Nvidia Engineer Asks How the Company Can Improve Linux Support

sfcrazy writes "It seems that recent comments made by Linus Torvalds have made the people at NVIDIA take Linux more seriously. Recently Nvidia employee Stephen Warren asked in the Kernel Summit mailing list what could be done differently to make Linux support better. 'In a Google+ comment, Linus noted that we have mainly been contributing patches for Tegra SoC infra-structure details. I'm curious what other areas people might expect me/NVIDIA to contribute to. I assume the issue is mainly the lack of open support for the graphics-related parts of our HW, but perhaps there's some expectation that we'd also start helping out some core area of the kernel too? Would that kind of thing help our image even if we didn't open up our HW?'"

Comments

Ugh, this makes me mad.

[intellitech]

[...] we'd also start helping out some core area of the kernel too? Would that kind of thing help our image even if we didn't open up our HW?

You seem to care more about NVIDIA's image than about what the Linux community actually needs.

I truly don't understand what the big deal is. Just open up your damn specifications already.

Re:Ugh, this makes me mad.

[Anonymous Coward]

Yea Just strictly license it just to linux and android and no other . I am sure that can be done.

Re:Ugh, this makes me mad.

[ethan961]

If this engineer knows that will never happen (through no fault of his own, higher-ups would have decided this) then at least he's making an effort. People can get upset all they want, but nothing other than good is going to come out of this, whether it's exactly what would be ideal or not. Some help is better than none.

Re:Ugh, this makes me mad.

[FridayBob]

... Some help is better than none.

Okay, so how about pointing him in the direction of the nouveau project? Even if his company refuses to share the full API, just a few hints here and there could make an enormous difference.

Re:Ugh, this makes me mad.

[lister king of smeg]

not if they plan on complying with the gpl

Re:Ugh, this makes me mad.

[SwashbucklingCowboy]

"You seem to care more about NVIDIA's image than about what the Linux community actually needs."

Of course they do. Why in the world would this surprise you? It's a public, for-profit company. The vast majority of contributions to open source by companies are not done out of altruism. They're done out of self-interest.

"I truly don't understand what the big deal is. Just open up your damn specifications already."

There's probably something in how they do things that they think (rightly or wrongly) that gives them an advantage over their competition that either isn't patented or can't be patented (or maybe violates someone else's patents).

Re:Ugh, this makes me mad.

[Anonymous Coward]

The big deal is that different graphics cards with different prices have exactly the same hardware, just different drivers.
There's no way to give the users what they want and maintain this important business practice.

Re:Ugh, this makes me mad.

[ethan961]

I agree, helping the nouveau project would be greatly beneficial. My post was mainly referring to the general attitude some hold that nothing is ever good enough, and while it is true that the driver state is rather poor, potential for improvement exists. Instead of attacking Nvidia for their efforts, people should actually help and show them where to direct their resources. No, it may not involve the release of all their specs, but they can help in other ways.

Re:Ugh, this makes me mad.

[Sir_Sri]

Or would give away something about how their hardware, or hardware development works that they don't want anyone to know too much about. Or at least not to wave around in public. I'm sure AMD know, enough staff move back and forth, but some random chinese hardware maker, that I'd be worried about.

Remember nVIDIA sells it's 'first silicon' runs, they don't usually prototype, they do simulations first, and from the simulations that are used to design the hardware they go straight to manufacturing. It's possible some of their driver support would have some hooks (or parts) of their simulation tools in it that they don't really want to be broadcasting around.

But broadly ya, for whatever reason, if their higher ups don't want to be telling the world how their drivers work, and so they won't. Hell for all we know it could be entirely paranoia from an MBA type who just doesn't understand and doesn't want his ass fired if something goes wrong.

Re:Ugh, this makes me mad.

[ozmanjusri]

They could have been a brilliant contributer, if they wanted to, but they don't.

Not to diminish Linus's effort, but there is a backstory to his frustration, and nVidia have been given other incentives to become a better contributor.

NVIDIA was approached by one of the leading Chinese CPU teams to use an NV GPU in a pilot school PC project. Linux would run on the Chinese CPU, while GeForce GPU would provide the graphics power. 'Pilot project' in this case means over 10 million PCs in one order, broken down - 100,000 schools with 100-150 PCs each.

To cut the story short, the NV team appeared there, and in very arrogant manner told the Chinese side that they are a large US corporation, and that recompiling the Linux drivers would cost the Chinese a lot of money.

http://www.brightsideofnews.com/news/2012/6/21/china-nvidia-loses-face-and-a-10-million-pc-order-over-linux-drivers-and-nres.aspx

They lost the relationship with the Chinese team, who have since approached AMD. The pilot project was worth an initial 250-350 million dollars, with the potential of much more to follow.

Re:Ugh, this makes me mad.

[bug1]

"You seem to care more about NVIDIA's image than about what the Linux community actually needs."

Of course they do. Why in the world would this surprise you? It's a public, for-profit company. The vast majority of contributions to open source by companies are not done out of altruism. They're done out of self-interest.

Good engineers dont think like marketing people, they think like engineers. How to do stuff rather than how to manipulate people.

Re:Ugh, this makes me mad.

[93 Escort Wagon]

Racist.

How is this racist? The only way you could construe it as racist is if you are ignorant of the origins of the phrase "yellow journalism".

Re:Ugh, this makes me mad.

[X0563511]

We're asking for specs, not driver code. There's no reason they would have to give us any of those simulation hooks, or any information about them.

Re:Ugh, this makes me mad.

[ozmanjusri]

Did you miss the bit where it was "a project that required drivers that NVIDIA needs to develop anyways (after all, CARMA toolkit is consisted out of ARM-powered Tegra 3 processor and a 96-core Quadro GPU)"?

Re:Ugh, this makes me mad.

[erroneus]

The writing is on the wall for Windows and generalized desktop computing. It's not going to "end" but it will cease being "the thing" and become one thing of many.

Our computers will become our mobile devices. The keyboards, displays and other I/O devices will connect to our mobile devices in whatever way is most appropriate for the environment. (For example, at work, you will probably have a display and a keyboard. In the car, you will have an enhanced dashboard and controls. Other places, you will have to use the touch display or mumble at your device to make it work... or even "think" at it and have audio feedback in the form of words or other sounds.) Lately, all of the newest and most cutting edge devices run Linux on the front end and Linux on the back end. All this crap about components and other things are going away.

NVidia can go away with them. Patents run out. The Chinese will put out clone hardware which out-performs or is better optimized for the environment, nVidia is TRYING to be an important figure in the mobile hardware arena, but if they want to participate, they had better join forces with Linux in a serious way. Microsoft is NEVER going to be a player in the mobile markets. And I seriously mean never. They can't do it. Their history of failure proves it. They keep wanting to adapt Windows into something it can't be and they seem incapable of creating something entirely new... and what's more, no one wants it if it comes from Microsoft anyway.

If nVidia wants to protect the software portion of their devices, then they need to figure out another way. Get involved with a different blend of hardware/software interfacing and take a lesson from other processor designers and start moving your precious software into microcode and let OSes use more generic interfaces to access the hardware. In other words, stop putting your "secret stuff" into the drivers. People have already disassembled your code... they just can't admit to it in public.

Re:Ugh, this makes me mad.

[ozmanjusri]

It is supposed to care more about itself rather than some open source project. By law.

Which law?

Link please?

Re:Ugh, this makes me mad.

[jvillain]

Telling them that it will be OK to slide on the documentation for the video drivers just as long as they do X, Y & Z would be a real disservice to them. The blob just isn't an option any more. Telling people which versions of Linux they can use the video cards with is equivalent to telling Windows users they all have to use Vista if they want to use NVidia hardware. For me right now all of their cards are the equivalent of a 10 year old 2D Cirrus card but with a stupid high price tag attached to it. Get it straight. Every thing takes a back seat to opening up the documentation for the video drivers. My money stays with AMD until they yank their heads out of their ....

Re:Ugh, this makes me mad.

[jvillain]

BTW we aren't asking them to open up the HW. We are asking them to document the interfaces to the hardware. Just like the PCI interface they use has been documented and like the entire PC spec they have been leeching off of and almost every thing else in the world of computing has opened up. This attitude just isn't acceptable any more. Period. Even Microsoft is starting to wake up to that fact.

Re:Ugh, this makes me mad.

[kllrnohj]

You seem to care more about NVIDIA's image than about what the Linux community actually needs.

I truly don't understand what the big deal is. Just open up your damn specifications already.

AMD did this, and the Linux community still recommends people buy NVIDIA cards. Vote with your wallet, or shut the fuck up. Keep buying from the company with closed specs instead of open ones, and you'll keep getting closed specs instead of open ones. Not a difficult concept, people.

Re:Ugh, this makes me mad.

[bky1701]

False. Completely false. The drivers for different cards are always the same, as long as they are supported. You are referring to differences in the card BIOS - which has nothing to do with the driver. Unlocking disabled features is not done via a driver, but via overriding memory registers in the card during runtime with an external program, or permanently altering the BIOS (after which it will have the features enabled, no matter what OS it runs on or drivers it uses). Documenting the API would not change anything since the information on how to enable unused card functionality is already well-known and widely used. Not sure how you got modded up.

Re:Ugh, this makes me mad.

[jvillain]

I guarantee some one in Asia has seen this brouhaha and is trying to figure out how to make a buck selling video cards to Linux users with proper documentation for the interfaces. They don't think like Americans there. They don't just sit on their arses and wait for the government to pass another law so they don't have to compete. Truth is soon Intel hardware with the open source driver will be more powerful than the NVidia card with the open source driver. But doesn't cost as much. Obviously Intel has learned the power of doing the right thing.

Re:Ugh, this makes me mad.

[Pausanias]

I'd put money on the likelihood that there is a GPL violation in their driver code that they'd have to own up to if they actually released the source.

Re:Ugh, this makes me mad.

[X0563511]

So the nouveau guys should just give up and delete everything they have already done?

Re:Ugh, this makes me mad.

[Anonymous Coward]

The term doesn't really fit here, "yellow journalism" was obviously making a stupid joke.

Re:Ugh, this makes me mad.

[kamapuaa]

The story you link to refers to nothing but rumor, and seems to have been copy-edited by a couple of disinterested high school students. It also mentions the idea that China is going to have the world's fastest CPU within a year, mentions the idea of "losing face" like they took "Asian Stuff 101" sometime in the 80s, and refers to the reasonable request for money to make a port as being incredibly unreasonable, some sort of slap in the face to the Chinese.

Additionally, the idea of China making a single order of more than 10 million Linux computers is a little hard to swallow, when the only support is "A rumor appeared from the heart of Beijing."

The article is merely an example of shitty web journalism.

Re:Ugh, this makes me mad.

[ozmanjusri]

I know companies that don't aim at maximizing profits can be sued by their shareholders

Not true. And even if you WERE trying to maximise value to shareholders, there's a distinction between long-term sustainability and short term profit.

The Myth of Profit Maximizing

        “It is literally – literally – malfeasance for a corporation not to do everything it legally can to maximize its profits. That’s a corporation’s duty to its shareholders.”

Since this sentiment is so familiar, it may come as a surprise that it is factually incorrect: In reality, there is nothing in any U.S. statute, federal or state, that requires corporations to maximize their profits. More surprising still is that, in this instance, the untruth was not uttered as propaganda by a corporate lobbyist but presented as a fact of life by one of the leading lights of the Democratic Party’s progressive wing, Sen. Al Franken.

http://www.alternet.org/economy/154789/whose_corporations_our_corporations!?page=entire

Re:Ugh, this makes me mad.

[rtfa-troll]

If this engineer knows that will never happen (through no fault of his own, higher-ups would have decided this) then at least he's making an effort. People can get upset all they want, but nothing other than good is going to come out of this, whether it's exactly what would be ideal or not. Some help is better than none.

Right; I'm not getting upset about him. We love him and if he ever gets kicked out by NVIDIA, I'm sure there are lots of us here who will want to hear about it and offer him a job. What upsets us is that NVIDIA clearly has people who understand that we are upset and why we are upset and they still don't do anything. From my point of view this makes things even worse.

NVIDIA; please: even if it's incomplete and misses your latest fastest newest stuff, please provide the Linux developers; both kernel and Noveau; with publishable open documentation which tells them how to set up the rest of the features and at the very least provide good, reliable and colour tunable 2D and low level support for your cards. Preferably provide full, documented access to any feature older than two years.

This can have no possible bad influence on your competitive advantage and will go a long way to making those of us using Linux happier.

Re:Ugh, this makes me mad.

[Anonymous Coward]

The big deal is that different graphics cards with different prices have exactly the same hardware, just different drivers.
There's no way to give the users what they want and maintain this important business practice.

When chips are manufactured, you don't end up with a stack of perfect chips and a stack of scrap. You end up with a stack of good enough chips, a stack of not quite as perfect chips, and a stack of scrap. So a chip designed to run at, let's just throw a number out and say 1ghz, might get 80% of the manufacturing run which performs at 1ghz and maybe 10% which won't handle more than 750mhz without errors but works perfectly fine if you keep it turned down. Or maybe a manufacturing defect in 20% of some auxillary chip prevents just those chips from being able to fully support some type of function. So you throw the semi-rejects into lower-price point models and disable some of the features in the BIOS so it doesn't overdrive the actual capability.

It's actually very rare for all the boards in the same model and version line to all have the same exact chips... and even when they do they are not usually all capable of the same performance. It's not done to gouge the customers who pay more or to make the lower-end models suffer, it's done to maximize the amount of money you can get out of a manufacturing run. Just because they didn't tell you specifically why board A is cheaper with less functions than board B when they appear identical doesn't mean they're playing a shell game, it just means you aren't privy to all the information.

Re:Ugh, this makes me mad.

