality video card than the entry level crap most people have. That's not "collusion", it's "Here are the requirements to do X" and then it's up to customers to decide if they want to do the required stuff or not.

Re:They wish they'd thought of it first

[Anonymous Coward]

Yes. How will they know how what adverts to show? By tracking web usage, surveys,
game usage metrics and other info they get by spying on you.
Also it'll probably add delays while the ads are downloaded.

If they offered a discount for games with ads it might be ok but they're making more money
by forcing us to put up with ads in games. The cost of developing games hasn't gone up in a while now.
What has gone up is how much money they waste on voice actors, making cut-scenes and other bits
that add little to the game.

Its just pure greed. They see how the RIAA/MPAA members repeatedly rip-off their customers and they want
to do the same.

Re:They wish they'd thought of it first

[Anonymous Coward]

Spoken like someone who has never had to...

a) build + animate + texture + shader 3D models doing:

- neutral pose
- run pose
- neutral with weapon
- neutral with shield
- running with weapon
- running with shield

For ALL THE FRAMES. I haven't even mentioned the other million permutations of the avatar+enemies in states such as poisoned, etc.

OR

b) programmed a graphics engine that has had to light said 3D models.

3D won because it's more immersive. I happen to be a 3D artist and for you to suggest that pure 2D is more difficult is simply ridiculous. 3D requires all of the work of 2D plus quite a lot more.

Re:They wish they'd thought of it first

[Cornelius the Great]

Not necessarily. Gamers, while a diverse stereotype across age, race and sex, mainly seem to have quite a few interests in common. In particular, in-game advertisements for other games, computer/gaming hardware, audio equipment, TVs, soft drinks and fast-food restaurants seem to be a popular favorite. EA didn't require spyware when they put in real-ad billboards in Burnout Paradise and Battlefield 2142. And certain games that cater to specific interests may even have more advertising opportunities: I'm sure Gran Turismo 5 (if it ever comes out) wouldn't need to poll gamers before putting in in-game signs for Honda and BF Goodrich.

In some cases, real product placement is not only okay, but preferred. In GTA, I'd rather see Pepsi/Coke and Taco Bell ads than some fictional "Sprunk" or "Cluckin Bell" ads- it pulls me out of the game when developers need to make up brands like that.

Re:They wish they'd thought of it first

[Thanatiel]

The day I start to see obnoxious/too much advertisement in a game is the last day I buy a game of that company.
The day they all do it is the last day I buy a game.
I'll do reruns of the last 25 years instead.

Re:They wish they'd thought of it first

[Xrikcus]

Unfortunately from the physics simulations I've been working with I'm pretty sure that what they've done there is simply removed physics activity from the non-accelerated version rather than adding it to the accelerated one. A sneaky way of making GPU-accelerated PhysX look better. I'd be shocked if those book effects wouldn't be just as easy at the same framerate on the CPU unless there are truly ridiculous numbers of books.

Re:They wish they'd thought of it first

[kalirion]

I very much prefer a quick splash screen of "powered by PhysX" and some mindless physics interactions than an in-game billboard

How about an in-game banner that realistically flaps in to the wind with the power of PhysX?

Re:They wish they'd thought of it first

[makomk]

In fact I disabled GPU based PhysX and had the driver fallback to CPU rendering because in most games I have spare cores but the GPU is already being pushed to the max.

PhysX doesn't use multiple cores very effectively, from what I understand. This isn't surprising - NVidia want you to use one of their GPUs, not CPU power.

Re:They wish they'd thought of it first

[im_thatoneguy]

The original assertion is BS imo.

"Developer should use OpenCL". Fine. AMD do you want to be the one who writes the actual physics code implemented in a multi-threaded OpenCL app? That's what I thought.

Nvidia offers a product which is relatively easy to implement.

Re:They wish they'd thought of it first

[im_thatoneguy]

So even if you play a modern game set in say... Times Square you would vow off the video game because it accurately recreated the real world?

This seems like a stupid (and dishonest) assessment of the world. "I can only have fun if there are no ads in view." Do you live in a black vault free of marketing's evil reach?

Re:They wish they'd thought of it first

[Miseph]

Seriously? The GTA parody products are hilarious. Easily the second best thing about playing them (behind, of course, banging a hooker then beating her to death with your bare hands to get your money back... in a playground... after you've amassed a fortune so vast that getting your money back is clearly pointless).

Besides... do you really think that any mainstream companies will want to be associated with that franchise? Can you imagine what that would do to whatever "family-friendly" image they might have or want?

Re:They wish they'd thought of it first

[icebraining]

Yes. How will they know how what adverts to show? By tracking web usage, surveys, game usage metrics and other info they get by spying on you.
Also it'll probably add delays while the ads are downloaded.

They won't track anything more than game usage because the game can't access anything else, on my PC. And they can *already* be tracking game usage. Of course, if I know it is, I won't buy it. And if it's an SP game, I'll just block it at firewall level.

As for delays in game loading, that's the kind of stuff that I can easily find out after reading a couple of reviews. Buying a game without doing so first is stupid anyway.

As I said, if it's unobtrusive and doesn't shut then game down if it can't download the ads, fine by me.

Re:They wish they'd thought of it first

[icebraining]

It obviously depends on the company, but e.g., gaming hardware vendors already use GTA pictures in ads for gaming magazines.

Besides, GTA was just an example. There are less controversial games that feature 'real world'-like maps and could feature publicity without looking awkward.

Besides, they're already doing it:

All of the cars in the video game Tom Clancy's Rainbow Six: Vegas 2 are manufactured by Dodge.

Re:They wish they'd thought of it first

[Anonymous Coward]

then I'm assuming that you think Apple should be sued to hell and back for locking the iPhone

I don't know that anyone should (can) be sued, but someone definitely needs the shit beaten out of them for this one.

Re:They wish they'd thought of it first

[Beardo the Bearded]

I was looking for a yoga mat last week. /. starting serving up Yoga class ads. What bothers me more than anything else about the ads is that they're never, ever useful. I can't buy anything online, since the shipping costs and duty are inane. I'm not going to take a yoga class in Arizona when there's a gym next door.

Anyway, you'll note that the newest DRM requires a constant internet connection. It's not going to take someone a lot of brainpower to require the permanent connection to be to the ad server, or put the hearbeat on the ad server.

Re:They wish they'd thought of it first

[Beardo the Bearded]

"All the songs you got tired of 20 years ago."

Re:They wish they'd thought of it first

[Toonol]

3D requires all of the work of 2D plus quite a lot more.

I don't believe that at all.

Obviously, a good 3d game requires more work than a bad 2d game. But sprite animation is more labor intensive, just like a hand-drawn animated movie is more work than a CGI movie of approximately equal quality.

The most beautiful game of 2009 was a 2d hand-drawn game... Muramasa:The Demon Blade.

Re:They wish they'd thought of it first

[hairyfeet]

Uhhh...dude? When was the last time you used ATI, 1998? I have been building a LOT of AMD based systems since Phenom came out and the "bang for the buck" got so crazy, and you know how many problems me and my customers have been having with the ATI chips? Zero. Zipola. Zilcha, nada, squat.

Frankly since being bought out by AMD their drivers and hardware have been nothing but good from what I've seen, and I've been building everything from single core semprons (great for ultra quiet HTPCs) to Phenom II quads, all with ATI GPUs, both onboard and discrete, and so far not a single lick of trouble. For users that only casual game and watch video the new onboards accelerate just about every format and will even play games like Bioshock at an acceptable framerate, and the 4xxx and 5xxx chips are cheap, fast, and can crank out the pretty.

Considering that Nvidia has pretty much been rehashing the same chips since 07, and from what I've been reading Fermi is gonna be a GP-GPU first and foremost with P4 era space heating capability, maybe it is time to try one of the new ATIs, hmmm?

Re:They wish they'd thought of it first

[PitaBred]

Playing catchup? Is that why the Fermi pre-release benchmarks just barely beat out the performance of the currently available in quantity ATI cards, at a much higher thermal and die-size budget?

Re:They wish they'd thought of it first

[jpmorgan]

Absolutely. If all you're interested in is raw benchmark performance, the Radeon 5870 is for you. But the Fermi is a much more interesting device. Feature wise, it is an enormous step beyond any of ATI's offerings. Not something your average moron gamer cares about, but important for anybody who knows what OpenCL and DirectCompute actually are.

Anyway, we'll have to see what 3d performance is like once drivers have been optimized for the new memory architecture.

Re:They wish they'd thought of it first

[Hurricane78]

Did you even read the summary to the end? It says so right there, that this is what they will start doing now:

AMD also announced today that it will be giving away free versions of Pixelux's DMM2 physics engine, which now includes Bullet Physics, to some game developers.

You were more right with your comment, than you could imagine. :)

Re:They wish they'd thought of it first

[UnknownSoldier]

Really shouldn't respond to anonymous trolls, but whatever...

> 3D won because it's more immersive.

Vector graphics, and Flat-shading wasn't immersive. 3D looked like shit when it first came out, and looked like crap for over a decade. Methinks you need to go play some old game and learn what low-poly modeling means. (Intentionally ignoring SGI because no one could afford texture mapping in hardware back then.) People were doing 3D graphics on 8-bit computers, such as BattleZone, Skyfox, F-15 Strike Eagle, in the early 80's.

3D in the consumer/games space took off in '96 when glQuake came out because everyone saw the potential for real-time rendering realism. The immersive factor was a side benefit of having complete control over the camera (and rendering the world from any view point.)

It wasn't until the poly count started taking off that focus to 3D happened. Grand Theft Auto 3 showed that gameplay was the driving factor rather than whether a game was 2D or 3D.

> for you to suggest that pure 2D is more difficult is simply ridiculous.

It _was_. If you can actually pull your mouse away from ZBrush for a moment you might want to review your history. Before you had to worry about normal maps, glow maps, reflection maps, and High Poly and Low Poly models, yes 2D was more difficult then 3D.

Re:They wish they'd thought of it first

[Anonymous Coward]

That was the GP's point. His video card (GPU) is already maxxed doing graphics while his 4 core CPU is sitting on it's 3 spare thumbs, so he can easily afford to turn off PhysX and have better graphics response while his spare CPU cores handle physics.

