Slashdot Mirror
NVidia Cripples PhysX "Open" API
An anonymous reader writes "In a foot-meet-bullet type move, NVidia is going to disable PhysX engine if you are using a display adapter other than one that came from their company. This despite the fact that you may have an NVidia card on your system specifically to do this type of processing. 'For a variety of reasons some development expense, some quality assurance and some business reasons Nvidia will not support GPU accelerated PhysX with Nvidia GPUs while GPU rendering is happening on non-Nvidia GPUs.' Time to say hello to Microsoft dx physics or Intel's Havok engine."
Havok is a better engine anyway.
But that's the problem with corporate buyings anyway. Even if its kinda wrong to stop supporting the other platforms, they have every right to do so.
Why is this not anti-trust? When you paid for the nVidia card to put into your machine why should its functions depend on whether or not a competitors hardware is present? What if Windows said uh-oh you have Linux installed on another partition, disabling Windows...
Shh.
At least he was 33.3% truthful.
...will it my $TERM faster?
Need an ISP in South Africa?
I was about to start using it, this announcement has saved me a lot of wasted effort.
No sig today...
Was Nvidia previously offering a software framework that could run on any GPU, but now only supports their own? Can ATI (or anyone else) not implement the standard in their own drivers?
bullet
First they scoop up PhysX and try to create a market for PPUs. Now the only way PhysX is ever going to get any use is out of pure coincidence. Not the smartest move for Nvidia to make when Ati/AMD is on their heels with a new line of cards.
I guess it never occurred to me to spend hundreds of dollars on a graphics/physics card only to not use one of the primary functions. I have an nvidia card, I don't notice the physics stuff, doesn't seem to make a difference anyway.
Heres some thoughts on the meaning of this. The PC is an open-architecture, you are free to put whatever you want into your machine. If nVidia can dictate what their hardware works with then they are effectively creating a "nVidia-Approved" list of hardware. First step down the slippery slope of closing the PC's openness. In the software world an equivalent would be Windows refusing to connect to network shares that were based off of Samba or the other way around a Windows box refusing connections from Linux machines. Standards apply to hardware as well as software and if any manufacturer gets away with an "approved" list then the platform as a whole will eventually suffer for it.
Shh.
physx seemed nice until they tried to close source it. Does Nvidia have anything left this round? Bad Yields, physx being stupid and abusive when disabled (it only uses 1 cpu core when on AMD for example instead of even all threads). Not to mention their crippling of batman as well.
So what's left for Nvidia? I don't see a whole lot.
Stop things like this from working?
There's a word in the article headline conveniently omitted from the Slashdot headline. That word is "reportedly"
Seriously, guys, can't we get any kind of standards here?
That aside, this is a pretty stupid move. If this news is accurate, I don't doubt a lot of users are going to be pretty vocal.
On the other hand, if they had made it work, but be horribly broken in the presence of an ATI/AMD graphics cad, they could easily blame it on something completely opaque to the user and get away with it. (cf. manufacturer graphics drivers on Linux.)
It'll get cracked
It'll get used
If it breaks,
they wont be to blame.
Locked in you say?
No Way!
I know nothing about PhysX other than what I've gleaned from the article..
If you buy an nvidia card to do some headless gpu grunt work, they will disable the functionality to do that unless the work is being shown through another nvidia card?
The displaying of the work is pretty much superfluous to the work being done, and they've already made their money on selling, support, etc the PhysX card.
Err?
Has anything changed with Windows 7 where you can run an ATI and Nvidia card at the same time? I know you could in XP, but I found out the hard way you couldn't in Vista. It was something to do with the new driver model.
I was trying this over a year ago to get dual monitors working while having SLI enabled under Vista. The recommended solution was to use an ATI card for the secondary output since the Nvidia drivers wouldn't see it and disable your ability to use SLI. When I tried to load the ATI drivers I couldn't get them to detect the card when the Nvidia cards were already active. I ended up being able to use a lower end Nvidia card which solved my problem.
Who on earth has a graphics card from two different manufacturers? Regardless though, it means they've directly tied PhysX to their hardware, and I just don't care for them anymore. ATI all the way baby!
I'm currently avoiding PhysX due to the fact that the license requires that credit be given to nVidia/PhysX in any advertisement that mentions the advertised product's physics capabilities. It's a real shame, because I hear that PhysX has pretty robust physics implementation.