[miknix]

I can give them some ideas:

What about starting to properly support XRandR and stop the TwinView crap?
What about continuing giving minimal support to old GPUs? My parent's computer has a GeForce FX Go 5300 which has currently no official driver that supports the latest Xorg. Unfortunately nouveau is not working very well with this graphic card either, being the 2D support buggy.
What about supporting KMS in their drivers?

Not going to happen, that's why my next computers will have Intel GPUs.

Re:Ugh, this makes me mad.

[Runaway1956]

Good Point. (sorry - lower case "P"s seem to be dead on this keyboard)

It is easy to imagine that Nvidia might make a contribution or more to the kernel, that would helP Performance. I'd much rather see everything oPen sourced, but if that can't be done - let them get their hands dirty, and make the drivers work better for Linux. Even if they never actually contribute kernel Patches, the attemPt to do so can only give them a better understanding of the whole system!

Re:Ugh, this makes me mad.

[rsmith-mac]

The article is merely an example of shitty web journalism.

Aye, BSN is a technology tabloid. They publish crap because people enjoy reading it, not because they believe in journalistic integrity or even the truth.

Anyhow, the article completely fails the capture the fact that NVIDIA's desktop graphics products are significantly invested in x86. Porting the whole thing over to MIPS (and then keeping MIPS in sync) just for China would be a massive undertaking.

Re:Ugh, this makes me mad.

[JustOK]

how so?

Re:Ugh, this makes me mad.

[Man On Pink Corner]

I truly don't understand what the big deal is. Just open up your damn specifications already.

The big deal is the army of patent trolls who will jump out of the bushes with baseball bats the minute NVIDIA or any other major GPU manufacturer publishes a full set of register-level specifications.

Don't like it? Write your congressman. NVIDIA can do nothing to help.

Re:Ugh, this makes me mad.

[Runaway1956]

Bottom line seems to be that AMD got the contract. It seems to me that your objections to the story are meaningless.

Re:Ugh, this makes me mad.

[philip.paradis]

This is utterly wrong. It's telling that you posted as an AC, and I'm assuming that's because you're not a developer involved in any open source work on this issue, and you don't have the first clue about how these things actually work.

However, you do sound like a user with an incredibly exaggerated sense of entitlement, or perhaps a college freshman with wild-eyed ideas that every line of code on the planet should be forcibly "freed." If the code for something hasn't already been written, I presume you're of the view that companies should be forced to pay developers to write it, and of course they should be sure to call you every day for your personal stamp of approval on the last 24 hours of coding. Naturally, the resulting code, which you didn't pay for and didn't write a single line of, must be licensed in precise accordance with your personal opinions on freedom, right?

Let me be absolutely clear on this: the core developers in the open source community working on this stuff want the specs. Just like anything and everything else that open source code was been written for, even back before the term "open source" was even in common usage, precise hardware specifications allow developers to write software that properly interfaces with the specified hardware.

I will be the first to stand up and strongly advocate for open hardware specifications. That said, the absurdly entitled position you're taking does far more harm than good in the community at large. In short, kindly shut the fuck up and stop ranting about things you have no understand of, and no honest stake in.

Re:Ugh, this makes me mad.

[Anonymous Coward]

I truly don't understand what the big deal is.

When designing hardware peripherals, it's both a skill and an art defining the interface to the CPU. On the one hand, you can use dedicated logic in the peripheral to do everything, but that comes at a high cost. Not only does every bit of hardened logic use some die area, but it also needs extensive test suites and simulations to be developed and ran which can take a lot of time. More hardened logic also adds risk, since if there is a bug, it cannot be changed. Therefore, if possible, it can be advantageous to construct the interface such that it is very flexible, at a low level and closely tied to hardware function. Microcoding is an example of these techniques where instructions can be redefined to workaround silicon bugs or to correct instruction implementation.

I suspect that if nvidia properly documented their interface, it would therefore give away key parameters of their chips as well as insights into what they harden and what they leave to the CPU. Things such as pipeline steps, interlocks (if any and where) and even the number of different hardware blocks they put on their die may become evident.

All of this benefits the competition greatly, and in an industry where designing and producing these high specification chips costs tens of millions of dollars, I fully appreciate why they hold their cards close to their chest.

Finally, what I don't understand is why Slashdot readers think they should open up the specs or drivers. Nvidia make and sell their cards and can do what they want while respecting law and licenses. They could very easily provide no driver for desktop Linux at all, and given the effort required to support Linux in all it's flavours and variants with frequent new kernel releases and ever changing APIs, I'm sometimes surprised they bother.

Re:Ugh, this makes me mad.

[bhcompy]

And GPL isn't the only license out there

Re:Ugh, this makes me mad.

[David Gerard]

This is a common myth in the geek world: that corporations are not merely incentivised, but actually required, to fuck over whoever they can in the pursuit of whatever arbitrary short-term buck someone on the internet is discussing at any given monent.

This claim is popular but incorrect - under the business judgment rule, directors' decisions cannot be judicially challenged unless they are grossly self-aggrandising or actually provably insane. The second standard is never met unless there is no deliberative process whatsoever.

The original source of this claim appears to be Cryptonomicon, where a shareholder doing this to a startup is a plot point. They can bring a suit for this, but only in the trivial sense that someone can bring a suit for anything. But as for being required by law to fuck people over in the short term no matter the long-term sensible consequences? It turns out reality isn't fiction, and that's why fiction is called "fiction".

So no, judgement on how much to interact with an open source project is entirely within the company's remit to decide. As would be sanely expected.

Re:Ugh, this makes me mad.

[GNUALMAFUERTE]

But it's the kernel license, so if they plan on releasing any useful code, it must be under a GPLv2 compatible license.

I really don't understand why they don't open up already. Their only real competition is ATI, and there's not a single technology that nvidia could expose through their source that ATI doesn't already have ... so, why don't just open up?

The way nvidia gets itself into the kernel is an ugly hack, and we're sick of it and the problems it brings.

Re:Ugh, this makes me mad.

[David Gerard]

Huh, didn't know the Franken one. That's from 2008 - perhaps he read Cryptonomicon too and mistaken fiction for fact. (And here I was thinking Franken was smart.)

Re:Ugh, this makes me mad.

[Instine]

Agreed. They should help this project. While the opinions of readers here clearly matters to *some* degree, I'd like to think NVIDIA would care about my take. As I'm right now designing and building a GPGPU compute system. And while it won't get in the Linpack this year, it genuinely might in a few years from now. Right now (I broke away to write this) I'm pawing over the OpenCL vs CUDA options AGAIN, to weigh up which to go for. Right now OpenCL is pulling me strongest, simply because I know I could move to something else if I had to. I don't want to. NVIDIA apear to be leading the race to me, for stable, enterprise ready, server standard GPUs. I prefer CUDA.

So guys, I prefer your products, I could spend hundreds of thousands, maybe million some day, but I'm nervous enough to choose a lesser option in order to leave myself an escape path - purely because your lack of openness at the interface with the hardware worries me. I don't want the hardware opened, I want its API. The lowest level software interface to it (and preferably the code to your supported drivers). If you opened up I'd know I could always get a driver that worked if I needed to move OS. Or that I could tweek for performance in the area I need it. etc... Open up. Why not? I've given you a bloody good reason to. Haven't I?

Re:Ugh, this makes me mad.

[TheRaven64]

I've worked on GPU drivers, and while much of what you say is correct, you seem to be conflating 'open source developer' with 'some guy working in his spare time'. When I was hacking on GPU drivers, it was for a company that sold a compiler targeting them for HPC. There are other companies with a vested interest in improving GPU drivers, for example those selling big GPU clusters, and they can employ competent people to hack on the drivers. There are also a few small companies that specialise in writing graphics drivers who are happy to take on contracts to write drivers for new cards (GPU makers often subcontract out some of this work, especially in the embedded space).

Re:Ugh, this makes me mad.

[TheRaven64]

Truth is soon Intel hardware with the open source driver will be more powerful than the NVidia card with the open source driver.

There's also the problem (from nVidia's perspective) of commoditisation of the market. Most consumers don't want the fastest GPU, they want the cheapest GPU that's fast enough for their needs. As with CPUs, some people don't need the latest and greatest, and this percentage of the market grows over time. As the niche of people needing the high end cards shrinks, so does nVidia's R&D budget and their ability to stay on the leading edge. This is something nVidia should be intimately familiar with, as it was how they killed SGI...

Re:Ugh, this makes me mad.

[Anonymous Coward]

I doubt anyone will even see this, but I think it's great that they took any interest at all.

The owe our tiny market segment exactly nothing, but they're asking anyway. That's certainly something and it seems awfully nasty to get all "fuck you, it's all or nothing" about it.

Just a thought.

Re:Ugh, this makes me mad.

[SplashMyBandit]

> The pilot project was worth an initial 250-350 million dollars, with the potential of much more to follow.
People always say this, "more work to follow" but it hardly ever works out like that. The Chinese are notorious (as in *consistently and repeatedly*) ask for samples with the promise of more orders and they take it apart and reverse engineer it. They do this with electronics, shoes, ships, MiG and Sukhoi jets, missiles, commercial agricultural strains even Austrian towns. I'm not anti-Sino, just reporting what has been documented as happening and appears to be a sustained, systematic and government condoned effort to make unlicensed copies of technology (that is, 'steal'). Perhaps NVidia were aware of this, even if many Slashdotters are not.

Re:Ugh, this makes me mad.

[Bert64]

Depends how the code is written... There shouldn't be a lot of architecture specific code in there.
To put it another way, i have used linux on various architectures, often using combinations of pci cards that i doubt anyone ever expected to be used together... I used to use sun ethernet cards in my alphastation for instance. I also used an (older, pci based) radeon card in that alphastation.
Similarly, i have used some old pci nics on a modern amd64 box, microsoft has 32bit drivers for these but didn't bother to compile 64bit versions so i doubt there are many people trying to use a similar setup to me.

Re:Ugh, this makes me mad.

[Bert64]

You could say that linux is significantly invested in x86 too, and yet it also has support for mips, arm, etc...
If the driver is even remotely well written, then it should require absolutely minimal effort to support other architectures. There are plenty of drivers for pci and usb devices in the linux kernel that work on other architectures pretty much by accident, try building a sparc or alpha linux box and throwing some random pci cards in.
And let's not forget that until a few years ago, there were nvidia drivers for macos on powerpc.

Re:Ugh, this makes me mad.

[GNious]

I would think that the Nouveau project would prefer to keep nVidia people/developers out of the source - they cannot risk there being any "polution" of intellectual property.
Lawyers should first classify what the company can divulge (and their answer: Nothing), before anyone from nVidia even thinks about looking at the Nouveau source.

Re:Ugh, this makes me mad.

[Tough Love]

NVidia should open their register specs like AMD or else fuck themselves. I believe that is what Linus meant.

Re:Ugh, this makes me mad.

[Tough Love]

that's why my next computers will have Intel GPUs

I'm favorably impressed with Intel's latest GPU efforts, but cheapo Radeon cards and embedded still clean their clock and obviously high end cards are completely in another league.

Re:Ugh, this makes me mad.

[Tough Love]

Either NV hands over the goods, or we have nothing

Crawl back into your cesspool, idiot. Like GP said, we need register specs. That's it. NVidia no doubt has a lot of years invested in their driver, but they can keep it. We have way more man years available than they do.

Re:Ugh, this makes me mad.

[Tough Love]

you wont ever be in a position where you have the required brain capacity to write a graphics driver

I am, and you're a moron.

Re:Ugh, this makes me mad.

[Tough Love]

The big deal is the army of patent trolls who will jump out of the bushes with baseball bats the minute NVIDIA or any other major GPU manufacturer publishes a full set of register-level specifications.

What a load of hooey. If it was worth anything they would have decompiled the NVidia drivers long ago. I don't know about you, but that would take me 5 minutes. Troll.

Re:Ugh, this makes me mad.

[philip.paradis]

I'd be delighted to see you back up any of your assertions with deep insights into my personal history, the histories of other interested parties on this topic, actual refutation of any of the listed points instead of generalized hand waving, demonstrated understanding of how practical open source / free software development works, and additional relevant citations.

Re:Ugh, this makes me mad.

[Anonymous Coward]

Companies don't like giving away the goods for fear of losing competitive advantages. Companies that invest a lot in their goods as NVIDIA does are especially fearful of creating new competition. This isn't a competition is bad argument, it's a giving away your tech and enriching your competitors is bad argument. Also, NVIDIA doesn't have to open it up to Linux. Seriously, what will the Linux community do? Threaten to get behind another chipmaker? Maybe a full scale boycott of NVIDIA? NVIDIA doesn't get much benefit from the Linux community.

Re:Ugh, this makes me mad.

[Anonymous Coward]

HW specs

would give away the competitive design of the underlying architecture blocks.

And having read H&P's Computer Architecture, my guess is that nV, ati/amd do not virtualize instructions like intel does
with their cisc -> micro-OP. I guess there is evon no such thing as a decoder that converts ia32/64 or amd64 instructions
to internal risc instructions.

You blame nV for being secretive. Try asking intel for their specs of tracecache or hyperthreading.
Ask them about their branch predictor. See what they will come up with.

The main reason that they are so reluctant to give away details is that the GPU arch is not open.
The architecture is not standardized like for instance CPU arch is.
amd and intel do have agreement which is guarded by corresponding cross-IP licensing agreement.

Plus,
it is all very competitive on GPU market. It is not like on CPU market, clearly segmented,
where you have intel trouncing amd on cosumer market, IBM busy with their super-computing-ass-kicking POWER chips
and sun/oracle is busy buiding on niagara for enterprise adding billions of cache transistors.

So, I would assume that giving away details of your hw spec would mean you give away months/years of RD to anybody willing
to look at your specs and figure out the organization of this or that pipeline engine. Think about it.
You know what ? nV execs are probably very worried with intel aka chipzilla now integrating video solutions onto their CPU.
It was a small relief for them when intel failed with larabee. Guess what, with that "failure", intel built up serious GPU know-how.
With their process development experience and their sheer might, they will easily turn nV( or name other GPU vendors) obsolete
in couple of years, at will.