Re:They wish they'd thought of it first

[bit01]

if it's unobtrusive

You're being foolish. An unobtrusive ad is a pointless ad. Unless you take subliminal advertising seriously the whole point of advertising is to gain your attention and to crowd out alternate points of view. Every single in-game ad is a lost opportunity to put in something entertaining. It reduces the quality of the game. And because the advertising industry doesn't know when to stop (it's an arms race to get mindshare) they can and do keep saturating every media they can get hold of until that media has a net value to you of just above zero. Broadcast TV is already a wasteland because of it and the parasites are trying to destroy every other form of media also.

---

The majority of modern marketing is nothing more than an arms race to get mind share. Everybody loses except the parasitic marketing "industry".

Re:They wish they'd thought of it first

[Anonymous Coward]

Uhh, no. Tell me, have YOU ever built an entire, intricately detailed and interactive world, created all of the models and characters to inhabit that world, uv unwrapped the models, created all of the textures and shaders for them and animated each frame of each model's movement by hand (including facial expressions). I have. I have also done a lot of 2D only sprite work, which is tedious but nowhere near the level of difficulty nor organisational nightmare that 3D work is.

Most beautiful according to whom? I could just as easily point to a large number of 3D games that have won random awards for their visuals too.

Re:They wish they'd thought of it first

[bit01]

So even if you play a modern game set in say... Times Square you would vow off the video game because it accurately recreated the real world?

Are you being deliberately obtuse? The whole point of a game is to escape from the real world. When the real world intrudes that's when it stops being fun.

This seems like a stupid (and dishonest) assessment of the world. "I can only have fun if there are no ads in view."

You're the dishonest one. Every single ad is a cognitive load to ignore, a cost to the viewer. The only way to have zero cost is to have zero ad's. Or useful unsolicited ad's. As if. Anybody who makes any purchase decision at all based on unsolicited ads is a fool.

Do you live in a black vault free of marketing's evil reach?

No, he lives in the real world where unsolicited advertising has reached ridiculous levels, costing billions, probably trillions, of manhours for fuck-all in return. And yes, most unsolicited advertising is evil, particularly on hot" media like TV, in that it actively attempts to influence people to make emotional, sub-optimal decisions. Spamming is is merely the extreme end of that spectrum. Unfortunately becoming less extreme all the time.

---

"Advertising supported" just means you're paying twice over, once in time to watch/avoid the ad and twice in the increased price of the product to pay for the ad.

Re:They wish they'd thought of it first

[Calinous]

Even Syndicate had billboards - but they weren't any more than some game backgrounds

Re:They wish they'd thought of it first

[makomk]

Except that unlike NVidia, they will actually let it run on graphics hardware other than their own...

Re:They wish they'd thought of it first

[Thanatiel]

Yes, ridiculous enough level to put me off TV already. (16 years and counting)

I also make a point avoiding labels whose ads were:
_ a sing-song
_ insultingly stupid
_ misleading (ie: "This TV uses 30% less power!" ... ok ? than what ?)
_ heard too often (more than a couple of times a day)

To make them "pay" does not cover all the annoyance but it helps.

Anyway ... moving on.

Re:They wish they'd thought of it first

[Creepy]

I'm going to quibble with you here, and there is a distinction to be made, but the game industry blurs the lines quite a bit so the distinction is more difficult than it would be for music or movies.

The RIAA and MPAA are there for one reason - to make sure PUBLISHERS get money. A publisher doesn't create anything, so they rely on license enforcement and sales to make all of their money. The main interest publishers have in collecting info on you is to know what kind of stuff you'll buy and know what to publish. A publisher's job is similar to a store manager knowing what you're going to buy and when so they can stock up.

The publisher usually buys content from a studio based on what people demand. Studio concepts for recording and motion picture industries diverge here - a movie studio often has in-house talent, pays everyone before expenses, etc. A recording industry usually hires talent by contract, pays themselves before expenses, etc (in exchange for a reciprocal license - the artist is allowed to play their music and sell related merchandise at their shows).

The reason I make that distinction is in-game ads would be put there to benefit the studio, not the publisher. In the case of games, however, often publishers own studios so indirectly they could be claiming a take or even forcing the marketing. In game ads are a lot like product placement on film sets - the studio gets paid for that, not the publisher.

Maybe

[Hatta]

I wouldn't be surprised if most game devs wouldn't implement PhysX if not for a subsidy. Only half the market is going to be able to take advantage of it after all. It may not be that they don't want it, just that it's not an economical use of their time otherwise.

Re:Maybe

[Vorknkx]

Exactly. Havok and in-house physics engine are perfectly fine for physics simulations in games. I don't see why we need another third-party physics engine. Flying boxes and wood splinters do not make a better game.

Re:Maybe

[hedwards]

If you noticed in the summary, AMD is advocating for a similar technology that works on their hardware as well as on nVidia's, seems like developers would prefer that for practical reasons.

Re:Maybe

[Monkeedude1212]

Flying boxes and wood splinters do not make a better game.

Well - it's the little things that make the differences though. I mean, you wouldn't think that flying boxes and wood splinters don't make a game any more amazing, but those were basically THE core elements of the Force Unleashed, using the Havok engine. Not surprisingly though, Havok was strictly licensed to Lucasarts for all of 2009 - no one else could use it. It's only just recently become available. So - for most of 2009, PhysX was the best choice - not only subsidized for using it, but because its competitors weren't actually available.

Re:Maybe

[Hatta]

Is PhysX just an API or is there hardware underneath supporting it? If it's hardware, then I'd say PhysX would be the better option practically. i.e. turning PhysX on would essentially be free in terms of resource usage. If it's just software then turning it on would take resources away from the rest of the rendering.

In any case, I've played a few games with PhysX. It's pretty fucking cool. Not cool enough to make a shitty game worth playing, but it makes a good game that much better.

Re:Maybe

[blahplusplus]

"Flying boxes and wood splinters do not make a better game."

But dead guys laying 180 perpendicular off a cliff makes them awesome? Does no one here remember the good old days of early FPS where if you died on the edge of a ledge your body would lay flat over the edge? Does no one remember the time when you hit dead bodies with shots and they didn't move or flail around? What about mass effect 1 the anti-gravity at the end with the geth/dead bodies floating and flailing around, not cool at all?

All that is physics and yes the do make a better game WHEN they are applied to things that need them and not over-used, especially not using physics as a gimmick.

Re:Maybe

[Improv]

I thought Havok has been used by SecondLife for years....

Re:Maybe

[EvilMonkeySlayer]

No, you're thinking of the Furry engine. It offers ultra realistic fur simulation, I hear it's quite popular with the Second Life crowd.

Re:Maybe

[soupd]

Maybe there is some subtlety I'm missing by Havok is part of every PS3 and Xbox 360 SDK and plenty of Havok games were released on PS3 last year, check out the list from Havok themselves: http://www.havok.com/content/view/584/96

Re:Maybe

[Aladrin]

Everyone seems to be glossing over a nice little fact:

Physx works on -all- modern Windows computers, whether they have a graphics accelerator or not. So yes, only have the market can use the hardware accelerated Physx, but the other half isn't barred from the game. They get to play, too.

Re:Maybe

[MobyDisk]

Open standards always win out over closed standards. Like OpenGL -vs- DirectX.... oh... wait... :-P

Re:Maybe

[Sir_Lewk]

Not everyone includes "pretty" in their "good game" equation. Doom can still hold it's own against modern games in terms of actual fun.

Re:Maybe

[jellomizer]

Well most devs would want to use their version because of pride in their work.

Re:Maybe

[DeadboltX]

It's only just recently become available.

I haven't been able to confirm your statement about Havok being exclusive to Lucasarts for 2009, but the Havok engine has been around since 2000 and has been used by over 100 different games, so it is by no means just recently available.

Re:Maybe

[Anonymous Coward]

Nvidia's stuff deliberately doesn't work on non-Nvidia hardware.

Swapping it so that it's something cross-platform that works equally for both will show people just how much Nvidia's Physx was a bunch of crap. Basically, you will see no Nvidia performance advantage in physx games as you see right now. This is not a new issue, but AMD's approach is.

Now, Nvidia will have no excuse as to why people should support PhysX.

However, Nvidia has moved on already. They're doing that bullshit 3d stuff - OMG 3d TV! Half the performance due to twice the refresh rate! must buy!

etc.

Re:Maybe

[VGPowerlord]

Not surprisingly though, Havok was strictly licensed to Lucasarts for all of 2009 - no one else could use it. It's only just recently become available. So - for most of 2009, PhysX was the best choice - not only subsidized for using it, but because its competitors weren't actually available.

I'm pretty sure the Source engine (HL2 and its derivitives) by Valve uses Havok. In fact, Havok is mentioned by name on the "Powered by Source" screen, near the bottom.

Re:Maybe

[H0p313ss]

Does no one remember the time when you hit dead bodies with shots and they didn't move or flail around?

Not everyone includes "pretty" in their "good game" equation. Doom can still hold it's own against modern games in terms of actual fun.

Clearly you don't get a kick out of shooting dead bodies and seeing them twitch.

What the hell's wrong with you?

Re:Maybe

[blahplusplus]

"Not everyone includes "pretty" in their "good game" equation"

No doubt but most people move on if a game has better graphics, otherwise you would still be playing wolf-3d, why did you move to doom? Oh yes, that's right it that darn graphics thing adds atmosphere and awesomeness to games. Why did people move on from super mario 1 for the NES, etc, etc? Not all games that come after are as fun but if you can get the same old fun in a shiny new package people will play it over the original.

Ever played the original Mechwarrior 2 dos games? No textures, all wireframe flat shaded practically, then 3D accelerators happened and every early 3D accelerator had a copy of Mechwarrior 2 with REAL textures that made the game look so damn awesome and took Mechwarrior 2 to a whole new level.

Re:Maybe

[Anonymous Coward]

Can PhysX even be used for non-gimicky stuff?

I say this seriously. Every implementation I've ever seen has been particle effects when you blow up trashcans, or papers flying around, or shrapnel littering the floor. Nothing is exactly repeatable and none of it is essential. You won't ever be unable to complete a mission because your trash didn't explode in the right direction.

This might be partially due to the fact that PhysX is a proprietary solution so anything that's important to gameplay is done in a more universal means but they could just use PhysX in cpu for gameplay elements, such as what halflife2 did.