The current state of physics acceleration reminds me of the days when hardware-accelerated 3D graphics (except for high-end OpenGL stuff) were only supported through manufacturer-specific APIs. Hopefully, DirectX physics will be good enough that PhysX will ultimately become mostly irrelevant to game developers -- I'm just not convinced that Microsoft can pull it off.
"In prison you just have to shut your eyes and take it. Here you have to shut your eyes and give it."
Between the full stack (CPU+Chipset+GPU) provided by AMD and the full stack that will be Intel (with Larrabee in 2010) Nvidia has no future in either Chipsets or GPUs. Any other outcome is a bet against integration and in electronics integration always wins.
Good thing too; both Intel and AMD are vastly more open (at least recently) with their hardware.
Techspot AMD has been working hard to develop Open Physics. Furthermore Bullet Physics has been shown running on Cuda. So that sounds to me like doom for physx...
It's kind of crazy that this is even going to get attention. This is only going to affect people using PhysX (which requires an nVidia GPU at the moment) with an ATI card for rendering. I'm sure the two people with this configuration are going to be crushed. Yes, I realize more than 2 will have a mix of cards, and 2 is probably a bit of a low guess, but only a handful are going to actually be affected by the lack of PhysX support for the config, so please, let's not get all in a huff about it. From a support perspective, I can understand where nVidia is coming from. This could be a true support nightmare for them.
Once the big game engines and physics libraries get generic support for GPU programming through OpenCL, this will all be pretty moot anyway. From what I can tell, the bullet physics library is already developing this, and I am sure closed source competitors are doing that as well. Relying on anything that will only run on a single vendor's hardware is just a losing business proposition (unless that vendor pays you for it, which I guess is how PhysX got going).
http://www.bulletphysics.com/
I don't have any affiliation with the project other than I've used it in my homegrown game engine that has never left my hard drive. It is however rather easy to use. When I was looking for a physics engine, Bullet turned out to be the best license, code base, and documentation set out there for no cost.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
According to a game console dev I know, he said that PhysX is pretty buggy as it is and pales in comparison to the superior Havok engine. On top of that Havok already works on many architectures.
If this report is true, this one more nail in the coffin for PhysX.
Nvidia releases announcement that they will no longer provide free driver support to ATI for interaction between Nvidia hardware and ATI competing hardware. Notes that software APIs are available for ATI to pay for and release their own damn drivers.
NEWS AT 11!!!
PhysX was trying to make a market for PPUs (and relatively failing). nVidia bought them up to make the technology another marketing bullet point for their GPU parts, not to sell GPU parts as mere physics calculations. Sure, they'll take the business as it comes incidently, but they have no interest in anything that could remotely be construed as putting something other than their role as a graphics adapter vendor first.
XML is like violence. If it doesn't solve the problem, use more.
You really can't blame them for dropping support for CRTs. If you can even buy them anymore, you'd have to be insane to want to.
You express a desire for an API from Microsoft to become dominant?
XML is like violence. If it doesn't solve the problem, use more.
A bigger issue is that Nvidia bought PhysX and was the one offering support for standalone PPU cards. These were sold to be used with Nvida or ATI GPU cards. Nobody told the ATI user they chouldn't buy the PPU cards.
:P
Then Nvidia decides to break their system and render this worthless... Sounds like poor support for their customer base to me.
Of course, if the scenario that I described above is not what's actually going on, then never mind.
If you think about it, physx works on all 8 series and up.
That's a $30 card for physx support. I wonder if I can do this since I have a spare x16 port on my machine.
I don't really know if this will work though.
They're using their grammar skills there.
I'd prefer something open, but at least a Microsoft API might be compatible across different brands of hardware.
"In prison you just have to shut your eyes and take it. Here you have to shut your eyes and give it."
Let's see you run old games made for Voodoo and Glide. Same thing with all those PhysX and CUDA and whatever.
OpenCL is the way to go. And before you all go up in arms saying FUD like "it's a proprietary Apple technology", no it's not. No more than AAC isn't from Apple, nor is H.264.
Death to DivX, the fucking AVI format, bullshit Windows Media formats, and closed proprietary crap like PhysX and CUDA!
You express a desire for an API from Microsoft to become dominant?