Talk about marketplace.
Since general purpose computing appealed much broader client base, the marketplace for general purposes computing is much bigger
than GPU market has ever been. It is a niche product for niche market. Their main targeted consumers are the victims of gaming
industry - like you and me(all of us who like playing videogames).
Economics kicks in here - the volume of produced CPUs has benefits to CPU manufacturers, like better learning curve, resources, and
leverage. You know, without a CPU there is no computer, but I can do without nV/ati chip as long as I have some kind of graphics.
Mind you, average person doesn't need to run crysis, rather they are much more interested in general computation.
Because of the market size you get so many more players on CPU market. Besides intel/amd you get guys, with all sorts of cpu
architectures: sun/oracle, ibm, fujitsu, tilera, arm and others who produce chips for specialized hw.
Compared to above the GPU vendors are rather sparse category.

And as suggested by Hennessy and Patterson, for a long time now ( since ver3 of Com.Arch ?) the time when chip makers could
increase speeds of their single core chips is gone. They can't scale performance by clock speed any more.
Also, it has been a long time when industry has relied on CPU's ability to extract ILP. There is not too much left in that direction
either. Think about it, these days, branch predictors are almost 95% accurate, and speculative execution allows schedule almost all
the chip hw resources.
Now, Amdahl's law is suggesting here that cpu manufacturers in order to extract performance are going the multi-core and parallel
computing way. No surprises really. Since there is no other way.

Today, it is the era of multi-core. Tomorrow(already we have these things) - heterogeneous/hybrids with GPUs integrated on the same
silicon chip. That is what Amdahl's law predicts. Oh and btw, there is nothing new in blending between superscalar cpus with verctor
processors. They had them in 80s or 70s. Not really new to chip makers.

Bottom line,
nV is understandable with their approach to disclosing technical information about their chips.

But that might soon become irrelevant pretty quickly. Think 5 years from now. Think chipzilla delivering 5-7Bn transistors chips
that do everything you need your computer to do. I don't see a place where I need nV.

Re:Ugh, this makes me mad.

[RabidReindeer]

"You seem to care more about NVIDIA's image than about what the Linux community actually needs."

Of course they do. Why in the world would this surprise you? It's a public, for-profit company. The vast majority of contributions to open source by companies are not done out of altruism. They're done out of self-interest.

"I truly don't understand what the big deal is. Just open up your damn specifications already."

There's probably something in how they do things that they think (rightly or wrongly) that gives them an advantage over their competition that either isn't patented or can't be patented (or maybe violates someone else's patents).

One of the very best ways to obtain a good image is to provide what your customers actually need. Or at least what they think they need. And in the long run, it's usually going to be a darn sight cheaper than spending a lot of money on PR types to repair damage that never should have happened to begin with.

Especially when it degenerates to the point where one of the most respected software developers in the world is publicly photographed aiming a rude gesture at you.

The current NVIDIA driver generates all sorts of annoying artifacts on my screen when ordinary web pages scroll. The only reason that I even buy NVIDIA hardware is that their competition is even less supportive. But the lesser of two evils is still evil, and I'd pay good money for something that wasn't evil.

Re:Ugh, this makes me mad.

[treadmarks]

I don't think NVIDIA is doing anything wrong, but I'm sick of hearing the profit motive used as a euphemism for antisocial or greedy behavior. That's not the way capitalism is supposed to work and everybody needs to know this. Any business textbook will tell you that the profit motive is not meant to be synonymous with greed, but in practice that is how most Americans think of it. I don't know how people got tricked or corrupted into this way of thinking but we're all paying the price now as fraud played a huge part in the economy's destruction.

Re:Ugh, this makes me mad.

[Rich0]

True, and I'm not sure how much profit there is in $250M worth of purchases compared to the up-front costs/etc.

I was talking to a guy in a defense contractor, and they were looking to develop some kind of smartphone for the US armed forces. They went to some phone vendor and told them that they could guarantee them a few million phones worth of business. The phone vendor explained to them that they sold a few million phones every day.

Gotta keep perspective. That said, would be nice if the nvidia folks mainstreamed their drivers...

Re:Ugh, this makes me mad.

[David Gerard]

Urgh. Yeah, it coulda been worse.

Re:Ugh, this makes me mad.

[Entrope]

Nvidia's distinguishing technology and products are not its driver software or the interface between software and hardware. They sell video cards, remember?

Re:Ugh, this makes me mad.

[Entrope]

High-end video drivers differ from most other drivers in that they do more "data path" type operations (especially on vectors and matrices), and as a result they tend to use architecture-specific extensions for speed. Sometimes they also dynamically generate host-side code -- for example, to handle shader operations that the hardware cannot.

Re:Ugh, this makes me mad.

[nashv]

And why shouldn't that exactly be an nVidia employee's priorities? You seem to suggest it's a negative point.

Re:Ugh, this makes me mad.

[FridayBob]

I would think that the Nouveau project would prefer to keep nVidia people/developers out of the source - they cannot risk there being any "polution" of intellectual property. ...

Not necessarily. We don't need their source code, just more information about their API. And as was recently ruled in the Oracle vs Google case regarding Java, API's can not be copyrighted.

Lawyers should first classify what the company can divulge (and their answer: Nothing), before anyone from nVidia even thinks about looking at the Nouveau source.

Not their source code -- just the API for their hardware! Not that I expect that Nvidia's management will allow their engineers to divulge any API information (even just a little), but for us that's what's most important.

Re:Ugh, this makes me mad.

[roman_mir]

You seem to care more about NVIDIA's image than about what the Linux community actually needs.

- duh! Is the "Linux community" going to PAY the company for their efforts?

Re:Ugh, this makes me mad.

[Jay Tarbox]

Actually, they don't even sell video cards. They make GPU's that go on video cards...

Re:Ugh, this makes me mad.

[smallfries]

If you can do that in five minutes then are you not being a little harsh in not providing the info to the nouveau project?

Re:Ugh, this makes me mad.

[Deorus]

Nvidia's distinguishing technology and products are not its driver software or the interface between software and hardware. They sell video cards, remember?

With a boatload of engineering in their drivers. By abstracting hardware access through their drivers they also prevent people from building on top of existing hardware and then complaining when NVIDIA changes everything. That's called encapsulation.

Re:Ugh, this makes me mad.

[drinkypoo]

This comment is an awesome rtfa-troll, you're living up to your name admirably, since the only rational response is "please stop asking nVidia to do things they are legally unable to do". nVidia is not the sole owner of "their" IP. They cannot do what you are asking them to do. They tied themselves tightly to Direct3D pretty much from the beginning of the Quadro series, and DirectX6.0b and NV2A (bits of which continued) are a matched pair.

Re:Ugh, this makes me mad.

[Deorus]

What would be the purpose assuming that they want you to use the official drivers? If they see a huge need (i.e.: money) to port the driver to another platform, they're the most qualified people to do it.

Re:Ugh, this makes me mad.

[drinkypoo]

If it was worth anything they would have decompiled the NVidia drivers long ago. I don't know about you, but that would take me 5 minutes.

Great, now you have a big ugly mess of assembler that won't tell you anything you couldn't have learned with a debugger. Now what are you going to do with that, that will actually save you time? And you haven't actually done it, but you say it wouldn't teach you anything? That is not and never has been the problem, the problem is that once you've done it you're tainted.

Troll.

God damn it, you failed at punctuation. It's not "Troll." at the end, it's "Troll:" at the beginning.

Re:Ugh, this makes me mad.

[Deorus]

Telling people which versions of Linux they can use the video cards with is equivalent to telling Windows users they all have to use Vista if they want to use NVidia hardware.

Fortunately, Vista drivers work perfectly fine on Windows 7, making that a non-issue and demonstrating that Linux itself IS the actual problem. Thanks for the Windows reference!

Re:Ugh, this makes me mad.

[Deorus]

There are plenty of drivers for pci and usb devices in the linux kernel that work on other architectures pretty much by accident, try building a sparc or alpha linux box and throwing some random pci cards in.

How many of those drivers have dynamic recompilation code in them (or even a compiler, for that matter)?

And let's not forget that until a few years ago, there were nvidia drivers for macos on powerpc.

Paid for, written, and maintained by Apple in a country where they can sue for damages should the NDA be breached.

Re:Ugh, this makes me mad.

[Electricity Likes Me]

Nonetheless, with open-source teams working on drivers for them, the problem here is not "it would require some careful thinking", the problem is "we have no idea how interfaces on the card were supposed to operate".

Re:Ugh, this makes me mad.

[jellomizer]

You probably have never worked in a high enough position in a software company.

Just releasing your specs or your source is hard.
1. Not all of the code you have is yours. Third party tools, and work done by a differen company. You are going to get all parties to agree.
2. Nvida has put a lot of money in making their equipment. Releasing the drivers may make an influx on Nvida compatible cards, that uses the Nvida driver. These companies can sell the card cheaper because they are not spending their money making the driver.
3. Imbarased to show the code. Even the best of us go back and look at your old code and wonder what the hell were you thinking.
4. Hooks for the next gen products can be in the code, and they don't want to realse what they are going to do next.

Re:Ugh, this makes me mad.

[ed1park]

The AC made a short "personal" statement and needed to be put in place. And he was. That's how it should be.

Re:Ugh, this makes me mad.

[SwashbucklingCowboy]

One of the very best ways to obtain a good image is to provide what your customers actually need.

True, but what percent of their customers need a fully functioning Linux driver? I'm betting it's less than 1%.

Re:Ugh, this makes me mad.

[swillden]

Eh? Smartphone activation numbers show that there are just over a million smartphones activated per day, combining Android and iOS (I don't know about the minor players, but they're minor). So one vendor selling a few million per day seems very unlikely.

Re:Ugh, this makes me mad.

[pipatron]

Maybe a reply to a reply to a reply to a slashdot post isn't the best way to inform the managment at NVIDIA of this situation. Try sending the same message to them directly, or to the google+ guy. Perhaps it could even open up the eyes for someone.

Re:Ugh, this makes me mad.

[Yvanhoe]

We can even point him to the exact page where Optimus developers complain about NVidia's attitude and ask for specific help

There should be ways to detect the wirings and whether there is a mux and where, but the documentation is not available to the developers (maybe you can help us figure out how to do this, have any ideas? You can also 'petition' nvidia for releasing these specs: nvidia customer help ? )

Re:Ugh, this makes me mad.

[gbjbaanb]

Telling people which versions of Linux they can use the video cards with

or guess what, we could create a stable ABI and then a single blob would work with (gosh) all versions of Linux.

The other benefit is that all other bits of hardware would work with all versions of Linux and could they start to be sold with shrink-wrapped drivers in the shops.

Re:Ugh, this makes me mad.

[Crosshair84]

My firewire cards last driver was for windows 2000 and yet it worked in windows 7 32-bit. I finally junked my last firewire peripheral so I didn't bother to see if it worked on Windows 7 64 bit.

Re:Ugh, this makes me mad.

[axlr8or]

Nuh UH!! Because remember now, the aren't happy making chips to serve customers. Oh no. That is beneath them. What they want is to retain the intellectual ambiguity so that you just have to up and say, "DAMN you guys are so SMART! I'm sending you my first born!" They love that.

Re:Ugh, this makes me mad.

[maxwell demon]

(sorry - lower case "P"s seem to be dead on this keyboard)

I wonder what sort of keyboard has different keys for uppercase and lowercase Ps.

Re:Ugh, this makes me mad.

[lordandmaker]

> You seem to care more about NVIDIA's image than about what the Linux community actually needs.

I find it hard to come up with a course of action that would improve NVIDIA's image among kernel maintainers without giving them what they need. He wants people to want to work with NVIDIA stuff which would obviously require both giving the maintainers what they need and doing stuff they they don't necessarily need but wouldn't mind at all.

Asking this question he doesn't just get suggestions of how to make it easier to run NVIDIA hardware on desktop Linux, but also on what about working with NVIDIA irritates the maintainers; fixing those means more maintainers *using* NVIDIA (and so testing it) if not specifically fixing it.

Re:Ugh, this makes me mad.

[axlr8or]

I'm sorry. You guys are way better peoples than me. But, I need a laugh here. Any time someone does this 'When I was hacking' thereby insinuating greatness. Well, let me point out, this is as Fn hilarious as my flaming narcissist boss saying that he runs this business because it really 9isn't about him. It's about you. And, I go off on my own, with a sick knot in my stomach but you have to laugh because its just so Fn hilarious. And he goes off in his new corvette comletely oblivious to how he actually got his vette. And there is no way in hell or high water you could coerce me into his self absorbed lifestyle. I needed a laugh, thanks.

Re:Ugh, this makes me mad.

[Runaway1956]

Same key of course. If I press the key without shifting, the cursor flashes, but I get no letter. Press shift, I get a capital letter. Strange . . .

Re:Ugh, this makes me mad.

[west]

Nvidia's distinguishing technology and products are not its driver software

What? There's at least as much useful (and reproducible) technology in the drivers as in the hardware. I can pretty much guarantee any open-source drivers would be gone though with a fine-tooth comb for useful information in exactly the fashion that companies purchase and disassemble their competitors products for manufacturing information.

In any half-way competitive industry, if you're competition *isn't* using your open source efforts to improve their own products, they're probably already out of business for a whole host of other blatant stupidities.

That said, in many industries, there isn't a lot of useful information for competitors in the software. Fearing to open-source most general purpose computing is usually pretty foolish. However, high-end video cards is definitely not one of those industries. A lot of money is spent on software R&D innovations and experience to invent or improve algorithms which will happily be taken by the competition to improve their own product.

Re:Ugh, this makes me mad.