Re:Maybe

[Opportunist]

Ya know, when you've seen it happen, it looks so fake in a game, no matter how good the engine...

Re:Maybe

[ASBands]

I've done some work with both PhysX and the things that AMD is pushing for. I try to keep with the Physics Abstraction Layer, which lets me plug in whatever physics engine as the backend, which gives a pretty damn good apples-to-apples performance metric. Personally, my ultimate choice of physics engine is the one which exhibits the best performance. My experience may differ from others, but I generally get the best performance from PhysX on with an nVidia GPU and BulletPhysics with an AMD GPU. Sometimes, the software version of PhysX outstrips the competition, but I have never seen anything beat PhysX in performance with GPU acceleration turned on. And with PAL, it is easy to check if there is GPU support on the machine and swap in the physics engine with the best performance (PAL is awesome).

Here's the thing: GPU-accelerated physics are just plain faster. Why? Because collision detection is a highly parallelizable problem. Guess what hardware we have that can help? The GPU. Another great part of using the GPU is that it frees the CPU to do more random crap (like AI or parsing the horribly slow scripting language).

AMD is working on both BulletPhysics and Havok so they can do GPU acceleration. But I have a feeling that PhysX performance will remain faster for a while: PhysX was designed to natively run on the GPU (technically, a GPU-like device), while these other libraries are not. Furthermore, nVidia has quite a head start in performance tuning, optimization and simple experience. In five years, that shouldn't matter, but I'm just saying that it will take a while.

So here is my message to AMD: If you want people to use your stuff, make something that works and let me test it out in my applications. You've released a demo of Havok with GPU acceleration. PhysX has been and continues to work with GPU acceleration on nVidia GPUs and will frequently outperform the software implementation. I'm all for open alternatives, but in this case, the open alternatives aren't good enough.

Re:Maybe

[FatSean]

Newer games offer more colors and higher resolution, nothing wrong with that. I'm sure the latest Gods of War would be just as fun in 8-bit 320x200.

Re:Maybe

[Sir_Lewk]

No doubt but most people move on if a game has better graphics, otherwise you would still be playing wolf-3d, why did you move to doom? Oh yes, that's right it that darn graphics thing adds atmosphere and awesomeness to games.

You are assuming it was because of the improved graphics, and not because of the improved gameplay mechanics and pacing. Polished turds are still turds and great games will always be great.

Re:Maybe

[Monkeedude1212]

Appears I was mistaken, the DMM by Pixelux and the Euphoria (for AI) was strictly to Lucas Arts, Havok is and was always available. My bad.

Re:Maybe

[Krau Ming]

Most games still have body parts of other characters appear through doors/walls when they're on the other side as close to the door/wall as possible. Is that part of the physics engine? Obviously it's better than the days of goldeneye on n64 when one could kill half an army from inside a closed room, but come on!!!

Re:Maybe

[blahplusplus]

"You are assuming it was because of the improved graphics, and not because of the improved gameplay mechanics and pacing. Polished turds are still turds and great games will always be great."

They will but sadly great gamers like ourselves are in the minority, assassin's creed 1 was like a poor man's prince of persia, but it sold millions. Sadly lots of people have junk gaming tastes, which means many aren't discerning enough to tell the difference between a mediocre game that missed the mark and a good game.

Re:Maybe

[Anonymous Coward]

Is it just me or did 4chan recently dump a big one on slashdot?

Re:Maybe

[blahplusplus]

You missed my point completely, go play mass effect 1 with the biotic powers, if you do not find the floaty physics of the 'anti gravity" animations of bodies floating around awesome then there is something wrong with you!

Physics is used as a special effect in games, sure games have clipping errors, most do. But these are HARD problems -i.e. they take time and money to develop and would not make for better games, games have limited budgets and they have to decide where to spend that money, people keep forgetting that.

Re:Maybe

[Anonymous Coward]

Euphoria has been around for a few years now -- the first triple A title to utilize it was GTA4.

Re:Maybe

[quantumplacet]

yes, what fools everyone is, they have no clue what they like at all. don't they know they're supposed to ask you first, and then only like the things that you like?

Re:Maybe

[jasko]

I think you're thinking of the Euphoria engine, a specialized physics engine for human(oid) bodies that's a big step over ragdoll. But even that's not exclusive to LucasArts. It's in GTA IV, too, at least.

Re:Maybe

[aztracker1]

But the hardware acceleration only works in nVidia products, which is what the point of the article is. Then again, I'm still waiting on OpenCL drivers for my Radeon HD 5770, but DirectCompute is there.

Re:Maybe

[Anonymous Coward]

Appears I was mistaken, the DMM by Pixelux and the Euphoria (for AI) was strictly to Lucas Arts, Havok is and was always available. My bad.

Don't worry. It's not the first time someone has been wrong on the internet.

Re:Maybe

[blahplusplus]

"yes, what fools everyone is, they have no clue what they like at all. don't they know they're supposed to ask you first, and then only like the things that you like?"

You're exactly what I'm talking about - first of all AC1 was a third person acrobatic combat game, i.e. cookie cutter and generic, nothing new except new art - that's what makes people buy the game "omg the art is so awesome and the AC1 assasin looks so badass got to buy this!". You could climb walls and jump on rooftops - you could do similar stuff in prince of persia running along walls, jumping on beams, climbing across roofs, etc. When you play enough games only then can you really understand what I'm talking about, your knee jerk "omg AC1 is s0 awesome!" response, is evidence of the correctness of my statements.

You act like the meta-game of these games is vastly difference when it is 1 to 1 with different coats of paint, AC1 had a lot of problems, 1 being the repetitive mini games, two being the boring travel between places. It wasn't a game an experienced gamer would not have seen before.

Re:Maybe

[DaleGlass]

I wish. All you can get right now is a flat texture. SL could really use a decent fur shader. And better support for quadrupeds and other types of avatars. And flexible tails that can move properly. That's just for a start.

SL still has a long way to go in that respect.

Re:Maybe

[PhxBlue]

What about mass effect 1 the anti-gravity at the end with the geth/dead bodies floating and flailing around, not cool at all?

I thought that was a feature. Gravity on the Presidium was 0.3G, and you're fighting on the outside of the station, so nailing a geth or krogan with high-powered biotics might be enough to accelerate it to escape velocity.

Re:Maybe

[osu-neko]

Exactly. Havok and in-house physics engine are perfectly fine for physics simulations in games. I don't see why we need another third-party physics engine. Flying boxes and wood splinters do not make a better game.

640K ought to be enough for anyone.

Re:Maybe

[Anonymous Coward]

Actually it is a quite accurate description of my experience in Second Life. All of those places to explore, but you never run into very many people and most of the ones you do see are furries. It seems to lend itself to the nature of the "game". It's a visual chatroom environment that you can't really do much with except look around and buy art of questionable quality.

Re:Maybe

[Anonymous Coward]

Everyone seems to be glossing over a nice little fact:

Physx works on -all- modern Windows computers, whether they have a graphics accelerator or not. So yes, only have the market can use the hardware accelerated Physx, but the other half isn't barred from the game. They get to play, too.

The issue is that when the PhysX API requires hardware (as in GPU) acceleration, it simply will not work on an ATI/AMD video card. Only nVidia GeForce 8xxx series and higher cards support PhysX Hardware acceleration.

Any game that has come out recently (Batman Arkham Asylum comes to mind) that touts PhysX as a feature requires GPU acceleration, and this locks that specific feature to nVidia only users.

AMD/ATI's approach is more likely to gain support, as OpenCL is, by nature, an open API like OpenGL is. Makes far more sense for a developer to support that as they can ensure that their features are hitting the widest demographic.

Re:Maybe

[quantumplacet]

You act like the meta-game of these games is vastly difference when it is 1 to 1 with different coats of paint, AC1 had a lot of problems, 1 being the repetitive mini games, two being the boring travel between places. It wasn't a game an experienced gamer would not have seen before.

I'm not acting like anything. I own AC1 and I think it fucking sucks. Every 3 or 4 months I put it back in to try to finish it just for the sake of finishing it, but I can't because it sucks. However, the fact that I don't like it doesn't mean that no one else is allowed to. Also, the recent Prince of Persia game (which was based on the AC engine), came out AFTER AC1, and also sucked. You'll notice that I just said that it sucked, and didn't make a poor attempt to belittle you because you didn't think it sucked, because you are allowed to like it, even though it did in fact suck.

Re:Maybe

[martin-boundary]

Can PhysX even be used for non-gimicky stuff?

I don't play recent games, but here are two games that use physics for non-gimicky stuff, and would therefore qualify:

1) Scorched Earth.

2) Thrust.

3) Now git off ma 'lawn.

Re:Maybe

[makomk]

If it's hardware, then I'd say PhysX would be the better option practically. i.e. turning PhysX on would essentially be free in terms of resource usage. If it's just software then turning it on would take resources away from the rest of the rendering.

PhysX runs using the same GPU hardware resources that games also use - there's no PhysX-specific hardware on NVidia cards, so it does take away resources from rendering the game.

Re:Maybe

[makomk]

Yeah, just so long as you don't mind either the FPS slowing down to single digits while it runs the physics using badly-optimised single core code, or disabling most of it.

Re:Maybe

[blahplusplus]

"I thought that was a feature."

Did you actually LOOK at what was happening to the bodies and how they were animating? Go reload the level outside the station and take a good look at what the bodies are doing as they float along.

Re:Maybe

[Stevecrox]

Your missing a something and thats cost. Uru: Ages Beyond Myst was developed by Cyan Worlds back in 2004 and they used the Havok Physics engine. Several years later Gametap approached Cyan and gave them money to remake the Uru:Live (online MMO) part of Uru: Ages Beyond Myst. Cyan had to ditch Havok because it cost an arm and a leg to license it, as a *Beta tester I can tell you most of the 1 year redevelopment was spent on ripping out Havok and putting Physx in (most of the other enhancements were done during development of Myst V).

Physx is cheap, it does 99% of what people are interested in and does it quite well. When your a small development house, price can mean a lot.