APIs are fairly easy to copy, and they cannot be copyrighted. API lock-in is definitely better than hardware lock-in.
Unless you can show that there's no technical distinction, a reasonable person would have to presume that Nvidia is on the up and up.
Nvidia doesn't sell LCDs.
Nvidia says they support any 100+Hz CRT.
http://www.nvidia.com/object/3D_Vision_Requirements.html
That's quite a contradiction you made there.
Now, what is the difference (tech-wise) between their shutter glasses and mine? Only the fact that theirs send a specific "yes I'm nvidia" signal back to the card.
I call this the "fill factor" as in landfill. I had a perfectly good Canon scanner that died when Canon decided it would never have a driver more recent than Windows 2000. Off to the landfill in near mint condition. (Or off to China in a mist of CO2 and particulate heavy metals). Generally my experiences with Canon have been good, but this one did not impress me.
I'm appalled at waste streams involving perfectly good manufactured goods that outlived its software. Stewart Brand was joking about "squanderable energy". You mean we don't already have that in the energy invested in squanderable applicances?
squand
I understand his inner mirth at the term. It's against nature to squander entropy. It's like saying "squanderable blood". Maybe from one vantage point. The sentiment is rarely universal.
I've never reconciled myself to some of the emergent stupidities associated with free market capitalism. The main argument is that most of the cures (regulation is the most cited) are worse that the disease--until we caught a bad case of trillion-dollar bail out. Taleb and Summers refer to this outcome as the "privatization of gain, socialization of loss". This was true for a long time about the medical profession (who were slow to grasp sterilization).
Ignaz Semmelweis
Semmelweis was puzzled that puerperal fever was rare amongst women giving street births. "To me, it appeared logical that patients who experienced street births would become ill at least as frequently as those who delivered in the clinic.
OK, so sometimes the cure *is* worse than the disease. But that won't last forever. Free market capitalism (and its champion of the hour, Nvidia) does such a good job of motivating economic Semmelweis's to contemplate the alternatives, if the system can't manage to eliminate it's own issues, we'll eventually reach the point of replacing the whole thing. God forbid it's any system we've tried already.
We haven't yet invented the germ theory of capitalism, so I'll have to content myself by referring to Nvidia's PhysX business decision as a cadaverous particle.
This is the official begining of the death of PhysX...
NVidia fought DirectX10 GPU computing/physics technologies, and got Microsoft to scale back DX10, which originally looked more like DX11, as DX11 is basically the technology the XBox 360 is using and what Microsoft wanted to make DX10.
However, NVidia threw a fit and because the 8xxx series of cards would not have been DX10 based on the first DX10 specifciations, Microsoft gave in when they should could have really messed up a full generation of NVidia GPU technology. (ATI has been ready to go with consume DX11 GPUs for several years, because they have the XBox 360 GPU technology Microsoft developed and gave back to ATI. -Although DX11 does have a few more features now that a few years have passed.)
When developers are already using Havok and DirectX technologies, and even DX10 will do quite a bit of GPU computing already, there is no need for PhysX anymore.
Developers will look at this move, and either abandon PhysX out of disgust or take NVidia at their word that PhysX is bugging, and can't be trusted to run on a system with another GPU rendering the video. (And neither view of NVidia is a good one.)
It's not like there were much more than a simple linear solver kernel and a radix sort kernel provided by the CUDA backend. No real loss. The trick, you see, is that nVidia provides an unoptimized and possibly buggy code-path for systems without nVidia GPUs present. It never ceases to amaze me how they find the least ethical solution to every problem. Anyone who has done any real amount of GPGPU and modern visualization software development could tell you that the granularity at which you need to dispatch physics simulation kernels would completely defeat the purpose of peforming simulation of game world physics on the GPU. nVidia is really just a giant marketing company that happens to have a chip design and marketing arm.
I know the last, probably 5, games i have played required the physX engine to play. does that mean that games that me and others used to play/own will no longer work? Because that definitely sounds like something that should be illegal to me, if that is the case.
Troll is not a replacement for I disagree.
Do you have any idea what DirectX actually does in its domain?
I am somewhat struck by how readily some posters here are willing to bend over, lube up, prepare with a series of ever-larger dildos, and wait for the inevitable. My own computing experience, because of choices I made two decades ago, involve working with a lot of "unsupported" hardware, perephrials, and the like, in combination with my choice of "not the majority" hardware & OS.