[Grishnakh]

You don't need license to make a copy of an Austrian town that's hundreds of years old.

As for agriculture strains, those are natural (though changed through generations of controlled breeding). Copyright doesn't apply there. Are you trying to tell me it'd be illegal to make a clone of someone's pedigreed pet dog? Bullshit. And if it is, that's one law I think the Chinese would be fully justified in ignoring. Western countries rely far too much on this "intellectual property" bullshit, instead of actually making money by making real things.

Re:Ugh, this makes me mad.

[Grishnakh]

Yes, this is called "binning". Intel does it for CPUs too.

The thing is, just because a chip is marked for a lower speed doesn't mean that's all it's capable of; companies usually sell many more of a low-end product than a high-end product, so if a run has lots of parts good for the higher speeds, they're not going to just mark them for the high speeds and have a shortage of the low-speed parts that all the budget-conscious consumers want, and a glut of expensive parts that very few people are willing to pay for. So many parts may in fact be capable of higher speeds than they're rated for. Of course, each one is different; with two chips marked identically, one might not be overclockable much at all, and the other might be able to get significantly better performance. The marked speed is what's guaranteed, that's all, and manufacturers are always conservative with their ratings to avoid warranty problems.

Re:Ugh, this makes me mad.

[Grishnakh]

Bullshit. Intel has fully open-sourced drivers (and therefore, documentation too), and there's no patent trolls suing them over it. I'm pretty sure AMD/ATI has also opened up much of their specs, and again, no patent trolls to be found.

Re:Ugh, this makes me mad.

[hairyfeet]

But whose fault is that when Torvalds flat out refuses to allow a hardware ABI that EVERYONE ELSE has? Solaris, BSD, Windows, OSX, hell even OS/2 has a fricking ABI now and the ONLY OS that doesn't is Linux. So why should Nvidia or anybody else jump through hoops to support a broken design?

You can use Win2K drivers on XP, you can use Vista drivers on win 7 and 8, that's over a decade of support for a driver. That means that in all likelihood the hardware will be long dead before you can no longer run it on your choice of Windows OS.

I'm personally using a 2008 Vista driver on my fully updated Win 7 simply because it interacts better with the software I prefer to use, yet on Linux its fully expected that driver simply won't work without a recompile or major futzing, that's just fucked up folks, and even one of the Red Hat developers says the current system simply isn't sustainable, you have too few qualified people trying to support too much and the QA and QC simply isn't there because of it.

I can tell you that as a retailer that is why I'll sell a system with XP despite XP being creaky as hell now, simply because I can't even take a 4 year old Linux and upgrade it to current without shit breaking. The entire driver subsystem in Linux is a God awful mess and if this finally gets people to talk about it, instead of just accepting fixes and forum hunts as par for the course? let me be the first to say i'm ALL for it. there is no damned reason why companies like Nvidia and AMD should have to keep releasing drivers for old hardware simply because Torvalds ego won't let him see the current system doesn't work. And PLEASE don't say it would encourage binary blobs because in case you ain't notice? they ALREADY DO THAT NOW, you just have shittier support.

So don't sweep it under the rug folks, this is simply a symptom of a MUCH larger problem, and that is the current driver model simply isn't sustainable. You can get by with it in server and embedded because hardware changes there are practically non existent and positively glacial, hell many servers still use Rage II for GPUs for the love of Pete, but to gain in the consumer market you need to be able to keep up with the constantly changing hardware and the current model simply doesn't work. Its always behind the curve, always way behind, hell by the time things are properly supported they usually aren't being sold in retail anymore. this really needs to change folks, and the only way it'll happen is if enough people get fed up with the status quo and complain. You can and DO deserve better fellas.

Re:Ugh, this makes me mad.

[lister king of smeg]

The owe our tiny market segment exactly nothing,

not so tiny our market share is only small in the desktop market in embeded (tv's blueray players streaming boxes etc) server, mainframes, super
computers, mobile, and pretty much everything else we rule the market.

Re:Ugh, this makes me mad.

[Anonymous Coward]

Yes. What stops companies to keep only the latest supersecret optimizations in the proprietary driver while letting the open version achieve 85% of performance or likely more? The fact that open source drivers make hardware "eternal".
That's natural. Evolution went from the hayes compatible modem to awful winmodems because planned obsolescence is possible with the latter only.

Re:Ugh, this makes me mad.

[crizh]

I cannot provide links but will dig out the relevant textbooks when I get home.

In the UK this absolutely is the case. Only 15% of a companies shareholders are able to drag a company to court and compel them to behave in a manner which is more profitable for them.

In practice this means that you need to control 86% of the stock to be able to compel a Company Limited by Shares to behave altruistically. As a Public Limited Company must trade a minimum of 20% of its stock on a recognised exchange it is impossible to retain total control of how your company behaves once it has been floated.

I'm also fairly certain that a board of directors could be compelled to perform any profitable activity that has not been clearly established as illegal by a court of law by only 15% of shareholders.

It might go even further than that but I am unsure to what extent the Proceeds of Crime Act applies to corporate entities.

If, for example, The Times could make £10M by reneging on a contract worth £1M Rupert Murdoch would be well within his rights to insist that they do so and, if it came to it, a Court would compel them to obey.

This is why I dislike Insurance Companies and no longer work for them. They literally exist to screw their customers as much as they possibly can without driving them to their equally corrupt competition.

£ = Sterling btw, dunno what's going on there...

Re:Ugh, this makes me mad.

[mrchaotica]

As for agriculture strains, those are natural (though changed through generations of controlled breeding). Copyright doesn't apply there. Are you trying to tell me it'd be illegal to make a clone of someone's pedigreed pet dog? Bullshit.

You should try explaining your point of view to Monsanto some time...

Re:Ugh, this makes me mad.

[DamnStupidElf]

Just to feed the troll: The REnouveau team already has nVidia's code (minus comments). That is not sufficient to write a high quality driver because the code contains a subset of the hardware interface specification.

Re:Ugh, this makes me mad.

[Tough Love]

I think you kind of missed the point. Whoosh. The GP trotted out the tired old claim that NVidia doesn't open the source because it would reveal patent infringement. Which is a load of hooey because any interested patent owner can trivially decompile it.

And sorry, if it looks like a big ugly mess of assembler to you, that's your problem.

Re:Ugh, this makes me mad.

[Grishnakh]

Well obviously I don't subscribe to patented genes either, but we're not even talking about GMO here, we're talking about traits acquired through selective breeding. Selective breeding has been done for millennia now; that's how we have dogs and cows and sheep. I've never heard of anyone trying to claim copyright on organisms produced by selective breeding.

Re:Ugh, this makes me mad.

[HiThere]

Even if there is, you'd never know it from the API, which is what is being requested.

Their position is silly, and both directly harmful to the company and harmful to their image. I'm much less likely to ask for a NVidia board in my next computer, which I'll be buying within a year or two. That doesn't mean I won't. This is just one factor that will enter into the consideration. But it decreases the probability considerably.

It's directly harmful because the drivers aren't a profit center for the company. If they can get someone else to improve their drivers, it's to their benefit. They probably don't believe that this can be done using only the the API, but I'd bet that they're wrong. Putting a wrapper around their driver will always be sub-optimal, but if you know the API you can often allow it to do things that it otherwise couldn't do.

My guess is that what they're doing is selling the same hardware with different drivers at different prices, and this is the reason they won't reveal the API, but that can be handled with fusible links or jumpers. Or even different FPLAs. (I'm NOT a hardware guy, so I'm not real sure what an FPLA is, just that it can be used in this way.) It's an old trick, and many companies use it, especially when they're circling the drain. Probably because it's favored by accountants, and when accountants take control of a tech company it usually dies soon after.

Re:Ugh, this makes me mad.

[cynyr]

And what happens when that ABI needs to be changed/updated to handle some new something?

Right, the only real way to make this work is what nvidia is doing, or going full opensource. Full opensource is probably not an option as they have likely licensed something from someone else.

Re:Ugh, this makes me mad.

[tincho_uy]

Well, the market they care about in this case is desktops, so all other niches are not relevant to the discussion.

Re:Ugh, this makes me mad.

[lister king of smeg]

because no one could ever want to use a gpu in a computational cluster oh wait.

Re:Ugh, this makes me mad.

[SplashMyBandit]

"As for agriculture strains, those are natural (though changed through generations of controlled breeding). Copyright doesn't apply there."
I live in New Zealand (NZ). Millions of taxpayer dollars were spent developing a strain of apples known as New Zealand Rose. It is a delicious apple. However, Chinese tourists were found with cuttings of the apple trees (they're always found with such things) and now the apples are grown in China - destroying some of the export potential of the apple, plus also destroying the brand in the quality control that NZ maintain for their product (since the Chinese can sell inferior quality apples it taints the NZ brand).

This is not unusual, in fact the Chinese Government has a department that asks its citizens going overseas to bring back anything useful they can find and they'll be rewarded. This is called the 'Thousand grains of sand' and is an interesting approach to spying (who cares if some tourists are arrested, since there are many more you can still get interesting stuff). It is not ok for it to be government policy to conduct industrial espionage - the Chinese joined the WTO (in 2001) which means they agreed to obey international property rights agreements. Now they are not the only ones, but they certainly are the greatest and more importantly they are *deliberate and systematic* offenders. Therefore, while your post has some good points I don't think it was made understanding the true 'state of play' at the moment.

Re:Ugh, this makes me mad.

[Sir_Sri]

All the cards have firmware on them, and yes, you can flash some cards firmware and get a better card when they down bin them in the firmware, but the firmware isn't the driver.

Even then, it's pretty rare that you get a much better card down tiered to a terrible card, it's a much smaller slide than that.

Re:Ugh, this makes me mad.

[gbjbaanb]

how often does an ABI of something as mature as Linux need to be changed? Microsoft manages quite well, and only changed their driver model when the big media firms persuaded them to put DRM into it for Vista. I should imagine stabilising the Linux ABI today would not change for the next 20 years.

Re:Ugh, this makes me mad.

[roman_mir]

Right, they bought the cards, but they knew what they were buying and the company sells what it sells. You want the company to sell something different and you think there is money in it?

The company doesn't care what OS you are running, but it knows that it must provide good support for the most common types of OS in existence - various Windows platforms, all other platforms are probably irrelevant in terms of total market share, so the company doesn't cater to those specific types of customers that much.

Re:Ugh, this makes me mad.

[LordLimecat]

You seem to care more about NVIDIA's image than about what the Linux community actually needs.

You know whats worse? I hear that Microsoft cares about their own bottom line more than they care about helping the BSD community.

THOSE JERKS!

Re:Ugh, this makes me mad.

[drinkypoo]

The GP trotted out the tired old claim that NVidia doesn't open the source because it would reveal patent infringement. Which is a load of hooey because any interested patent owner can trivially decompile it.

Decompiling it will tell you what the driver is doing, but not why.

And sorry, if it looks like a big ugly mess of assembler to you, that's your problem.

It's going to be a problem for anyone who is actually trying to write a driver that takes into account what is going on inside of the GPU. If seeing what the driver is doing were all it took then we'd have great Free drivers by now.

Re:Ugh, this makes me mad.

[rtfa-troll]

This comment is an awesome rtfa-troll, you're living up to your name admirably,

Thanks. Like all Slashdot posters, I am here only to please.

since the only rational response is "please stop asking nVidia to do things they are legally unable to do". nVidia is not the sole owner of "their" IP. They cannot do what you are asking them to do. They tied themselves tightly to Direct3D pretty much from the beginning of the Quadro series, and DirectX6.0b and NV2A (bits of which continued) are a matched pair.

To simply document register names etc. does not document how things work. That is not IP. The recent Oracle vs. Google case says clearly that APIs are not protected matter and I see no reason why the same does not apply to hardware binary interfaces. NVIDIA could also just provide alternate interfaces to (at least some of) the same functionality. NVIDIA could also simply start by providing the 2D and basic registers and avoiding the Direc3D related stuff.If some idiot in NVIDIA signed a contract which excludes them from giving any help whatsoever to Linux, then visibly and cearly firing that person would go a long way to showing willingness.

There is no way that NVIDIA could be stuck and unable to do anyhthing at all helpful. And yet what they do is nothing.

Re:Ugh, this makes me mad.

[west]

The choice is yours!

Um, the choice isn't theirs (the engineers) - it's the company's.

Moreover, just about everything you've posted there are reasons to NOT give out the source code!

(I apologize if this is meant to be humorous - if so, I'm missing it.)

when I see a concept in action in actual hardware ..I could figure anything out.

Allowing you to produce better products that will help make the original company's products less attractive.

These people, having gained lots of knowledge and insight from their experimentation and other related studies, would be much better positioned and more confident about going into the business of creating their own state-of-the-art video cards.

Come on, this has to be humor. You're saying companies should release the source code to help others start new businesses that will destroy them?

I think the video card market could really use some more competitors at this juncture.

Thus with more competition profit margins decline even further, helping enrage the stock owners. Are you sure you're promoting open-source?

If NVidia and ATI/AMD truly are confident in having the best engineers, processes, etc, they will surely continue to dominate this market

Companies *aren't* certain they'll continue to dominate (at least the ones that last - there's nothing as doomed as a company that is certain they'll retain their leadership status). You're basically saying idiots will release source code.

the customers would get better and cheaper accelerators in the bargain.

Um, that's likely *bad* for the companies trying to make money selling video cards...

There needs to be new competitors entering the market, and there certainly will be sooner or later.

and again...

ATI/Nvidia may have the upper hand at this present moment, but other technological advances, computing in general, simulation and manufacturing capabilities, etc, have caused us to reach a point where much lower end players can now enter the graphics accelerator market and compete successfully.

So the big two should help to speed in their own destruction?