*My 2 year NDA agreement expired last year

Re:Maybe

[fast turtle]

The problem is, Nvidia bought the company that created the PhysX chip. The then incorporated the capabilities into their GPU and their drivers while refusing to allow a true PhysX card to work with anything except an Nvidia GPU in the system. In this case, Nvidia has done a major Dis-service to everyone as the original PhysX did run on anything that supported it but now it's a closed sorce, Nvidia only app/feature.

Re:Maybe

[Anonymous Coward]

It's a combination of the two - hardware based physics rendering if you have an Nvidia GPU above the 8000 series (i think). Basically, it uses the GPU to do all the physics calculations in parallel if the GPU supports it - GPU's are massively parallel processors anyway, so it works really well without eating too many of the GPU's cycles.

If a GPU with support isn't available there's a software backup available, but not all games use it. Mirrors Edge, for example, disabled the PhysX stuff if it couldn't run on hardware.

Re:Maybe

[Anonymous Coward]

Technically, PhysX was designed to run on the PhysX standalone card (somewhat similar to modern graphic cards in design), which maybe even isn't supported anymore. Though graphic card is more logical place to have it - because it cuts down back and forth copying of geometry data from CPU to PPU, back to CPU and then to GPU.

Re:Maybe

[Anonymous Coward]

And after all those great graphics came in, everyone stopped playing mw2, cuz all the crappy, special graphics accelerated versions added lag.

Nothing, NOTHING was as good as dropping with your team and blasting some AC20's or doing an LRM "bloom" at the enemy.

Nothing.

Re:Maybe

[Anonymous Coward]

But dead guys laying 180 perpendicular off a cliff

Geometry parser fail (conflicting angle specification) Perhaps you need a better physics library?

Hint: perpendicular != 180 degrees.

Re:Maybe

[Gothmolly]

Yet you still play FPS games, so clearly those artifacts were not a detriment to the enjoyment of the game.

Re:Maybe

[Jthon]

PhysX can also run on the CPU (and PS3, Wii, and Xbox360) but can't handle the same amount of workload in that case since the CPU doesn't have as much raw compute power as the GPU.

The thing which irritates people is that it doesn't start to get really cool until you crank up the object density to the point the CPU can't handle it anymore. At that point you need a GPU or PhysX card, and right now that means NVIDIA only. (Note that I've heard NVIDIA would license PhysX to AMD if they want to code their own back end, but AMD has no interest.)

Re:Maybe

[BikeHelmet]

Only half the market is going to be able to take advantage of it after all.

1) PhysX runs on the CPU if no nVidia GPU is present. A $100 quad-core CPU easily handles it for most games.
2) According to the Steam Survey, nVidia is approximately 66% of the PC gaming market. Two thirds.

Re:Maybe

[JernejL]

Except that PAL - Physics Abstraction Layer contains some flawed and bugged support for half the physics engines it supports. And even those only support ancient old versions, the pal benchmarks are pretty much leaning towards bullet, you can see for yourself which engine support is given most work there. If you cant see that pal favorizes bullet on purpose you are blind.

Re:Maybe

[PhxBlue]

Well, they weren't doing this, so I didn't really pay too much attention ...

Re:Maybe

[sixsixtysix]

he is obviously not judging games based solely on looks.
are you saying that, for example, serious sam hd's updated graphics in no way improve over the original, that they shouldn't affect the fun factor in any way?
if your precious doom was redone exactly the same, save updated graphics, it would only be equally as fun?
imo: (fun game + good graphics) > (fun game + not-as-good graphics)

Re:Maybe

[KillShill]

There are discussions and evidence that Nvidia disables multi-core PhysX processing , most likely to make its hardware-based implementation look better.

http://www.google.com/search?q=nvidia+disable+multi-core+physx
http://www.bing.com/search?q=nvidia+multi-core+physx
http://search.yahoo.com/search?p=nvidia+disable+multi-core+physx

The "official" excuse is that it's "up to the developers to enable it"... but if Nvidia is paying them... then it only makes sense they wouldn't allow multi-core software implementations as that would make their hw less desirable.

PhysX is the GLide of 21st century, except it offers much less and performs poorly (even on high end hardware).

Open standards are the way to go.

Re:Maybe

[Anonymous Coward]

Havok and in-house physics engine are perfectly fine for physics simulations in games.

640k should be enough for anyone.

We don't know what could be accomplished by developers with good physics technology.
Physics in games is still TERRIBLE. Even "simple" things like ragdoll physics are completely unrealistic.

I can't wait to see realistic water, vehicle and, explosion simulation in-game.
And I'm really excited to think about how some hot-shot developer can make a revolutionary game with it.
Like what DOOM, Quake and Portal did for 3D games, I want to see what someone can do with real physics.

Hell, that's just for games, what other applications could benefit from commodity hardware able to perform realistic physics calculations in real-time?

Re:Maybe

[Hurricane78]

Doom? Doom was stupid it was a total joke compared to System Shock (released at the same time). It only got the publicity.

Same as Half-life 2 and Doom 3 got the publicity, while Far Cry and Return to Butcher Bay were the much greater games.

If you want to see a great ID game, look at Quake 3 Arena. It’s still played. Over 10 years after it was released. With CPMA and Defrag, it’s still challenging and modern. Plus there are new engines for it in development, that get close to Crysis looks! (xreal!) Best. shooter. ever.

Re:Maybe

[BikeHelmet]

I'm more inclined to believe it to be incompetence rather than malice.

When multi-core rendering was added to the nVidia drivers, it was unstable on most games. Crashes, etc. unless disabled.

I suspect a similar thing is going on here.

But I fully acknowledge that nVidia has marketing deals with game developers. I just can't say what they actually cover. I doubt there's an actual conspiracy, based on their past prowess (or lack thereof) when it mattered with multi-core rendering and SLI. (multi-GPU)

Re:Maybe

[Anonymous Coward]

I blame the ARB and Khronos for that. OpenGL used to be cutting edge and DirectX was the one playing catch up.

DirectX used to really suck until around DX6 where the 2D acceleration interfaces got good enough to be programmer friendly (Using OGL for 2D is possible but not everyone had 3D accelerators in those days which would have made performance suck for most people). DX7 had reasonable 3D support, people ported their 2D engines to use Direct3D as an addon but were generally still 2D at the core [simpler migration path]. The problem arrived with DX8 -> MS added hardware accelerated vertex and pixel shaders, that came out of left field and allowed DX to overtake OpenGL on features; OGL has been playing catch up ever since.

OpenGL3 was supposed to be a complete redesign to make OGL object oriented like DirectX (but better since it isn't as in-your-face low level) but that got kicked in the nuts at some point and so we're stuck with the retarded 90s era State Engine system.

It also doesn't help that ATi, the kings of shitty drivers, have worse OpenGL performance than DirectX on Windows (A program that can use both DX and OGL will run slower using the OGL API). Intel's (as though it is worth mentioning...) is flaky but their chips suck on both APIs so it doesn't make much difference. Only nVidia's OGL support is about as good as their DirectX support.

[Of course, there was also the usual fun anti-trust stuff from Microsoft where they deliberately redesigned the OpenGL bindings on Windows so that OpenGL drivers were more complex and ran slower but that isn't new or surprising]

Re:Maybe

[Carewolf]

But dead guys laying 180 perpendicular off a cliff makes them awesome? Does no one here remember the good old days of early FPS where if you died on the edge of a ledge your body would lay flat over the edge? Does no one remember the time when you hit dead bodies with shots and they didn't move or flail around? What about mass effect 1 the anti-gravity at the end with the geth/dead bodies floating and flailing around, not cool at all?

All that is physics and yes the do make a better game WHEN they are applied to things that need them and not over-used, especially not using physics as a gimmick.

No, NONE of that is physics. It is all fiction. What physical happens when a bullet hits a body, is that it goes through, or just inside and does purely internal damage, but it doesn't affect the body as a whole except for making bullet-holes. So what you have here is a acceleration of non-physics...

Re:Maybe

[Anonymous Coward]

You forgot one important point - bullet physics is shit. One of Irwin Couman's benchmarks is a pyramid of blocks which PhysX can actually simulate - the blocks fall then stay in place. In bullet physics they slide off each other, totally contrary to the laws of physics. Havok is moderately better, but the friction models in both systems are just wrong. Also, PhysX handles up to 128 contact points between each object while Havok and bullet manage only 4. Just stacking a pair of cubes needs 8 contact points in the general case, so you can see the problem.

Bullet was written for speed, for simple physical reactions (not real "physics" as such), but since upping the ante, PhysX have gone and optimized their solution a great deal more, and it's now as fast as Havok and almost as fast as Bullet for physical reaction stuff, while being accurate and stable for more complex uses. (This is on the CPU and Cell SPU; you're saying PhysX is faster anyway on GPU, which is very interesting, although not many triple-A games have spare GPU to spend on physics.)

I don't know what AMD are bringing to the table. Bullet physics is free anyway, isn't it? As is PhysX, if you only want libs and not full source code.

Re:Maybe

[Anonymous Coward]

No....not 66% of the PC gaming market. 66% of the Steam market. I, for one, refuse to install Steam on my computer. If the game needs Steam, I don't need the game. Having a downloadable version of the game is great and I'm glad that the option is available to me should I need it, but it's not the option I choose (at least not in the form Steam games are presented; feel like giving me an ISO and I have no problem buying your downloadable game). My point being that perhaps the type of person who purchases an NVidia card (or an OEM computer with an NVidia card) is the same type of person who would get their games from Steam. When you can point me to a study from an independent group showing statistics (and making the raw data and study methodology available, anything less is just posturing), then we'll talk Until then, your statistics are about as useful as my metric that shows that 100% of the direct parents of this post are morons, NVidia owners, and put way too much faith in statistics.

Re:Maybe

[makomk]

PhysX can also run on the CPU (and PS3, Wii, and Xbox360) but can't handle the same amount of workload in that case since the CPU doesn't have as much raw compute power as the GPU.

Actually, IIRC it's not too far off if we're talking lowish-end GPU hardware and a mid-to-high end quad core - remember that few people use high-end cards for PhysX acceleration. The actual problem is that NVidia's CPU fallback is extremely unoptimised and can't make use of more than one core. (In fact, there are plausible rumours that PhysX uses slower code even for the bits that always run on the CPU if you don't have GPU acceleration enabled than if you do.)