For all of those two decades, "unsupported" meant that I figure out if it will work, I do the config myself, and if it doesn't work, I'd better come up with a story that jives with a failure mode on the supported hardware (in other words, sometimes stuff is actually broken, and sometimes I can't get it to work ... If it's actually broken, I need to convince them it's broken on a supported platform; if I can't get it to work, I eat it).
This is a long way ... A VERY LONG WAY ... from disabling the thingy if I don't use the very specific hardware and/or OS that actually is supported.
Unsupported means I get to figure it out by myself and no refunds ... this is not "unsupported", folks. This is extortion.
I can't think of any sane reason to use this configuration.
So, you've got two PCI-express slots in your system. If you bought the graphics card for physics, and you're worried about drain, go buy a second one. And then use SLI. You'll get better graphics performance since the physics processing doesn't use the whole card's processing power.
This isn't a knock against ATI; it's just that if you want to use physics and have high performance, it just makes sense to have 2 cards do the rendering rather than 1.
The reason for doing this isn't likely some evil hatred of ATI; it's probably testing. I'm a developer of commercial openGL software...and we target certain graphics cards. We test the bejesus out of them and then, sometimes, have to go in and fix code or provide support to the vendor when a bug pops up. I think nVidia thought the developers/testing manpower would have better been spent on something else...even porting the physics engine to ATI would be a better use of time.
-- Political fascism requires a Fuhrer.
They've decided to stop supporting old tech. That's hardly shocking.
The "original source" of the perplexing xbitlabs article. An internet forum, not confirmed nor corroborated by anyone. Here's hoping that xbitlabs publishes a retraction for their non-existant fact checking and/or technical editor oversight: http://www.ngohq.com/graphic-cards/16223-nvidia-disables-physx-when-ati-card-is-present.html
(For what it's worth, NGOHQ was also the source of the rumor about ATI adding PhysX support themselves, along with 'Russian Hackers' publishing info on how to enable PhysX on ATI 'next week' last year.)
"A Goddess rarely smiles for she is forced by others to be an island unto herself." - Zephiris
Citation required.
About the only dangers I can think of are dropping the CRT on your foot or maybe if you crushed up the leaded glass and ate it. A modern CRT doesn't shoot X-Rays out of the front.
“Common sense is not so common.” — Voltaire
If there's copyright on the code, there's a monopoly.
By definition of patent/copyright.
"For a variety of reasons some development expense [10%], some quality assurance [10%] and some business reasons [80%] Nvidia will not support GPU accelerated PhysX with Nvidia GPUs while GPU rendering is happening on non-Nvidia GPUs."
But currently *everybody* can get their hands on it, and they don't even have to pay money to use it in a commercial product anymore! Because Intel bought them and is basically giving the technology away to anyone who will use it, because Intel will optimize it more for their stuff and so it will help make Intel look good with gamers.
Compare that to PhysX which only works on a subset of hardware, and whose owner (Nvidia) seems to be exhibiting control-freak tendencies towards it that rival those of Apple.
I know which one I'd choose. Disclaimer: I've used Havok on several AAA game projects, and I've never used PhysX. And that trend is likely to continue after this move by Nvidia.
How's that any news? NVidia has plenty of software that's limited to their HW.
AMD+ATI and Intel+Havok seem to remain "good" even though they could disable things with competitor's HW.
Anyone has evidence of evil there?
In other words, "dont be evil" (not like Google).
It seems like they use an IR transmitter to sync the glasses with a pass through for the monitor. Do your glasses have an IR receiver?
It seems that most LCD's that claim to run at 120Hz are using shenanigans to get that high; they aren't really showing 120 images per second.
My guess is that the wired ones had a clock signal on that wire, and no logic in the glasses. I would expect the Nvidia IR glasses would use a start/stop signal with onboard logic. Maybe the old IR glasses do something else?
I know its domain intentionally excludes OSX, Unix, and Linux (Wine comes closest, but always is subject to the whims of MS). Mono trails MS in .NET at all times and that's when MS is ostensibly *trying* to play nice. MS isn't trying to play nice at all with DirectX.
Microsoft's agenda is clear, they have a lockin technology and they want to keep it that way.
XML is like violence. If it doesn't solve the problem, use more.
So that's a no.