Which of you wants to be the inspiration to future graphics engineers, who will be influenced by your designs and your way of doing things

You don't want to influence you competitors to be more like you, you want to be unique so that customers flock to you alone!

Again, my apologies if this was meant to be humorous.

Re:Ugh, this makes me mad.

[gmack]

Companies don't like giving away the goods for fear of losing competitive advantages. Companies that invest a lot in their goods as NVIDIA does are especially fearful of creating new competition. This isn't a competition is bad argument, it's a giving away your tech and enriching your competitors is bad argument. Also, NVIDIA doesn't have to open it up to Linux. Seriously, what will the Linux community do? Threaten to get behind another chipmaker? Maybe a full scale boycott of NVIDIA? NVIDIA doesn't get much benefit from the Linux community.

First off, NVIDIA is the last of the hold outs, almost everyone else is playing ball on the open source front.

Secondly: All of the world is not the PC market. NVIDIA wants their GPUs to be used on the HPC market. That's mostly Linux. NVIDIA is releasing cellphone chipsets and a lot of that is Linux(android). So yes, maybe PC users on Linux are something they can ignore but they need us for pretty much every growth market they want to be on so we do matter.

Re:Ugh, this makes me mad.

[Zaelath]

Nice rant, but this would be the antithesis of how Linux has grown.

Yes, you have issues with backward compatibility, but it also means things like not being stuck with a 640Kb memory limit for 10+ years.

I did particularly enjoy how you held out that you run a 4 year old video driver as a strength of Windows and and (nVidia's?) Windows drivers, when in fact it's likely it "interacts better with the software you perfer to use" because it's broken in unique ways that software worked around and now updating the driver would break the fixes. OMG, wait, that's __exactly__ what you're complaining about in Linux.

BTW, referring to the rest of us as folks twice makes you sound like a politician trying to sell us reintroducing prima nocta as a "return to the good old days".

Re:Ugh, this makes me mad.

[shiftless]

(Actually, I think you've not even compiled a kernel yourself so do us all a favor and hold your opinions, child.)

Enjoy fading into obscurity while the world marches on without you and your precious kernel, pops.

Re:Ugh, this makes me mad.

[shiftless]

How many of those drivers have dynamic recompilation code in them (or even a compiler, for that matter)?

What about NetBSD drivers, which run in userspace and are easily portable between architectures almost seamlessly? There is more than one way to skin a cat, and the Linux developers are being really closed minded with their insistence on changing up the driver ABI all the time. PC technology is certainly mature enough that they can settle on a standard ABI and stick to it, if only there were the political will to do so. Linux's culture is its own worst enemy.

Re:Ugh, this makes me mad.

[hairyfeet]

I take it you've never been south of the Mason/Dixon line friend? That's just how we talk and unless i take the time to concentrate over it, which why the fuck would i care on a forum, I WILL be slipping the occasional Y'all, ain't, and folks because that is simply common speech here, just as its common to have a checkout girl tell you to have a blessed day or everyone in line talking it up about the weather, that's just the way we do thing here.

And surely the BEST argument you can come up with isn't 30 damned years old, yes? the 640K lasted so many years because frankly RAM cost more than a fricking car and every damned thing went bare metal. it didn't have a damned thing to do with Windows or DOS, early OS/2 didn't play nicely with large memory and most programs would positively freak, even on DR DOS or GEM if you tried to go above 640k.

Finally when did I say anything about it being a video driver? I have an older USB capture card and I prefer the older version of the video software that came with it over the newer one, but the older one doesn't like the new driver. luckily since I'm on Windows and not Linux I actually have that choice, whereas you DO NOT. You can either spend days futzing, live with an old kernel, or stay bleeding edge, that's the only "choice" the broken linux driver model gives you. Look at how the forums were filled last week with those that couldn't use their Radeon HD3200 IGPs in Linux because AMD doesn't support hardware that old, look at how Nvidia STILL has to put out drivers for the 6200 series, even though that hasn't been sold in years.

So don't blame anyone but yourselves for the lousy support, the lousy adoption numbers, the broken driver model, or the frankly piss poor upgrade system, because the community won't call Torvalds to the carpet so he gets away with giving you crap. remember a free shit sandwich is still shit, and you will NEVER get the majority, or even a decent sized minority, to accept CLI fixes and forum hunts as par for the course. i could wallpaper this post with page after page of craptastic QA and QC simply because there aren't enough devs to do the work, but read the post by the Red hat dev i linked to or this post with over 100 links of broken shit in Linux.

So don't blame anyone else when your numbers don't get any better, don't blame anyone else when MSFT locks down the bootloader and nobody cares, it'll be the lack of action of the community and their willingness to accept half assed as status quo that will doom any chance Linux had on the desktop.

Re:Ugh, this makes me mad.

[Tough Love]

I wasn't talking about people writing drivers. We certainly do not do that by decompiling code, though running the driver to see what it outputs to the card is fair game.

I was talking about the highly theoretical possibility, which to put it bluntly, I regard as a pile of hooey, that NVidia refuses to open their hardware because the driver is doing something for which they might have to pay patent fees. Did I say hooey? I'll say it again. Hooey. Whoever advanced that theory, please do not insult my intelligence. Shipping binary drivers in no way protects against discovery of patent infringement.

My personal theory about why NVidia does not open its hardware specs is that they simply have their corporate head shoved so far up their collective ass it goes around twice.

Re:Ugh, this makes me mad.

[DarwinSurvivor]

The ABI doesn't just have to change with the kernel, it also needs to change as new hardware comes out. For instance, if someone invented a hologram projector, the ABI would have to be updated to handle the new protocol that would be pushed over the new hologram-out plug on the new video cards.

Re:Ugh, this makes me mad.

[DarwinSurvivor]

OSX hasn't run on anything but x86 and AMD64 since they adopted the name "OSX". iOS does, but I doubt there's nearly as much in common between those two as most people think.

Re:Ugh, this makes me mad.

[garyebickford]

I don't know about any statutes, but there is a long standing body of civil litigation regarding the fiduciary responsibility of corporate management, primarily to shareholders but also to others, to a lesser extent. Companies are sued very often for failing to maximize shareholder value either in the short term (such as agreeing to sell the company at too low a price) or longer term (following ineffective strategies that result in a drop, or even too small an increase, in share price.)

So while there may be no 'law' on the books, there is plenty of legal muscle behind it. As one SCOTUS justice put it over 100 years ago, (paraphrasing another common sentiment from, IIRC, one of the signers of the Declaration of Independence), the corporate management should pledge its wealth, its life and its holy honor to protect the shareholders.

I personally think that the managementhave a wider responsibility as well (perhaps not as strongly) to the other stakeholders - workers, community, environment, etc. But bottom line - any organization (whether for-profit or not-for-profit, or even governmental) has to 'make a profit' (even if it's only 1c), as if an organization loses money each year it can not survive to do whatever its mission is. If a policeperson is not paid enough to eat to sustain him/herself and starves to death, how much policework can he/she do?

Re:Ugh, this makes me mad.

[Hamsterdan]

Give them time. Their new Ivy Bridge GPUs are a big improvement over Sandy Bridge. Probably enough for AMD to be a little worried.

Right now an AMD APU will wipe the floor in graphics, but intel could probably fix it if they wanted to. Of course integrated graphics will *never* beat a discrete card, but as long as it's enough for most people, it won't matter.

For most people buying OEM boxes the integrated GPU won't break a sweat decoding HD movies and playing little games. A 5-600$ Dell will blow the doors off a 4-5 year machine even on the graphics side unless the old machine has a high end card.

Re:Ugh, this makes me mad.

[Raenex]

The owe our tiny market segment exactly nothing, but they're asking anyway.

Regarding Linux on the desktop, it's in their self-interest to make a token effort (good will among influential geeks, and a hedge in case the Linux desktop ever takes off), but since Linux is one of the core pieces of Android, it goes way beyond that now.

Re:Ugh, this makes me mad.

[Man On Pink Corner]

Which is a load of hooey because any interested patent owner can trivially decompile it. And sorry, if it looks like a big ugly mess of assembler to you, that's your problem.

You know how I can tell you wouldn't know an XOR from your own ASS?

Re:Ugh, this makes me mad.

[Anonymous Coward]

You can and DO deserve better fellas.

I have no fucking idea who you're talking to. The problems I've experienced are few and far between, and when I do I have absolutely NO problem finding the solution on my own. This is slashdot -- I know it's been cunted up quite a bit in the past couple of years, but most of us are still cut from this cloth and we can see through the bullshit you spout and straight to the heart of the real issue: the quality of OSS drivers has less to do with a missing ABI and more to do with resource allocation.

If a corporation like AMD can pay developers to release driver updates on a MONTHLY basis, I don't think development is the problem. Hitting a moving API is CAKE compared to fixing obscure bug Y exposed by game X of 1 billion on Windows. It's clearly a matter of resource allocation; you spend money where you make money and that's the main reason OSS drivers are behind. This has been the case _forever_.

WDM and Windows drivers in general aren't some miraculous panacea; Windows drivers can (and do) suck and when they're closed you're basically operating on the whim of the hardware manufacturer. This is why OSS is great -- I don't have to wait for my issue to become "worth it" to fix, because me, and a shitload of other geeks that are much smarter than me can dig in and get shit moving immediately. Sorry, not giving that up mate, because I deserve better.

Re:Ugh, this makes me mad.

[deek]

Tiny market segment? Well, that depends on whether you think more than 10 million GPUs is tiny. It could actually be in their best interest to open the programming specs for their hardware. Stranger things have happened.

Re:Ugh, this makes me mad.

[diego.viola]

Fuck stable kernel API.

Re:Ugh, this makes me mad.

[epyT-R]

In the case of the nvidia driver, the closed source nature of it prevents debugging.. that's why the kernel 'taints' when nvidia.ko is loaded. The kernel guys don't want to be responsible for fixing bugs in code they can't see. ..and who wants to spend their time tearing apart blobs?

Also, the tools used to compile the binaries for a given distribution release can cause ABI changes too.

Re:Ugh, this makes me mad.

[diego.viola]

Stable hardware ABIs are a cancer. Thank you but I don't want my Linux to be flooded with a bunch of unstable blobs. Give us code or specifications or GTFO.

Re:Ugh, this makes me mad.

[diego.viola]

File a bug about that nouveau issue to the nouveau folks.

Re:Ugh, this makes me mad.

[epyT-R]

I can almost guarantee that the driver is heavily optimized for x86/SSE and friends at this point.. yes, they have it running on other arches but I doubt it's anywhere near as fast..

Re:Ugh, this makes me mad.

[diego.viola]

No fucking way. We want specs and docs for nouveau.

What good is the nvidia blob code if the nvidia blob code is just a port from the Windows nvidia drivers?

Re:Ugh, this makes me mad.

[diego.viola]

Stable ABI sucks, I don't want Linux to be flooded with blobs that make the system unstable.

Re:Ugh, this makes me mad.

[diego.viola]

That "patent troll" is just Nvidia' excuse.

Nvidia are pussies, period.

Re:Ugh, this makes me mad.

[Mabhatter]

The cornerstone of the issue is that they legally cannot just publish the source to their firmware. They have licenses for OTHER people's IP they have signed agreements to use. So much of what makes the "GPU" work is tied up in software run on the local CPU.

There was a point in the 80's when PCI bus was created that cards were going to have enough of their OWN firmware and smarts they could talk to any type of CPU via API.

The IBM PC wrecked that. First, IBM wanted cheaper systems. Then when Intel and Microsoft took over having cheaper parts where more of the work was done on the CPU (used up cycles) by custom Drivers (tied to the leading OS) became the base model of computing.

That was one reason Apple's PPC add-in boards cost more because they followed the cross-platform hardware specs. They had the proper guts to be used in cross-artecture systems which went extinct about the same time.

The issue with Linux is that Nvidia HAS to keep all this stuff in a BLOB for license sake. But then they have to go through more layers to obfuscate the IP used AND properly follow the license to link to the Linux Kernel.

Kernel Devs won't support because they can't touch it or open up the hood... Nvidia's engineers can't seem to get unstuck from the land of compiling the interfaces themselves... And Linux moves too fast for them to keep up versions.

What they really need ate a handful of Linux Devs they could build a Chinese Wall with. But it has to be in a manner that doesn't require NDA. Linux Kernel Devs don't sign those... You can download their BEST work in the open for free!

That is where the empass lies. Nvidia knows the "Chinese Wall" doesn't really work with guys at the level of Kernel devs... They are clever enough to figure out what you left OUT of the documentation, and that leads to legal trouble with Nvidia's IP partners.

Re:Ugh, this makes me mad.

[Mabhatter]

NO, NO, and NO

They already played that game with LGPL drivers and such in locked-down devices that nobody could touch (like the old TiVo battles).

If Devs kept an API for 5 years, then their kernel would just be support for a series of binary blobs hanging off. Nvidia would share their blob info with Intel (under NDA) and intel would share with HP (under NDA) and HP would share with Autodesk... Then Linux would only be "usable" with 90% closed modules.. Kernel devs would be working on the OS for FREE and not getting any intellectual sharing for their trouble.

Not to mention, moving to a new kernel after 5 years that does break things means all the interactions the devs have no way to debug, at that point even third-party software will rely on the closed modules more than open ones. Then the Fevs would NEVER be able to upgrade their OWN work without corporate PERMISSION. Like Microsoft does trying to make win7 run winXP programs... Only without the monopoly money to go with a crappy job.

Re:Ugh, this makes me mad.

[Mabhatter]

We have this OS... It's called Android. Linux devs washed their hands years ago.

Re:Ugh, this makes me mad.

[strikethree]

But whose fault is that when Torvalds flat out refuses to allow a hardware ABI that EVERYONE ELSE has? Solaris, BSD, Windows, OSX, hell even OS/2 has a fricking ABI now and the ONLY OS that doesn't is Linux. So why should Nvidia or anybody else jump through hoops to support a broken design?