Re:Maybe

[Anonymous Coward]

" I own AC1..."

That's all that needs to be said in your case, people buying games on hype thereby allowing devs and publishers to rape you of your money for an unfinished game. You probably finished that sucky game too right? It was mediocre but you played to the end, thats how you know it sucks. Yet your poor taste made you buy it to begin with.

Re:Maybe

[Anonymous Coward]

Except nVidia artificially limits the number of CPU cores that they allow to run PhysX. Got a Quad Core CPU? Too bad, PhysX won't use it. Often, while the main gameplay will run on both software and hardware, they add extra eyecandy that only works on nVidia hardware even if it would run just fine if you had enough fast CPU cores.

Is anyone

[Anonymous Coward]

surprised?

It's a new riff on the old joke

[idontgno]

"You're so ugly the only way to get the dog to play with you is to tie a steak around your neck."

Says the kid the dog without a dog to play with.

Re:It's a new riff on the old joke

[idontgno]

Wow, that's so badly edited it's surreal.

This is one of those days where even the "Preview" button doesn't help.

That should read "Says the kid that the dog isn't playing with."

Re:It's a new riff on the old joke

[Anarki2004]

"You're so ugly the only way to get the dog to play with you is to tie a steak around your neck."

Says the kid the dog without a dog to play with.

Try again please. That statement is a grammatical failure. I'm not even sure what you were trying to say.

Re:It's a new riff on the old joke

[Minwee]

Try again please. That statement is a grammatical failure. I'm not even sure what you were trying to say.

If you use Google Translate to translate it back into Korean, then Portuguese, Russian, Welsh and then finally back into English everything becomes much clearer.

"Children play with a puppy."

Re:It's a new riff on the old joke

[MartinSchou]

That should read "Says the kid that the dog isn't playing with."With whom the dog is not playing.

Re:It's a new riff on the old joke

[Anonymous Coward]

I got: "Say hello to your son without a dog"

Re:It's a new riff on the old joke

[Anonymous Coward]

Funny thing... it's perfectly acceptable to end a clause or sentence with a preposition in English, and always has been. Feel free to examine the works of any published author ever.

Re:It's a new riff on the old joke

[mestar]

For some reason, I got this picture:

http://craphound.com/images/translateservererror.jpg

Re:It's a new riff on the old joke

[Rob the Bold]

That should read "Says the kid that the dog isn't playing with."With whom the dog is not playing.

Yes, that's something up with which I cannot put.

Re:It's a new riff on the old joke

[dbcad7]

I must be defective. Even with his mistake, I got the gist of what he was trying to say... Life may make more sense to you at this web site.. http://forums.about.com/n/pfx/forum.aspx?webtag=ab-grammar

Re:It's a new riff on the old joke

[osu-neko]

I must be defective. Even with his mistake, I got the gist of what he was trying to say... Life may make more sense to you at this web site.. http://forums.about.com/n/pfx/forum.aspx?webtag=ab-grammar

I'm even more defective -- not only did I understand what he was trying to say perfectly, but if replies hadn't called attention to it, I would never have noticed there was any error in the statement to begin with.

Re:It's a new riff on the old joke

[Endo13]

I've seen it before. As a result, I read over it so quickly that not only did I get the gist of it, I didn't even notice he'd butchered it until I read his next post. I'm not sure which is worse, his typing or my reading.

Re:It's a new riff on the old joke

[MartinSchou]

The issue wasn't about ending the sentence with the word "with", it was with not using "whom".

What does PhysX do anyways?

[Anonymous Coward]

I've never figured out what PhysX is supposed to do. More realistic physics I suppose? Well I can't say I've ever noticed any difference between a game that uses it and a game that doesn't. So, what, the corpses flop differently?

Re:What does PhysX do anyways?

[binarylarry]

duh, it's got what gamers crave!

Re:What does PhysX do anyways?

[Pojut]

Ask, and ye shall receive: http://en.wikipedia.org/wiki/PhysX#PPU

Re:What does PhysX do anyways?

[Pojut]

Electrolytes?

Re:What does PhysX do anyways?

[religious freak]

http://lmgtfy.com/?q=PhysX

I [heart] this site... makes me happy every time I provide a link :)

Re:What does PhysX do anyways?

[postmortem]

The ones with PhysX are more likey to crash because nvidia drivers aren't prefect.

Not a trolling attempt: I have played Mirror's Edge game where flag vawes realistically on PhysX, but it crashes on same spot always. Just before crash flag looks weird.

On Radeon, flag doesn't wave so naturally, but game does not crash either.

Re:What does PhysX do anyways?

[Anonymous Coward]

duh, it's got what gamers crave!

Electrobytes?

Re:What does PhysX do anyways?

[Jeng]

I think it makes you look like a dipshit everytime you provide a link.

You do not provide any sort of answer, and you assume that the person did not already google it.

So lets assume that the person did google this, and did not find an answer that helped him understand the issue. Wouldn't a good next step be to ask others for help? Is Slashdot not a good place to ask the question?

Re:What does PhysX do anyways?

[Spatial]

It can be hardware accelerated on the GPU. That's it.

The benefit: Physics is one of those easily parallelised problems so a very large increase in complexity is possible.

The drawbacks: There's less GPU time available for drawing stuff so your framerate suffers. And of course, it's limited to Nvidia hardware only.

The latter leads to a drawback of its own: the technology can't be used to its full potential because many people who buy a game won't have the necessary hardware. So it can't be used in ways that would affect gameplay.

Re:What does PhysX do anyways?

[david_thornley]

Girlfriends?

Re:What does PhysX do anyways?

[Anonymous Coward]

When a two year old or even a five year old asks you "Why?", it makes sense to explain the reason to them because they're still learning basic skills. When someone older than 13 asks you why and the answer is simple enough to be within their grasp (as opposed to being based on quantum field theory), it makes more sense to tell them "Look it up!". There probably shouldn't be too many 5 year olds on /. So giving someone a lmgtfy link is basically telling them, "You're older than 5 years and this is now a basic skill." If someone has enough interest to ask the question, then it will be faster for them to look it up than waste someone else's time. Also see this link.

You do not provide any sort of answer, and you assume that the person did not already google it.

If they had googled it, their question would show that they had at least read the first article (or even the google excerpt) returned by the search. They would ask for some sort of clarification if they hadn't understood. Expecting someone else to digest it for you is an intellectual laziness that is at the root of many ills in modern society. The real dipshits are the ones who expect the world to cater to their every minor whim.

Re:What does PhysX do anyways?

[Fastball]

White Castle

Re:What does PhysX do anyways?

[Jeng]

Excellent reply.

Slashdot brings in geeks from all realms. Some may be smart in one area, but clueless in others with no real knowledge of a subject. Even to the point of "Hey, anyone mind helping me understand this issue?".

I should have re-read the original post on this thread, the original poster was an idiot, but even so telling him to google it would do nothing, especially in this case since he has already used PhysX in games and was mainly stating that he saw no point.

My reply was mainly a knee-jerk against the lmgtfy posts, didn't really fit this thread though.

Re:What does PhysX do anyways?

[Nyder]

duh, it's got what gamers crave!

We crave no DRM also, but I don't see that happening that much...

What else is new

[dvlhrns]

How is this any different then what Microsoft does ?

Sour grapes?

[cbope]

Sounds to me like AMD is just taking pot shots at NVIDIA. They probably wish they had either invented it or bought up Ageia before NVIDIA. Are there any games out that use OpenCL for physics? Or DirectCompute?

Trust me, NVIDIA will flog the Physx horse as long as it can. Eventually something will replace it anyway, so who gives a shit. Apart from software-based CPU physics, I haven't seen too many Physx titles and nothing for OpenCL or DC yet. I do have a 9600GT dedicated for Physx in my gaming rig, for those few titles that support it.

Re:Sour grapes?

[jandrese]

Are you seeing any benefit from that 9600GT? My complaint about physics in games is that because not everybody has the hareware for it, the developers have kept the number of physics objects down to a bare minimum (a create here, a rock there, sometimes a ragdoll deadguy), so you really don't get a feeling like there is physics in the world, just sometimes bodies will flop around and boxes will bounce a bit. Because of this, dedicated physics hardware goes to waste almost all of the time.

It would be cool to have a game where nearly everthing was physics enabled, and you just spent most of your time applying forces to objects to get stuff done--Oh' no! a badguy! I'll stop time for a second, apply a push to a nearby door, and cause the door to slam open in his face. Or maybe a clock will fly off of the wall and bonk him in the head. Maybe I'll knock a bookshelf over instead. And not in the pre-scripted "Hmm, I just got my gravity gun, and oh my, apparently a woodshop exploded over the town and littered the place with sawblades..."

Re:Sour grapes?

[Jeng]

Actually this sounds fairly familiar, there are strong parallels between this and AMD's issue with Intel.

Nvidia is using their marketshare to push forward their software that can only run on their cards by paying companies to use it. If the developers are using Nvidia's solution then they are not using the competitors.

Re:Sour grapes?

[dvlhrns]

Exactly what M$ does...use their marketshare to push their software that can only run on their PCs, etc, etc, etc...

Re:Sour grapes?

[Jeng]

If Microsoft made computers you would have a point.

I guess your point may have been that Microsoft only made software that works with their operating system? If so, that would also have been slightly incorrect in that Microsoft does put out a version of Office for Macs. They don't do much beyond that for other operating systems so you would have been close.

Re:Sour grapes?

[dvlhrns]

Well, more to the point of them using their market share to influence people and companies, but sure

Re:Sour grapes?

[Jeng]

It is the way its being done that is questionable, not that it is being done.

Use of market share as a marketing tool is completely legal and common as dirt.

Re:Sour grapes?

[Anonymous Coward]

There is such a game, where the developers are dedicated to everything being physically correct. If you design something that looks like a car it'll behave like a car, the same goes if its an airplane... etc. The game/physics engine is open source and is called Rigs of Rods:

http://repository.rigsofrods.com/

Is it actually allowed to also BE better?

[DarkkOne]

Even before hardware accelerated PhysX was on CUDA and you only got it with the standalone card, I always thought PhysX looked a bet nicer than Havok in action. I've been wishing more games used PhysX for a while, but it seems that if a game is going to be cross-target to the consoles as well, Havok is just a lot more likely. It may just be my own perceptions, but things seem to have a bit more consistent behaviour in regard to momentum and mass in PhysX whereas Havok seems a bit "floaty" a lot of the time. This may just be a result of constants designers pick, or something, I don't really know the details. But I personally just like PhysX better, from a player standpoint, hardware accelerated or not.

come on, AMD...

[toastliscio]

A few months ago I bought a new pc after years, and it has an AMD cpu but an nVidia gpu (it's assembled by myself). For hardware compatibility reasons it would appear obvious to buy an AMD/ATI gpu, but the problem is, I use Linux. And AMD graphic drivers on linux still suck compared to nVidia's. Why don't they shut up and strive to make decent drivers? They would get new customers, including me.

Re:come on, AMD...

[Anonymous Coward]

because, yet again, catering to the linux desktop community would be like gmc catering to the double amputee community. there are next to no users and the few users that are there seem to revel in running their machine off of 80 dollars worth of parts that they only upgrade after every other president.

sorry guy but the linux community is notoriously cheap and when you have a niche market like that you need for them to be big spenders to make it worth while. how else do you think that bentley and rolls royce get away with making less cars than most manufacturers use in crash tests in a year but still maintain a profitable business?

Re:come on, AMD...

[Anonymous Coward]

...only upgrade after every other president.

I'm shopping for an AGP graphics card to go with some new RAM I just put into a dual Opteron machine I built just after the start of the Iraq war on Fedora Core 1. That's 12 Fedora releases, 1.5 Bush terms and a third of Obama.

It was a pretty sweet machine six-seven years ago, and I really splurged on it. Now it's pretty much exactly as you describe, and likely to continue serving me and my family through the current Obama administration.

Re:come on, AMD...

[osu-neko]

because, yet again, catering to the linux desktop community would be like gmc catering to the double amputee community. there are next to no users and the few users that are there seem to revel in running their machine off of 80 dollars worth of parts that they only upgrade after every other president.

Hey! I upgrade at least once every president! (...or at least I did for the last two, who both served two full terms.)

Re:come on, AMD...

[makomk]

I'm currently using a Radeon 2600 HD under Linux. With open source drivers, in fact. It's good enough for what I want to do (though the 3D support isn't as good as the closed source driver) and I don't have to deal with the NVidia binary blob and its bugs. (Seriously, they managed to break 7300GS support in a minor patch on several occasions, then took ages to fix the regression. Plus, I'm not sure that KDE works quite right on NVidia hardware even now.)

Re:come on, AMD...

[Jesus_666]

there are next to no users and the few users that are there seem to revel in running their machine off of 80 dollars worth of parts that they only upgrade after every other president.

The obvious solution would be to ensure that no president survives more than half a year in office.

Is it true?

[Improv]

It seems a lot of people are kvetching at AMD for this because they're criticising a competitor. I think it's really more relevant to consider if what AMD says is true - if nVidia is paying people to use their proprietary stuff and then claiming it has broad industry adoption (and therefore is good), that's pretty shady.

I'm not sure how we really can tell if the criticism is valid unless we're in the industry though.

clutching at straws

[obarthelemy]

GPU makers are in a bind:
- IGP are now enough for 90% of users: office work (even w/ Aero), video, light gaming, dual-screen... all work fine with IGPs
- the remaining 10% (gamers, graphic artists) are dwindling for lack or outstanding games: game publishers are turned off by rampant piracy, mainly online games bring in big money nowadays
- GPGPU is useless except in scientific computing: we already have more x86 cores than the devs know how to use, let alone use a different computing paradigm
- devs have to target the lowest common denominator, which means no GPGPU for games

I'm actually think of moving my home PC to one of the upcoming ARM-based smarttops. They look good enough for torrenting + video watching + web browsing, consume 10 watts instead of 150...

Re:clutching at straws

[Anonymous Coward]

I second that

Re:clutching at straws

[Ironhandx]

Tell that to AMD who have sold 2 million directx11 GPUs since release. (http://www.dailytech.com/ATI+Sells+Over+2+Million+DirectX+11+GPUs+Celebrates+With+Radeon+Cake/article17349.htm)

IGP are sufficient for 90% of users... but that hasn't changed since back in the Pentium 1 days. Many PCs were equipped with IGP or something that amounted to the same thing but in card form even then.

Also: GPGPU is NOT meant for gfx processing on the fly at all, so it has absolutely nothing to do with devs having to target the lowest common denominator. You even state that its useless except for scientific purposes in your own comment. The entire purpose of the GPGPU move is towards scientific purposes where vast quantities of repeated calcs have to be done. Something that GPUs excel at.

At least get SOME of your facts straight before spouting FUD.

Re:clutching at straws

[LWATCDR]

"GPGPU is useless except in scientific computing: we already have more x86 cores than the devs know how to use, let alone use a different computing paradigm"
Well maybe for games but GPGPU will mean a lot for transcoding.
Home HD video is going to be big soon and it takes forever to transcode. However you can do that with an ARM. The TegaII and the OMAP line have enough GPU power to use it for transcoding.

Re:clutching at straws

[obarthelemy]

There are about 25 million PCs sold per month. I guess ATI is happy to have sold 8% of that monthly amount over the several months their 5xxx have been available, that's 3-4% of PC sales. Congrats to them, but still, fairly marginal.

Discrete cards have always been better than IGPs. I don't really get your point. Only recently (definitely way after the pentium 1) have IGPs become good enough to display all video files, or handle Aero.

PhysX is about making physics computations, not directly putting pixels on screen, so it's a kind of specialized GPGPU.

Re:clutching at straws

[Anonymous Coward]

- the remaining 10% (gamers, graphic artists) are dwindling for lack or outstanding games: game publishers are turned off by rampant piracy, mainly online games bring in big money nowadays

Piracy, or bad games? Many games these days are just pretty graphics pretending to be quality gameplay -- I think I can count on one hand the number of games I've felt have had excellent gameplay in the last 5 years, enough to be worth paying full retail price for.

Also as far as the move to consoles, there's the standardization issues on PCs to be dealt with given any engine -- Borderlands couldn't run on certain machines (It had two separate hardware incompatabilities, one with the cpu one with graphics card) even if they met min specs. On consoles, that's not an issue. Also, the console market seems to be on the whole less demanding of quality products before they'll throw gobs of money at companies.

I think you're being too quick to blame pirates.

Re:clutching at straws

[Ironhandx]

The point is that its no more marginal than it always has been. Theres actually been an increase in market share for dedicated GPUs, and market studies are now seeing a large demand for discrete video cards for laptops even.

You sir are the one clutching at straws. Also: The recently for that would have been the P1, I had an old one, and while it definitely wouldn't do anything even remotely high res, but it would play most of the video files I needed it to... because standard resolution sizes were much lower back then anyways. The 800x600 desktop was even still relatively new back then. The IGP has really just evolved with the needs of the lowest common denominator consumer ever since it was first spawned. First one for most of the current(excluding HD) formats was probably the P3 generation of PCs.

Aero was actually a speed bump in the IGP road. Up until then IGP had been plenty of gfx rendering power for anyone not doing a lot of 3d gaming. I may be stretching your memory here but 1080p, even 720p video files have not even been the norm for all that long. There has been no "major advancement" in IGPs, in fact they're just catching back up recently to the utility they had before the advent of Vista and h.264 etc.

Sure the library of games you can play on them is larger, but that will happen regardless as all of the hardware gets more powerful and the games do not.

You make points that are not actually points, then also blame some sort of mythical lack of dev ability to cope with multiple CPU cores when devs actually have their hands tied by target audiences. Multi-threaded game engines have to be built from the ground up to BE multi threaded. This doesn't work well when you want your game to run on single core systems as well(See recent problems with Mass Effect 2 on single core systems for reference) The developers are just now moving away from programming for single core because the larger gaming community, which includes many of your casuals who won't upgrade for 5 years+ after buying a new PC, because their systems CAN actually still run the new games, if slowly, on low settings, are just now getting to the point where those guys are being forced to upgrade, and publishers, and developers both, don't want to jump the gun on cutting out potential markets.

Also: The Phsyx processor more closely resembles a CPU than a GPU, as that is where it is actually taking load from. The CPU used to handle the vast majority of physics calcs so Nvidia got the bright idea to lend a helping hand.

Re:clutching at straws

[obarthelemy]

I've got studies that say IGPs actually rose vs GPUs in 2009, I'm interested in your sources.

Indeed, the difference is between IGPs playing some videos, which they always did, and IGPs playing all videos, which they now do. Glad you got that.

Mmm, are you saying that IGPs, especially Intel's much dominant ones, have done hardware-assisted video decode for all formats for a long time ? That's just not true.

And finally, my point exactly: if devs can't do multi-core even now, when do you think they will do physx or GPGPU, which is even more marginal ? The reason is rather immaterial, the result is clear. In the end, difficulty = money anyway, choose which one you prefer.

Re:clutching at straws

[Ironhandx]

Who's ever needed hardware assisted decode? CPUS have been overpowered enough to handle this without breaking a sweat for ages now.... The only minor problems along the way are the same ones that have always been... some codecs etc relying on some random bit of hardware support that the IGPs don't have currently. Which by the way, has seen issues from way back then, right up to more recently, and the next video format will see it happen again, its just cyclical.

As for the IGPs gaining market share: Yes they did, if you include Laptop/Netbook sales as well. If you include only desktops, GPUs are seeing an increased market share, and if you take the more powerful IGPs that are made more for gaming in laptops and include them in the discrete GPU list(because technically, the higher end ATI/Nvidia ones mostly are, they just share memory) You also see an increase. I'm probably looking at the same studies you are, but I'm interpreting the data differently(I would say more precisely).

Also, Intels most IGPs STILL don't do hardware decode for most stuff. They tacked on h.264 a little while ago and thats it.

One of the trends over the course of 2009 is seeing a lot of people go for more portability in their gaming, which means laptops. That inevitably skews any results in favor of IGPs because thats really the only option. Even the dedicated GPU+Memory solutions are still technically IGP because they're fully integrated with the motherboard etc. Looking at it on the surface is a bad way to look at things. As is the case for many things, the surface is highly deceptive.