IIRC, Linux does have an ABI. They just keep changing it around ever so slightly so that closed source drivers have to be recompiled each time a kernel update happens. They do this so it is a pain in the ass for you, the end user, to use closed source drivers.

In other words, this would not be such a problem if some kernel devs were not religious about forcing YOU to use only open source stuff.

I understand their position about wanting everything be open source, but considering the way the world works, they are only being a pain in the ass to users. They will NOT be forcing any companies into open sourcing their drivers through these shenanigans.

Welcome to the world of the end user being a political football. :)

Re:Ugh, this makes me mad.

[hobarrera]

The difference is that real documentation would help any OS run better on nvidia, while just giving pointers to the guys at nouveau would only improve linux support.

Plus, drivers designed from proper documentation tend to be more efficient and well-finished that reverse engineered drivers.

Re:Ugh, this makes me mad.

[hobarrera]

You seem to have it all wrong. There's no need for the manufacturer to even SUPPORT drivers. If they just release documentation, each OS's developers will gradually develop, and maintain drivers for their respective OSs. Period.

Re:Ugh, this makes me mad.

[rastoboy29]

It's not as tiny as you think, especially in some commercial and scientific spaces.

For example, their cards are used very heavily in Linux environments in Oil and Gas exploration, I happen to know.

Those guys buy lots of cards, and need to drive them hard.

Re:Ugh, this makes me mad.

[wed128]

Well, the market they care about in this case is desktops.

Is it? Tegra is an embedded platform...

Re:Ugh, this makes me mad.

[gbjbaanb]

nonsense, otherwise Windows would have to be re-released with a new ABI everytime too. If MS can have a stable binary driver model, there's nothing stopping Linux from doing the same.

Re:Ugh, this makes me mad.

[RabidReindeer]

One of the very best ways to obtain a good image is to provide what your customers actually need.

True, but what percent of their customers need a fully functioning Linux driver? I'm betting it's less than 1%.

I think it's an advertising adage that one dissatisfied customer has the leverage of 10 satisfied ones when it comes to spreading the word. And that's assuming that the dissatisfied customer is just an ordinary person. Piss off people who make decisions or who have influence (like Linus), and the damage can become far greater.

I'd also very much doubt that the number of people who at least think they need a fully functioning Linux driver is anywhere near that low these days.

Re:Ugh, this makes me mad.

[HiThere]

Indeed, the firmware isn't the driver. But I was talking about something a bit more basic. Jumpers or fusible links can change the hardware connections permanently. FPLAs used to be able to, but maybe now they're rewritable. It's good practice if some of the production run doesn't measure up to top specs to either underclock it or disable features (depending on the problem) until it does measure up, and then sell it as something that doesn't have those features. This can be done at the driver level, but it can also be done at the hardware level, for minimal expense. (Just don't connect some wires, or connect others.)

The "accountant driven" procedure that I was talking about is to sell your good cards as a lower end chip because you can't sell them all at the price you want to. You can still degrade them by not connecting some wires or connecting others (i.e., jumpers or fusible links). So my point was that that's not a valid reason for refusing to release the API. (Very few software developers are going to want to get in and muck around with internal jumpers that MAY result in their computer refusing to work properly. And fixing fusible links requires replacing a chip, which is even less likely. There'll be a few hardware types that do this, but that's basically free advertising. They're a small fraction of a small fraction of the market, so it's not sales to them that you are interested in, but what they say about you.)

As I said, to me the decision appears stupid, as well as unfriendly.

Re:Ugh, this makes me mad.

[GuB-42]

This is correct but only if said developers are available.
Writing and maintaining a driver requires time and/or money. And people, especially skilled programmers, won't do it without some kind of compensation. Sometimes it is just some computer geek/hacker that wants his hardware to work but most of the time drivers are written by employees paid by a company that expect a return on investment.
And of course, the companies that benefits the most from linux drivers are the hardware manufacturers themselves. And if they release their specs but don't contribute to the driver itself, chances is that we won't see a good driver anytime soon. Also keep in mind that spec are not perfect, there may be some undocumented features and limitations that are much easier to address if the hardware guys are just next door.

Re:Ugh, this makes me mad.

[ryanov]

The trouble is, try to stop them? I had the option on this laptop (where I did not choose the vendor) of nVidia or Intel. I've had even more trouble with Intel drivers, so where am I left?

Re:Ugh, this makes me mad.

[Lodragandraoidh]

Until a limited battery life device with limited processing power can do the following:

High performance video gaming (FPS, Combat Flight Simulations, MMORPGs etc) where FAST immersive real-time visualisations are critical (particularly for those who are competitive).
High speed simulations (Science, Virtual Servers and Networks, and Game Servers).
Software Development (compilation cycles faster on powerful multi-core systems).
Multimedia Development (3D rendering and animation etc).

There will be a need for powerful general purpose computer systems that can be personalised for a given application domain.

Now - if these capabilities are removed from the consumer market - then individuals, schools, and small businesses will not be able to compete - which would impact the level and pace of innovation in these areas, as established companies with a monopoly are not prodded to improve.

General purpose computing may become more of a niche - but it will be a large niche market that would be stupid to ignore while there is money to be made - so I don't see it disappearing.

Re:Ugh, this makes me mad.

[couchslug]

Great!

When bad companies self-damage, everyone else wins.

Re:Ugh, this makes me mad.

[hairyfeet]

I'm sorry but you are wrong, and here is why: writing, providing QA and QC and maintaining said driver is NOT a job Joe basement coder can do, in fact there are only an EXTREMELY small amount of people actually qualified to do that work. your argument might as well be "Just show us a picture of the brain and we'll develop our own brain surgeons" which as you can see would just be insanity. it takes years of education, and years of experience after that to become a good enough coder to work in such a low level environment without cocking it all up, and most of those people that ARE qualified? Well they actually already have jobs and won't work for free, sorry.

Maybe you should read this fine article by one of the RH devs that points out why that simply won't work. it hasn't worked in the past, won't work now, and will not work in the future, because in the end there are simply too few qualified to do the work and too much work that needs to be done.

Re:Ugh, this makes me mad.

[drinkypoo]

I was talking about the highly theoretical possibility, which to put it bluntly, I regard as a pile of hooey, that NVidia refuses to open their hardware because the driver is doing something for which they might have to pay patent fees.

I am talking about the highly non-theoretical fact that nVidia crawled into bed with Microsoft in order to get the contract for the Xbox GPU. nVidia shaped Direct3D in their image and Microsoft got a foot up in nVidia. It's not about patent fees, it's about retaliation from their former ally. Mission accomplished, now Microsoft can move on to poisoning the next company.

Re:Ugh, this makes me mad.

[kling0n]

I think the "tiny market segment" is wrong. Nvidia is aiming for a dominant position for mobile processors. A functional relationship with the Linux kernel community becomes increasingly important and while Linus may not be the most diplomatic of sorts, they do need to address that kind of direct telling off if they want to appear as a viable partner ofr handset manufacturers.

The very thing that they don't want to do.

[sethstorm]

Open up the hardware such that proper drivers can be written for any card (recent or not) and platform w/o the need of binary blobs.

That shouldn't be an impossible task given how much weight NVidia has towards third parties.

Re:The very thing that they don't want to do.

[Anonymous Coward]

Isn't the solution "curated APIs". They approve and distribute what developers make for custom and legacy hardware and software. Apple could learn from this too.

JJ

Re:The very thing that they don't want to do.

[Gaygirlie]

Open up the hardware such that proper drivers can be written for any card (recent or not) and platform w/o the need of binary blobs.

That shouldn't be an impossible task given how much weight NVidia has towards third parties.

People keep spouting these things without actually taking into account that NVIDIA most likely has all sorts of contracts and license agreements they just can't break. Yes, it's fun to bash NVIDIA and try to paint them as some sort of a big, black demon worth all the hatred in the world, but in the end that doesn't actually *help* anyone, least of all F/OSS movement. Intel has the advantage in that that they entered the GPU-market a lot later than NVIDIA et.al., so they could avoid many of the mistakes their competitors did and they also have the whole, complete chain needed to produce and ship GPUs all by themselves and that really makes the comparison feel more like apples and oranges than anything directly comparable.

NVIDIA could possibly open up future hardware, but they'd still possibly have to then develop some areas completely from scratch in order to not run afoul of their previous engagements and such a move obviously would cost them huge amounts of time and money. And well, Linux-users simply ain't worth that. All this is to say that NVIDIA should open up as much as they can, that I agree with, but that it is also highly unlikely they could open it all up even if they wanted to.

Re:The very thing that they don't want to do.

[Gaygirlie]

I see why you're writing as an AC.

because nothing prevents nVidia from collaborating

Copyright, patent and contract laws quite likely do. Oh, but let's not let real-life facts bother us, eh? Ignorance is a bliss and all that.

Re:The very thing that they don't want to do.

[Sir_Sri]

Intel are also not competing on high end. They're competing on simple, integrated and low power. If someone could, in the worst case, duplicate an Intel GPU it wouldn't really hurt intel. nVIDIA on the other hand...

Re:The very thing that they don't want to do.

[sethstorm]

Then how does that address multiplatform code availability/readiness? Consider that Nvidia hardware does run on non-PC hardware as well - such as PowerPC/POWER platforms as an example.

Re:The very thing that they don't want to do.

[MrLint]

Apple has ascended beyond learning. Steve taught them that 'not invented here' is a mantra to burn into your soul.

Re:The very thing that they don't want to do.

[aklinux]

I tend to agree here. I know that when we look at HP scanners, the stand alone scanners are difficult, if not impossible, to use under Linux. HP has released all kinds of stuff about the all-in-ones. I suspect this is something similar.

There was also the Philips Webcams a couple of years ago. Philips worked out something with that one former employee or contractor, don't recall which he was, to where he could release binary blobs that actually worked pretty well. The kernal people were pretty well up in arms over the situation and the hooks in the kernel for those binary blobs were coming and going for a while.

If this the case, I don't see why nViida doesn't just say so...

Re:The very thing that they don't want to do.

[Anonymous Coward]

Patents cover all software, including open source. Copyrights and contract - you mean nVidia doesn't own _ANY_ information about its _OWN_ hardware that it can share with Nouveau? Everything is contracted to someone else??? For crying out load, you must be an nVidia employee to spout such nonsense. Then, let me then repeat it until you get it, Fuck You, Nvidia, Fuck You!

Re:The very thing that they don't want to do.

[oakgrove]

Do you have any actual evidence that this is completely stopping them from collaborating in any way or are you just repeating shit you read on the internet? I can accept that those things would prevent then from opening up completely but to make out like it shuts them down all the way seems a bit much.

Re:The very thing that they don't want to do.

[jvillain]

I guess we missed the part where 100% of their technology is licensed from some one else and NOTHING is developed in house. Maybe thinking about how licensing some thing will effect your ability to document you hardware would be a good idea. A lot of patents can be easily worked around. One of the problems is every time there is a standard the hordes come from far and wide to get as many patents into it as possible to help jack up the profits. The price gets passed onto the consumer and because it is a standard there is no competition to push back against IP bloat.

Re:The very thing that they don't want to do.

[Phroggy]

nVidia is violating somebody's patents. I don't know which ones. They don't know which ones. It's really not possible for them to know which patents they're violating, and it could be that they're actually not violating anything, but it's probably a safe bet that they are. That's how broken our patent system is.

The flip side of this is, whoever holds those patents that nVidia is violating also doesn't know that nVidia is violating their patents. However, if nVidia were to release the specifications and source code that we all want them to, it might become easier for somebody to recognize that nVidia has unknowingly violated their patent, at which point nVidia might be very screwed. Or maybe not, who the hell knows? In any case, a series of patent infringement lawsuits could be very very bad, and could potentially destroy the company. The only defense nVidia has against this is 1) building up a patent warchest of their own, so that if somebody sues them, nVidia can look for patents they own that the other company is violating and work out a trade, and 2) shrouding their hardware in mystery, so nobody knows how it works and therefore nobody can figure out if it's infringing on someone's patent or not. Note that #1 is not possible against a patent troll that doesn't actually produce anything, so #2 is really all they've got.

Re:The very thing that they don't want to do.

[MichaelSmith]

Well yeah maybe. There are degrees in these things. AVR outsourced their entire developer toolchain for working on their microprocessors so that you can compile for any of their chips with gcc. Its very convenient and it sells a lot of hardware for them.

Re:The very thing that they don't want to do.

[SplashMyBandit]

"Yes, it's fun to bash NVIDIA and try to paint them as some sort of a big, black demon worth all the hatred in the world"
Lol. I suggest you go look up the definition of the Latin word "invidia". It's fits nicely with the Slashdot lynchmob.

Re:The very thing that they don't want to do.

[TheRaven64]

Intel has nothing to lose when they do it, because their graphics parts are substandard to even 6 year old parts from nVidia

I just looked up some benchmarks and the first ones I found said that running Skyrim they were about half way between Geforce GT 520 and Geforce GT 440 performance. The 520 was released about 18 months ago as an entry-level GPU and the 440 a bit earlier as a mid-level GPU. They're still a bit behind, but I think 6 years is an exaggeration.

I've just finished reading the Ivy Bridge ISA docs, and I've read the reverse-engineered nVidia docs (and worked on a compiler targeting nVidia GPUs) and Intel's design is pretty clever. The way they lay out their register set means that you will be able to generate much shorter instruction sequences for some very common operations than with nVidia hardware. Given Intel's track record, I'd imagine that their shader compiler is not yet doing this, but it will in a few driver revisions. I wouldn't be surprised if we see a fairly significant speedup in shader performance over the next 6-12 months from driver improvements in Ivy Bridge.

Re:The very thing that they don't want to do.

[Tough Love]

People keep spouting these things without actually taking into account that NVIDIA most likely has all sorts of contracts and license agreements they just can't break.

That sounds like a load of selfserving bullshit to me.

Re:The very thing that they don't want to do.

[Anonymous Coward]

There is a #3 - pay some lobbyists to get the patent system fixed. All in all, money invested in lobbying has some of the highest rates of return in the USA. Nvidia would be sure to recoup the costs of lobbying for patent reform manifold.

Re:The very thing that they don't want to do.

[zixxt]

nVidia is violating somebody's patents. I don't know which ones. They don't know which ones. It's really not possible for them to know which patents they're violating, and it could be that they're actually not violating anything, but it's probably a safe bet that they are. That's how broken our patent system is.

The flip side of this is, whoever holds those patents that nVidia is violating also doesn't know that nVidia is violating their patents. However, if nVidia were to release the specifications and source code that we all want them to, it might become easier for somebody to recognize that nVidia has unknowingly violated their patent, at which point nVidia might be very screwed. Or maybe not, who the hell knows? In any case, a series of patent infringement lawsuits could be very very bad, and could potentially destroy the company. The only defense nVidia has against this is 1) building up a patent warchest of their own, so that if somebody sues them, nVidia can look for patents they own that the other company is violating and work out a trade, and 2) shrouding their hardware in mystery, so nobody knows how it works and therefore nobody can figure out if it's infringing on someone's patent or not. Note that #1 is not possible against a patent troll that doesn't actually produce anything, so #2 is really all they've got.

Keep in mind that has not stopped AMD from releasing a ton of specs and docs and nobody has come forward over the years to sue AMD over their releases. Nvidia is just being stupid at this point and time.

Re:The very thing that they don't want to do.

[Bert64]

Compare that to all the data needed for Intel's CPU...
All the information needed to write programs to run on Intel CPUs is out there, and yet other processor manufacturers are still behind. Intel may produce lowend GPUs, but their CPUs are more than competitive on the high end.

Competitors can employ people to reverse engineer if they really want to copy... While keeping the specs closed might make copying slightly more difficult, it also makes it harder to prove.

Re:The very thing that they don't want to do.

[drinkypoo]

That sounds like a load of selfserving bullshit to me.

You mean, like everything you've said in this thread? It is a simple fact that nVidia got into bed with Microsoft in order to be the best Direct3D GPU, and also that they got further into bed with them in order to be in the Xbox. One does not simply "walk away" from Microsoft, and then go slide into bed with a Penguin. There will be consequences. I know (from your comments) that you want this all to be happening in a fantasyland where you get what you want, but we're not there, and thank goodness.

Re:The very thing that they don't want to do.

[Gaygirlie]

Do you have any actual evidence that this is completely stopping them from collaborating in any way

If you'd bother to read the comments you'd quite clearly see I never said NVIDIA cannot collaborate in any way or form. I merely responded to the Anonymous Commenter's claim that there is absolutely nothing preventing NVIDIA from collaborating which fairly obviously is incorrect. I also did comment that NVIDIA should open up as much as possible, but expecting NVIDIA to open up everything is most likely impossible for them to do, atleast for current hardware.

The thing is that NVIDIA quite possibly use licensed code in their drivers, code that they do not own copyright over, and thus they cannot contribute any such parts. Yes, it is a mere assumption on my part, but given how old a company NVIDIA is they most likely are bogged down by several decades-old contracts they themselves would like to get rid of already; Intel has the advantage here that they entered the game with enormous amounts of IC - design experience in-hand and they could glean a lot of insight and knowledge from F/OSS - projects into all kinds of things, projects that simply weren't there back when NVIDIA started. We also do not know how much of the actual in-depth design NVIDIA has licensed from some 3rd party, and they may well even be disallowed from telling us such details at all due to contracts.

As for my opinion on contributing to Nouveau -- even though you didn't ask for my opinion: I personally would prefer NVIDIA to just release proper programming specs instead for whatever parts they can. While contributing to Nouveau would benefit Linux-users, releasing proper programming specs would get the other alternative OSes up to speed much faster than if they had to study Nouveau source-code first and then try to implement it themselves.

Re:The very thing that they don't want to do.

[oakgrove]

I did read it. In fact I'll quote it for you since apparently you werent paying attention to your own weasel words.

because nothing prevents nVidia from collaborating

Copyright, patent and contract laws quite likely do.

The second part is yours and its completely unqualified. Now go ahead and play dumb.

Re:The very thing that they don't want to do.

[evilviper]

People keep spouting these things without actually taking into account that NVIDIA most likely has all sorts of contracts and license agreements they just can't break.

That was a fairly good excuse, say, 10 years ago... If they licensed this stuff on lousy terms, it doesn't stop it from being their own fault. If they haven't solved these licensing problems in over a decade of people shouting about it, they're either incompetent, or just not trying.

They had one hell of a lead back then. Now Intel and AMD are vying to be the king of Linux GPUs, and doing it with open source drivers, and soon nvidia's blob will go from being a necessary evil, to an inferior option for a host of reasons. If they don't get the hint before then, they could well forever lose the entire server and embedded systems markets, and any others where Linux becomes a dominant player.

Re:The very thing that they don't want to do.

[Tough Love]

What you suggest would be tantamount to illegal market control. Of course Microsoft would never do such a thing.

And you also suggest that NVidia is a Microsoft whore. That may be. I don't need their hardware.

Re:The very thing that they don't want to do.

[gl4ss]

no, but it is implied that by opening the specs would release enough information for some people to find out how exactly the cards achieve some things, thus opening them for patent lawsuits.
I just happen to think that's bullshit, but it's a good excuse anyways.

I just think it's more likely they don't want to release enough information for people to build quadro-drivers. and because they're embarrassed about how much they do on the cpu they say they do in the hw. there's also the possibility that as a team NVIDIA has no fucking idea how their cards work since they've outsourced and out-department so much of the development to random teams around the globe with subsystems arriving to other teams pre-compiled, subsystems which had been actually outsourced even further by the outsourcing company, so that they wouldn't have a specs bible they could just throw at nouveau and say catch.

Re:The very thing that they don't want to do.

[hackus]

I would like to point out, Nvidia has had many opportunities to put a stop to these agreements.

So I am not arguing lawful IP agreements can prevent Nvidia from cooperating, but you also have the right not to sign also and pursue a more profitable business which brings in open source customers as well as buyers into your product lines.

This cuts both ways and you can't tell me Nvidia management has had its head in the sand with regards to market penetration of LINUX, and what and why the engineering process calls for: source code.

People don't pick Windows when they need 5000 GPU's for scientific computing. Furthermore, they don't want ancient kernels with binary blobs running on very expensive dedicated hardware when advances in kernel technology happen almost weekly.

They want to have the latest, and they want it when they have engineering and administrative and engineering timelines they want to accomplish for their organizations, not what Nvidia dictates.

I mean who does Nvidia think they are? When I buy a piece of hardware, its mine, and I expect to be able to use it how I want.

This secret crap is B.S.

-Hack

Re:The very thing that they don't want to do.

[drinkypoo]

And you also suggest that NVidia is a Microsoft whore. That may be. I don't need their hardware.

I do, in fact, suggest that nVidia is a Microsoft whore. And I'd sure like to not need their hardware, but so far I have. Also, there is some hope that Nouveau will finally support NV2A and then I'll dust off my Xboxen and do some cool tricks with them.

Re:The very thing that they don't want to do.

[diego.viola]

Great comment, my thought too, I wish I could mod you up.

Re:The very thing that they don't want to do.

[rastoboy29]

Then it would be a simple matter for them to simply tell us, that that is the case.

They have not done this.

We are their customers, dude.

Nvidia Open Source Policies

[hackus]

Eventually given how global markets expand and change, Nvidia will have to change its policies.

After all, they sell a GPU.

Consider how silly it is not to publish register information so you can write software for it. It is silly. Especially if Intel did the same thing with its chipsets because AMD might get some information on how a instruction works and try and copy it therefore making assembly language illegal and silly IP laws.

Eventually this policy is going to cost Nvidia in the pocket book, and they will either come around or sink into nothingness.

The exact same direction Apple is going to be heading to very shortly.

-Hack

Re:Nvidia Open Source Policies

[Z00L00K]

By opening up the hardware platform you will also allow people to create new solutions and applications that utilizes the hardware in new interesting ways.

But one problem that Nvidia may suffer from is that the division of the Quadra and Geforce cards will suffer so that the professionals select Geforce cards instead.

As for Nvidia going to sink - it's not likely as long as they actually provide top notch hardware. If they start to lag behind in performance then they will see more problems.

Re:Nvidia Open Source Policies

[VortexCortex]

I agree. I mean... maybe not tomorrow, but soon, it won't fucking matter, there won't be a GPU, the CPU will have massive parallelism too and another CPU would take the GPUs place and power consumption. The more they close up their shite, the faster it accelerates the transition away from dedicated hardware. It's happened before, it'll all happen again.

Re:Nvidia Open Source Policies

[Anonymous Coward]

The difference between Apple and Nvidia is intent. Nvidia keeps the specifications that would allow low level programming on their products obfuscated because of licensing, for competitive advantage, and to uphold the image of distinct levels of performance by series by soft capping product features and performance. Apple does it in order to give a specifically tailored and homogenous user experience and so provides the APIs which it tunes to meet its internal performance and aesthetic criteria. Also, I don't have any Apple chipsets in my computer's generic hardware, my microcontrollers, or my other electronics, while I have an Nvidia NB chipset in a laptop and ATI chips in a video card, neither of which are properly supported under Linux. In a sense, devices containing Nvidia's chipsets are components which should integrate as seamlessly as possible into the systems they are added to, while Apple's products are meant to be complete platforms, usable and programmable without ever knowing what's inside.

Re:Nvidia Open Source Policies

[Chibi Merrow]

Eventually this policy is going to cost Nvidia in the pocket book, and they will either come around or sink into nothingness.

And yet, our new deployment of thirty Linux-based image generators I just setup all have GeForce cards in them... Do we live in the same world?

nVidia cards, with nVidia's binary blob driver, are the only way I can get any work done in Linux. That doesn't seem to be changing anytime soon, either.

I wish it would. My home PC has a Radeon card... But the support under Linux just isn't there.

Re:Nvidia Open Source Policies

[Osgeld]

its worked for the first 30 someodd years. In the past the OS makers had to pay to get hardware supported, now they just wine cause its not perfect and free... do you work for free? Do you pay other people to work for free?

Seriously whats the god honest advantage here? 0.1% of nerds will be happier until the kernel changes again in a few months ... fucking up all unpaid efforts? Its not a charity or priority.

Please get with the times, linux is dumping numbers in the user market, now? they want to have a hissy after over a decade

good luck bud
 

Re:Nvidia Open Source Policies

[Tough Love]

Consider how silly it is not to publish register information so you can write software for it. It is silly

It's stupid beyond belief. A firing offense in my opinion.

Re:Nvidia Open Source Policies

[Bert64]

High end hardware is already a niche, for the vast majority of users the integrated Intel GPU is more than adequate, while also being far better supported. Intel chips are more than adequate for general desktop use, and even playing lighter weight or older games...

AMD are in a slightly different boat, since they are able to offer cheap integrated lowend CPU/GPU combinations to compete with Intel, nvidia aren't able to do this so the only market they can go after is the high end.
And when it comes to GPUs, high end basically means hardcore gamers and those using GPUs for processing, with Linux taking a significant share of the latter.

Re:Nvidia Open Source Policies

[sir-gold]

I think that if Nvidia was willing to write better linux drivers for a price, then that price would be published by now

The problem is not not that Nvidia won't do it for free, it's that they won't do it AT ALL, for any price.

Re:Nvidia Open Source Policies

[diego.viola]

I fully agree with you.

Re:Nvidia Open Source Policies

[benthurston27]

I keep trying to think if you are right that there will be no need for the dedicated graphics card, the way the FPU moved into the CPU a while back.. What would you put in the PCI-Express socket? Did I read you right that you think it would be another CPU? I wonder what kind of specialization it would have.

Re:Nvidia Open Source Policies

[bingoUV]

"Processing" into PCI-Express (or a future variant of it) is too latency inducing. So CPU, FPU and GPU all should be part of same chip, talking to other subsystems using system-bus. Bandwidth of this system-bus would need to be expanded to accommodate higher speeds and integration of FPU.

PCI-Express and variants can be used for I/O - e.g. network and storage. Network already fits in PCI-Express, storage would also need to increasingly become low latency, so better fit into PCI-express or variants.

Re:Nvidia Open Source Policies

[Rich0]

When the kernel changes any mainline drivers are updated along with it. That's the carrot in the contribute-your-source problem. Nouveau isn't suffering from kernel changes as a result.

Nvidia is paying the price every time the kernel changes already.

So score +1 Linus Torvalds?

[Anonymous Coward]

I find your lack of faith disturbing...

A vulgar finger gesture got results.

Demand the moon, get a moon base.

Re:So score +1 Linus Torvalds?

[jvillain]

I have to agree. What ever else people want to say,the man does get results.

Re:So score +1 Linus Torvalds?

[Tough Love]

Haiku, what?

Re:So score +1 Linus Torvalds?

[ae1294]

nVidia is worried that Linus will replace their binary software with a small well written shell script...

Same thing as always

[macemoneta]

It's unnecessary, and likely impossible, for Nvidia to open source its proprietary driver, due to licensed software they don't own (they have stated in the past). All that's needed is for Nvidia to release the documentation on the components they manufacture, as AMD/ATI did in 2008 (and Intel has always done). The existing nouveau driver team will take it from there. Nvidia can also choose to provide funding, salaried developers, or sample cards for the team. That would put them in a parity position with AMD and Intel.