Re:clutching at straws

[dkf]

Multi-threaded game engines have to be built from the ground up to BE multi threaded. This doesn't work well when you want your game to run on single core systems as well(See recent problems with Mass Effect 2 on single core systems for reference)

That was a problem with the video renderer; everything else about ME2 worked just fine on single core. Moreover, ME2 is now fixed and works just fine (provided you've got enough memory and a good graphics card, of course, but they're easier to upgrade).

Re:clutching at straws

[mxs]

I wonder why you attribute the lack of outstanding games to piracy being rampant -- the industry has been bitching and moaning about that for over 20 years now. That can't be the reason or we would not have a videogame-industry at all.

Few game developers are willing to do risky things though, and countless remakes of the same games just don't really appeal to all that many gamers -- add to that that gaming itself is being transformed (or rather, the marketplace is changing with mobile games becoming a pastime of millions, there actually being a LOT of games out there from the years prior, etc.), and I can see some reasons for that. Add to that the fact that asset-development (ugh) for modern games can be magnitudes more expensive than for old "outstanding" games and you see that the financials have changed quite a bit as well -- you can no longer just produce an AAA title in a team of 2-4 people in a basement -- the tools simply have not caught up yet.

I would not count out GPGPU as a niche product just yet -- it's true, paradigms change, but the mere fact that x86 cores are becoming more plentiful is leading to more tool support, more heads thinking about the problems and solutions involved, and more people getting used to concurrent programming. Once you know how to use 6 or 8 cores well, it will not be too much of a jump to the 300-500 threads a GPU will handle. I hope for great things in this area though admittedly I personally like it for the sciency stuff :)

Lowest common denominator, not. Degrading gracefully to it, yes. Games traditionally push the envelope, and will continue to do so -- unless the console-model and mobile gaming overtakes the entire market.

ARM-based devices have their uses, but to be quite honest -- I like a spiffy desktop.

Re:clutching at straws

[Chris Burke]

There are about 25 million PCs sold per month. I guess ATI is happy to have sold 8% of that monthly amount over the several months their 5xxx have been available, that's 3-4% of PC sales. Congrats to them, but still, fairly marginal.

Oh I'm quite certain they are exceedingly happy, and I think you'll find that margin is exactly why. ;)

Discrete cards have always been better than IGPs. I don't really get your point. Only recently (definitely way after the pentium 1) have IGPs become good enough to display all video files, or handle Aero.

Aero didn't exist in Pentium 1 days, nor did anything consumer level that required OpenGL, nor could the highest-of-high-end consumer cards handled it anyway. Videos were generally small and low resolution, and only had to be decoded, not decrypted. People were doing different things with their computers.

The point is that IGP (or before integrated graphics, the bottom-of-the-line cards that came with budget computers) has been fine for most users for most of the history of the PC.

Re:clutching at straws

[Jarik_Tentsu]

It has nothing to do with what is 'enough'.

People always like what is out there. A $20,000 family car is more than enough to transport you from point A to point B. Yet hundreds of thousands of people buy cars worth over $150k boasting amazing lap times around Nurbugring, fast 0-100 time and all that...and the funny thing is, probably a tiny, tiny, tiny portion actually race their cars. Most just drive around normally.

Marketing means that when people go to buy something, the mere existence of something better lulls them to it. So while your 5 year old graphics card renders your web pages and email fine, when you go to buy a new graphics card, customers would be drawn away from the dull, cheap rigs to the more expensive ones. Then they'll remember some lagginess, some slowness in certain instances and justify the expense.

Hell, I'm a prime example of this. Unlike my mates, I rarely game. I used to video edit a lot, but not really more. But I spent huge money on my current computer, buying massive heatsinks, overclocking it to the max, buying top end parts, etc. Certainly didn't 'need' it. But I did 'want' it.

And as long as people and society are like this - GPU's are not in a bind. They just have to keep releasing something better and people will want it.

Re:clutching at straws

[Ironhandx]

Yeah, it was, but it was due to the way the game is programmed to handle things, and it was programmed to take better advantage of multiple cores, little to no testing was done on single core machines. Once the game engine is set up to handle multi-threading to begin with, patching on support for single cores isn't a huge huge deal, but it does add another layer for things to potentially break.

I think you are confuzled

[Sycraft-fu]

Intel owns Havok (since 2007) and licenses it out all over the place. There's a page that has all the titles using it (http://www.havok.com/index.php?page=available-games) and it is not a small list. Havok also runs on the CPU exclusively (and will probably continue that way since Intel wants to sell quad cores) so works no matter what your graphics card.

It's also not just physics anymore, there's Havok animation libraries and so on.

Re:I think you are confuzled

[Monkeedude1212]

I was mistaken, it was the DMM from Pixelux that was licensed - which AMD is also giving out, according to the article.

Is 'Incentivizing' Anti-Competitive?

[mpapet]

This kind of incentive is anti-competitive.

1. It eliminates competition by feature/functionality.
2. It meaningfully constrains innovation. A novel product without capitalization to participate is shut out. (That's the goal anyway)

That said, this kind of incentivizing is everywhere. (game consoles, mega-retailers, mobile phones) No one seems to care about the increased costs consumers assume or constraint on innovation.

I have my bias, what is yours?

Re:Is 'Incentivizing' Anti-Competitive?

[KillShill]

Nvidia is very anti-competitive and has been for a very long time.

The recent "making physx stop working when AMD gfx card is present" is just one of the more public outings of their unethical behavior.

I wish someone would expose all of their shenanigans and anti-competitive practices so people can realize how badly these things affect the industry and consumers (ugh, hate that word).

The most recent thing I read about their practices is from the upcoming PC game, Just Cause 2. There's a trailer showing off Nvidia-only effects ...(something which is dead standard DirectX code) and artificially blocking out AMD/others from getting the benefits. The Batman Arkham Asylam scandal was one more people may recall. They claim (and their users/shills) that TWIMTBP is just "marketing"... more like bribery and blocking out the competition. They've been caught on many occasions but the public rarely sees anything negative about them.

Nvidia is the Intel/Microsoft of the video card industry but unlike them, isn't quite as dominant (thankfully for us) but they still do a hell of a lot of damage. (The Jupiter of the computer industry... too small to become a sun but still an 800 quadrillion ton gorilla).

I've stopped buying Nvidia cards since the Geforce 2. At that time for performance reasons but since then I vote with my wallet and let others know to support fair and legal competition.

Not yet

[Sycraft-fu]

Though the reason for that at this point is the newness of the APIs, not because they can't be used. We'll have to wait a couple years to see if one or both of the technologies take off. Please remember that the OpenCL API didn't get finalized until the end of 2008, and GPUs didn't implement it until several months later. So there has been less than a year that one could develop on real hardware using it. DirectCompute was released with DirectX 11, October of 2009. Also it requires DirectX 10, 10.1, or 11 and as such requires Windows Vista or 7.

There hasn't been the time to develop a physics engine using either of the technologies yet and implement it in a game. Also the two big middleware engines have no interest in using it at this point. One of them is PhysX which is nVidia of course. They want to use it to help sell hardware. The other is Havok, owned by Intel. Well they too want to use it to help sell hardware, meaning CPUs in this case. As such it'll probably stay all CPU based.

Means is we are to see an OpenCL/DirectCompute physics engine in a game it'll either need to be custom developed for that game, or a new middleware solution from someone else.

Some truth to that.

[eddy]

While I don't think it's super dire, it's certainly a concern. I can add another point. Steam confirmed for Mac.

Problem? Macs don't take the latest and greatest off-the-shelf graphics cards, and generally are a fair bit behind the curve, way back in 'casual land'.

On the other hand, maybe if Apple open up a bit this is a way to sell more and better cards rather than another spike in the coffin.

Re:Some truth to that.

[Anonymous Coward]

While I don't think it's super dire, it's certainly a concern. I can add another point. Steam confirmed for Mac.

Problem? Macs don't take the latest and greatest off-the-shelf graphics cards, and generally are a fair bit behind the curve, way back in 'casual land'.

On the other hand, maybe if Apple open up a bit this is a way to sell more and better cards rather than another spike in the coffin.

There's a larger issue with that; nVidia's PhysX only works in Windows.

Another reason why OpenCL should be the way to go as it's cross platform, and also vendor agnostic.

Pure conjecture, but--

[Anonymous Coward]

What strikes me as the real issue here, is that game devs dont want to invest time on a proprietary graphics API that only a portion of their potentially targeted demographic will have access to.

EG, They don't want to use PhysX, when some non-trivial percentage of their customers will have AMD/ATI video.

It doesn't matter if PhysX would allow them to do real-time particle simulations of shrapnel, and thus create a more immersive FPS-- If some non-trivial percentage of their target audience cannot make use of that technology, then they need to be compensated for that loss.

To me, it isn't that PhysX is bad; it is that it is nVidia Only.

Take a short history lesson from the 1990s, when DOS games were the rage. Prior to the VESA standards people creating industry standards for high resolution video modes from DOS, certain games would only work on certain hardware; Same situation with audio capabilities. You wanted wavetable audio? Too damn bad unless you have a genuine MT32 plugged into your soundcard, or you have an actual AWE soundcard.

All that changed when Microsoft proposed DirectX for windows gaming.

Early versions of DirectX were indeed; total shit. Now, however, it is a well mature API, and is the primary target for game developers, because of its uniformity and ubiquity. It doesnt matter what random POS hardware is in there, as long as it has a directX driver; the game will at least start, and display a picture.

What needs to happen with "Processing on the GPU" taking off, is that a standardized implementation that is hardware agnostic needs to be drafted and approved.

Otherwise, it's just a case of yet another patent trolling, market playing pissing match between two or more squabbling children.

"He has to BUY PEOPLE OFF! His stuff is OBVIOUSLY crap! (Use my free license!)"

What we actually need is a platform and hardware agnostic API for doing GPU processing tasks. Not vendor kickbacks, like paid incentives or free licenses.

I find both players equally culpable in this debacle.

Best example with the MMORPG UTOPIA

[SharpFang]

A friend told me about his experience with Utopia. It implemented GPU-accelerated physics in one of recent patches. But try hard as you wish, he failed to notice any difference for weeks of gameplay. Until he entered the central city. With flags by the entrance fluttering smoothly in the wind, instead of the old static animation.

Yep, that's it. Many megabytes of a patch, a game of hundreds of miles of terrain, hundreds of locations, battles, vehicles, all that stuff... and physics acceleration is used to flutter flags by the entrance.

Re:Best example with the MMORPG UTOPIA

[VeNoM0619]

It may not have changed gameplay, but tell your friend to un-patch (if possible?). Then see if he notices anymore changes.

Hell, adding new sounds to a game and people rarely notice, further eye-candy and people may still not notice: However, there are some people who will notice and appreciate the attempts at realism.

So it may not be a game-breaker, but it can help the look and feel, like that paint job on your car.

My complaint

[w0mprat]

Nvidia has failed to engage the coding community in the right way. Any hardware-accelerated physics API needed to be openly available at the DirectX/OpenCL level from the begining. AMD has kind of seen the light here.

The original intention of Ageia and their PhysX set up seemed to be just to sell the company, rather than try to make a viable business model of selling hardware. Ageia would have been more open with API and code right from the start if they intended to make a business selling hardware.

Its all Hearsay

[KharmaWidow]

We don't have any proof that developers don't want PhysX. What we have is spokes person from company A saying that no one wants company B's technology. There are no scientifically obtained statistics only one guy's - a competitor - opinion.

Nor did the article state *why* it may be unwanted, or any specific why-nots for using PhysX

Re:Its all Hearsay

[Creepy]

Here's what I have to say:
nVidia bought the proprietary PhysX hardware acceleration. At that point it was a separate card, now it is integrated into GPU technologies. nVidia wants to push the technology to recoup their investment.

PhysX wasn't popular as a separate card and few game developers would want to include a solution that doesn't apply for all cards unless they have a fallback. Most write to the DirectX API because it works on all consumer graphics cards (and even being an OpenGL fan, their ponderous movement has made Microsoft the technology driver). If they want to support PhysX, they need to devote a team to implementing hardware accelerated physics and a team to either GPU shader based or CPU physics. Its an unwanted double effort.

Unlike DirectX vs OpenGL where OpenGL developers can capture maybe 10% more users at best (and have to suffer an API that is ponderously updated), the graphics card market is much more competitive. About 30% of game players use ATI according to steam. If that were 5 or 10% it may be acceptable, but 30% is a lot of income. Intel GMA (integrated graphics) actually leads the market, but they don't really make game capable hardware yet (and I am still skeptical about Larrabee - I'll believe it when I see it and it isn't a $1400 card).

Physics is coming and is needed in hardware for realistic realtime cloth and hair (for example). I would prefer it to be a "shader" with access to the vertex and texture buffers (to access height or bump maps)

AMD is pushing bullet, and while the current released version of bullet is still software (as is Open Dynamics Engine last I checked, #4 on the list), the dev version is OpenCL based. I'm not currently working on anything that needs physics, but when I do I will likely look into this. It is the first non-proprietary library to offer hardware accelerated physics that I know of.

re: Steam for Mac

[King_TJ]

Yeah... this is more of a solution than a problem, any way you slice it. Why? Simple ... Many of the games they'll deliver to Mac users via Steam will offer cross-platform network play. So regardless of the specs they're constrained to for a native Mac version of the game, it will help keep a title popular having more people playing it. They can always support higher-res graphics capabilities in the Windows version, if they so desire. And if they do? All the more incentive for Apple to start releasing better graphics options for their own systems.

STOP IT!

[Anonymous Coward]

in Korea, only old people use PhysX for the money, because Netcraft confirms that Apple is dying

Re:STOP IT!

[chronosan]

Oh crap, did I just cross the line into trolldom?

Re:STOP IT!

[Thanatiel]

No but certainly achieved sarcasms on a Sheldon-detectable level.

Wha?

[rgviza]

"They're not doing it because they want it; they're doing it because they're paid to do it."

Doesn't this describe just about any paid project? Just sayin'

Took first Linux/ATI plunge on 5870

[Anonymous Coward]

Because I heard that the driver quality was up, and because ATI has decided to embrace the fully open-source driver model (even if their fully open driver is just nascent), I took the plunge last fall.

I can thus far report the driver itself is working fine - roughly comprable to nvidia, with one exception where they regressed the ubuntu package build (and this could be fixed with a workaround that was relatively simple once you knew it).

Driver quality has IMO decimated ATI's business; many gamers and enthusiasts buy nvidia so that they can dual boot - i.e. are not stuck on Windows. Frankly even on Windows there were ATI driver issues for a while.

But if you were looking to pick up one of the hot new ATI cards, I can give you one data point that it's doable now.

Re:Took first Linux/ATI plunge on 5870

[toastliscio]

Ok thanks, I'll give it a try ASAP, but still I see a big difference between nVidia and ATI drivers on linux. For example, XBMC that in my opinion is the top for building up an HTPC, only supports hardware accelerated video decoding on nVidia, using VDPAU. That's an intresting feature if wou want to see 1080p video.

Optional Features are expensive

[happy_place]

The game development cycle is already airtight, because competition is fierce, and every new feature is old news in a few months, when your competitor games catch up. They hardly have time to test games, these days. Every day the game isn't on the market is money lost. And it's hard enough to debug a game with all the standard set of PC's, now add to that specific hardware configurations with specific feature sets, and you've got a testing nightmare. And what if there's a bug? what sort of support comes, if at all? it's more likely the game project management will more likely instruct the testers/devs to turn off the feature and go gold. --Ray

The physics business

[Animats]

Ageia's innovation wasn't their technology. It was their business model. Havok gets a fixed fee per title. Ageia's "physics chip" got revenue for each graphics card. Both Havok and Mathengine had serious revenue problems as standalone companies. The original investors did not do well. Both were eventually acquired. The basic problem is that game middleware isn't a good business.

Physics in the GPU is mostly useful for visual effects like water, snow, fire, explosions, etc., where the motion doesn't feed back into the game engine. Ragdolls and vehicles are usually still done in the main CPUs.

Duhhh!!!

[frank_adrian314159]

They're not doing it because they want it; they're doing it because they're paid to do it.

I can say the same thing of just about everyone who is employed, even the folks at AMD. Though, it's only in the "creative" arts where there's always this odd shiny coating of "fidelity" that seems to be desired and added on as a last step. In reality, this coating is as faux as the images and sounds that these arts provide. The bottom line - it's a business, any art is just an afterthought. If they can make more money with a feature plus marketing kickbacks than by leaving the feature out, they'll do it. I guess AMD is getting its ass kicked and that's why they're whining. Please, spare us...

Open standards, and nothing else

[unity100]

i wouldnt even care if physx was the biggest software innovation of the century - in gaming, especially in regard to graphics, we have suffered a lot because of proprietary shit in the last 2 decades. i dont want to see that again. even if its coarse, inadequate at the start, everyone should push for open standards so that we wont get in deep trouble later.

little things make a difference only if

[unity100]

they are implemented good. so far what i have seen in all those 'physics' engine games (irrelevant of whose engine is it) has been geometrically constructed splinters and pieces flying around. it takes more away from realism than it delivers, because the visuals they create (ie the distribution and nature of the destruction) is generally unrealistic.

At least they aren't directly sabotaging ATI cards

[mahsah]

Like they did with Crysis.

For the money?

[Darinbob]

Um, don't game developers only develop _games_ for the money?

But this is GLIDE

[Chris Burke]

Like OpenGL -vs- DirectX.... oh... wait... :-P

Ah, but see, that right there tells you why PhysX is doomed. The PhysX-equivalent from the early days of consumer 3D was 3DFX's GLIDE library. By the accounts I've heard a fine library, and it certainly did its job of enabling game developers to support 3DFX cards. Using it made good sense, even though I'm sure 3DFX payed some developers to do it, because at the time Voodoo was the only game in town. Nobody else provided consumer/gaming 3D capabilities on par. So that was good, for a time.

But then as soon as other graphics card manufacturers started making cards with similar capabilities, suddenly the value proposition for a game developer changed. Soon, you basically the choice of OpenGL, Direct3D, and GLIDE, and they broke down like this:

OpenGL: Runs on any platform.
Direct3D: Runs on any Microsoft platform.
GLIDE: Runs on any platform with a 3DFX card, and then still mostly Microsoft-only.

You can argue back and forth about the merits of Direct3D vs OpenGL and giving up non-MS platforms -- if you're including consoles, then maybe not a great idea, but if you're targeting the PC, then MS-only is still 95% and hell lots of games decide to be MS-only before the graphics API decision even gets made. But to go after only the subset of PCs that have a 3dfx and not an NVidia, ATI, or Matrox card? Yeah right! GLIDE dropped off the face of the planet.

PhysX is in the same boat. It's a library that should be replaced by an open standard like OpenCL that AMD is pushing, but may instead be replaced by Microsoft's DirectCompute or whatever they call it, or possible an ecosystem where both exist. Either way, unless PhysX somehow becomes the new MS library, it's going away and in 10 years nobody will remember it.

For the money ?

[morcego]

They're not doing it because they want it; they're doing it because they're paid to do it.

So we are to believe that working for AMD, we can do whatever we want, and not be ordered to do stuff we don't want ?

And what ? Developer companies doing things for the money ? No, I can't believe that. I mean, all their games are free today, right ?

This simply can't be true. I never heard of any company doing their stuff for the money. They all do it for the sheer pleasure of whatever it is they do.

In my experience, PhysX has only been a hindrance!

[Roger Wilcox]

I have to disable PhysX in the nVidia control panel to get HL2 or any of the Source engine games to run properly! I had no idea what was causing these games to crash. After disabling PhysX they work right every time!

Apparently it doesn't do anything crucial or even noticable as my games run just fine with it turned off. And now I'm told the game devs don't even want to use it?

This "feature" has caused me nothing but grief!

Is that "Both country and Western"?

[jonaskoelker]

AMD is advocating for a similar technology that works on their hardware as well as on nVidia's

That sounds a bit like "both country and western" to me.

Ah well, I guess it isn't AMD's fault that when they make open standards there are only two (maybe three) implementers...

Re:At least they aren't directly sabotaging ATI ca

[CyberDragon777]

With recent driver versions if you put an ATI card in your machine next to a nVidia one PhisX gets disabled.

http://www.tomshardware.com/news/nvidia-physx-ati-gpu-disable,8742.html