Re:Same thing as always

[Anonymous Coward]

The minimum nVidia needs to do is release any and all specifications required to develop source code to utilise the card. If there are parts of the card they don't hold ownership over they need to indicate which parts these are and WHO does. Then with nVidia's help we need to go after them and/or reverse engineer these parts. The idea that you can use an excuse like this to justify not releasing the majority of the code/specifications is ridicules. Intel releases the code and nVidia and AMD both could as well.

nVidia goes out of its way to explicitly reject free software. The completely disavow themselves of anything to do with free software.

Until nVidia makes a minimum of effort we should avoid nVidia. I won't buy ATI graphics cards either until AMD release sufficient specifications for free software drivers/firmware. Right now the way AMD operates is total a PR stunt. They wrap a free driver around non-free code and there is no way to operate it in a free environment (due to the non-free dependency).

Re:Same thing as always

[Anonymous Coward]

Here are the initial steps that can be taken:

1. Document the hardware. Releasing full specs would go a long way. Intel just released thousands of pages for its new stuff.
2. Have your engineers work on the nouveau project to support all of your cards in 2D.
3. Make nouveau your official linux driver for all 2D stuff. Initially this would be hard, but this is a case of if you bite the bullet, you will see rewards both short and long term.
4. We understand that the 3D stuff is where things are exciting. Move all of the 3D stuff to being a closed module of nouveau, or even to userspace.
5. Work with the virtualization people. You can do it, we know you can.
6. Work towards being able to open all of your cards. Can you work with people to set release dates for older cards? How about the ones you don't really support anymore?
7. Keep talking to the community. It does help.
8. Contribute to an open hardware project. Perhaps by agreeing to manufacture an open video board?

Re:Same thing as always

[symbolset]

NVidia doesn't even have sole ownership of their own hardware interface specifications. There will be no documenation. Ever. It is not theirs to give.

Re:Same thing as always

[Crosshair84]

All that's needed is for Nvidia to release the documentation on the components they manufacture, as AMD/ATI did in 2008 (and Intel has always done). The existing nouveau driver team will take it from there.

Yea, how well has that actually worked. Oh yea, it's failed miserably. 4 years later and ATI Linux drivers are still garbage and the only reason Intel drivers work is because of the server market and because chips don't change often. It's not rocket science, writing graphics drivers are HARD and Linus continually poking his pecker into the kernel means they continually break and have to be tinkered with to work again.

The VOIP visitation recorders at work have to have updates turned off because Ubuntu craps itself whenever it tries to update. When the box is 4 hours away that is a big deal. Our windows machines have auto-updates turned on and never have problems. We've tried having a development machine at the office to test the updates, but for whatever reason, the number of VOIP boxes and minor hardware revisions make it impossible to develop a standard update profile for us to reliably push out to all machines. So we have to just lock them down as much as we can and hope for the best. We are working on a Windows based solution so we can eliminate the Linux boxes in the future.

It needs to be repeated so it sinks in, Linux needs a reliable ABI so ATI and others can just slap a Linux driver onto the CD that comes with the card and call it a day.

Re:Same thing as always

[macemoneta]

Posting anon because I've already modded this thread - I've clearly been way too far out of the loop for way too long. Is this to say that ATI is now the Linux-friendly manufacturer whereas NVidia is not? I thought that in the past, NVidia had the lead by way of better drivers, better stability, and VDPAU. Did ATI/AMD leapfrog ahead or is NVidia still the better way to go when building a PC with the intention of running Linux on it?

ATI/AMD's New Open-Source Strategy Explained:
http://www.phoronix.com/scan.php?page=article&item=826&num=1

AMD Releases 900+ Pages Of GPU Specs:
http://www.phoronix.com/scan.php?page=news_item&px=NjA1Mw

AMD Releases Additional R600 GPU Programming Documentation:
http://www.phoronix.com/scan.php?page=article&item=960&num=1

AMD Releases 3D Programming Documentation:
http://www.phoronix.com/scan.php?page=article&item=amd_tcore_release&num=1

Re:Same thing as always

[Alex Belits]

Either you are lying, or Ubuntu installation at your work is managed by saboteurs.

Re:Same thing as always

[Z00L00K]

That assuming that Nvidia has an agreement regarding the ownership of the other parts - what if they don't and are infringing on patents and that's why they are afraid of revealing how stuff works?

Re:Same thing as always

[Anonymous Coward]

AC call's truth on his story. I've had Ubuntu installations on a variety of proprietary hardware devices pitch SHITSTORMS in upgrades.

Last year it was actually my nvidia card that caused it in fact. Years ago, it was some weird dell raid controller in the non-poweredge line...

We had to stop updating two office servers entirely as a result..

Re:Same thing as always

[bug1]

"4 years later and ATI Linux drivers are still garbage"

Seeing AMD trying is all it takes to make me a loyal customer for life.

NVIDIA dont understand their own buisness, or the basics of Free software.

NVIDIA is a HARDWARE company, not a software company, placing restrictions on software essential to operate their hardware makes it more difficult to sell their hardware.

You would think they would have worked it out by now.

Re:Same thing as always

[wrook]

4 years later and ATI Linux drivers are still garbage

Actually, I have an HD6950 in my box. I had been running the proprietary driver basically because it was set up by default and I was too lazy to change, but yesterday I decided to give it a go. In terms of performance, it is not garbage (I haven't looked at the code). There are actually quite a few advantages to the open source driver.

For one, the 2D operations seem to be significantly faster. I had to screw around with Catalyst on the proprietary driver to get good desktop performance, but the open source driver is considerably snappier out of the box. Also, Gnome Shell was crashing on me frequently with the proprietary driver (usually when doing an expose type event), but this seems to have stopped completely with the open source driver (it has also never crashed with my Intel card on my netbook). Finally video playing seems to have been improved. No matter what I did there would always be some situations where I would get tearing with the proprietary driver, but I never get tearing with the open source driver. There is probably a way to fiddle with Catalyst to get everything working well on the proprietary driver, but I could never seem to find the sweet spot in terms of performance and stability

I'm not a big gamer, but I have a few games. Some games work flawlessly. Some have reduced framerate. One game (the World Forge Ember client did not run at all due to driver problems.

Apart from Ember (which is kind of screwy most of the time anyway), every game I've tried is playable at a reasonable framerate and resolution. I suspect that hard core gamers would not be happy playing some of the more modern windows games under wine, but I don't have any of those to test. On the whole, for a casual open source gamer, the open source driver actually has a better user experience for me. Admittedly, I have a fairly high end card, so I don't know what it's like for a cheaper one, but I don't get a dramatic drop off in performance. The improvement in other areas more than makes up for it in my mind.

Quite possibly for a specific application you have in mind, it's not acceptable, but that's a far cry from "garbage".

Re:Same thing as always

[jvillain]

So I should screw my self and use a 3 year old kernel so I can use the blob, I should undo all the work the open source community has done in creating an open OS by supporting these guys because they are thieves and they just might get caught? My moral compass is swinging like I'm at the North Pole.

Re:Same thing as always

[jvillain]

Ever? Patents last how long? A LOT of the stuff they are doing has been around for a very long time. Please tell us what hardware interface specs they don't have ownership of. I am willing to bet they could easily document more than 50% which is 50+% more than they are doing currently. The problem right now is far more about just having a shitty attitude than it is about ownership of interfaces.

Re:Same thing as always

[Anonymous Coward]

And still the Nvidia drivers give the best performance compared to the other cards and open source drivers. You can argue until you turn blue, but at the end of the day this is just a fact..

If you depend on heavy graphic 3D performance nothing beats the Nvidia cards and closed source blob. I think most users are just pragmatic and use this combination. At the end of the day you have a job to do, and being up in arms about the openness of the drivers and rejecting them for that does not get this job done.

Nvidia over and over again explained they have reasons not to open their specs (you can speculate about third-party entities restrictions, but this is not a fact so far). While this is not an ideal situation it is the best we got.

Just imagine Nvidia would say: "Considering the negative reactions we decided to stop supporting Linux at all. Only manufacturers making specific soft- and hardware (like tablets and Android or programs like Maya) will get support at a small fee". Do you think that would hurt Nvidia? Do not forget the bulk of the graphic cards Nvidia sells is for the gamer market on Windows machines. While they would lose a small segment of the user market, I do not think it would it would hurt them too much. I am just wondering how long Nvidia keeps supporting Linux. I could be they are on the brink on giving up support at all. If they do that I think Linux would certainly not be helped by that, closed source blob or not...

Re:Same thing as always

[jvillain]

You can get a distro with 10 year stability if you want it. But that leaves you frozen in time. Linux has made hellacious progress compared to all other OS's because it is willing to ditch the bad and move forward. I remember one year where we were all fretting about having no journalled file system. 6 months later Linux had 5. How many can you run now? Telecom companies and others run their Linux machines with 5 nines of uptime. Maybe the problem isn't actually the ABI.

Re:Same thing as always

[arose]

If nVidia sees no benefit they needn't be doing any of it. Wait, they wouldn't be, their attitude makes it quite clear that they care preciously little for goodwill. Desktop users likely only get support as its a good bugtesting pool for their workstation and supercomputing clients. The only thing they'd accomplish by pulling out is sending developers to the FLOSS Radeon drivers, something that could turn very bad in the long term. So yeah, pull the desktop blob... if you dare.

Re:Same thing as always

[Miamicanes]

Or do the next best thing... come up with some kind of binary thunking layer to allow binary drivers to do an end-run around Linux's endlessly-changing ABI and allow a binary driver written for an older kernel to limp along under a newer one for the time being. And team up with Google to do it, so every fscking new version of Android won't catastrophically break every .ko driving a Qualcomm chip.

Someone seriously needs to take charge of this and just do it. I mean, for god's sake, if NDISWRAPPER can make *WINDOWS* drivers usable with Linux, how fscking hard can it be to build your own drivers in a way that everything they do interfaces with the kernel at arm's length through your own totally 100% open-source thunking layer, so that whenever a new kernel comes out, end users can recompile the thunking layer and keep using the older loadable kernel module? Hell, at this point, I suspect even Linus would grudgingly cooperate and throw some suggestions at such a project... not exactly *encouraging* it, but peeking at it once it a while just to make sure they don't accidentally do something totally bad.

Once... just fscking ONCE... I'd like to be able to upgrade to a new version of Android without having the camera, GPS, 4G, and half the phone's other features catastrophically break until someone manages to get their hands on leaked drivers from either a phone with compatible hardware and newer version of Android than mine, or leaked drivers that are being beta tested for my own phone's official release.

Re:Same thing as always

[im_thatoneguy]

And yet their open source drivers still are unusable...

I don't give 2 shits about "Theoretically" working drivers. I need reliable, stable and fast drivers. The AMD Drivers are none of those things.

Re:Same thing as always

[im_thatoneguy]

Seeing AMD trying is all it takes to make me a loyal customer for life.

Am I the only one who doesn't care how hard a company tries but how well they succeed?

AMD has been "trying" for a decade and in my experience never really worked right. So whatever they're doing in my opinion is the wrong approach or they would have something working by now.

Re:Same thing as always

[rsmith-mac]

It needs to be repeated so it sinks in, Linux needs a reliable ABI so ATI and others can just slap a Linux driver onto the CD that comes with the card and call it a day.

Agreed.

I much prefer the philosophy of having everything open sourced right down to the drivers, but we've been doing this dance with modern GPUs (read: pixel shaders) for what, more than a decade now? GPU driver development is just too damn complex to have good drivers developed in a reasonable amount of time by outsiders - you're writing an OS for a quirky processor that runs inside of another OS - which means drivers need to be created by the hardware manufacturer. And neither company is ever going to open source the "good stuff" that we need for higher performance since so much of their competitive advantage is in the software/driver stack in the first place. Which is why we get these half-assed AMD FOSS drivers that are more lip service than serviceable, while NVIDIA doesn't even pretend and just keeps hammering on binary blobs.

In short, the open source driver philosophy just doesn't work here; it has failed in the real world. Rather than beating this tired horse for another 10 years we need to just suck it up, implement a stable driver API/ABI, and make it practical to have binary blob video drivers while spending every waking moment extolling the virtues of FOSS drivers.

Looking at the long game, at some point these GPUs will evolve the point where they have stable ISAs and at which point writing drivers will radically change for the better. But until then the message is clear: the binary blob is here to stay. So if we want to improve the deplorable state of Linux graphics we need to stop fighting binary drivers with tooth and nail, and embrace it so that we can get good graphics now while setting ourselves up to dump the binary blob back to the curb in the future.

Re:Same thing as always

[tstrunk]

To sum up what I write downstairs: The open source ATI drivers coming from this effort are the best drivers available currently for Linux if you count out 3D performance.

For all cards except the two newest series (6xxx and 7xxx) they work flawless; they are the most stable drivers I ever had for any card (I used nvidia and ati; no intel chipset yet). No tears, everything works out of the box without tampering with anything. They 'work' better than the catalyst drivers, because there you will still get problems with Gnome3 sometimes; also the tear-free mode is not working with some compositing setups.
The only thing they cannot match is 3D performance. If you are working as a sysadmin in an office environment using no 3D apps, these drivers will give you the least headaches.

Whoever modded you insightful: Try the open source drivers first. Slapping a binary blob on a CD is a horror scenario I don't want to live. The radeon drivers are living proof that staying up to date with the kernel ABI is very much possible even with a small-ish development team.

Re:Same thing as always

[bug1]

Am I the only one who doesn't care how hard a company tries but how well they succeed?

No, lots of people, lots of people care about practical measures, rather than an ideology. But without ideology Free Software wouldnt even exist.

ATI have succeeded in releasing the specs, NVIDIA havent even done that.

The problems i have had with graphics under linux have been more to do with the whole stack, i dont blame ATI for whole thing.

Re:Same thing as always

[SuricouRaven]

I've had one: The Apple bluetooth keyboard on my mac pro ceased to work after a kernel update. I don't know how this keyboard works, really - it's some type of bluetooth-supported-in-EFI voodoo. I do know that the window during which you need to press alt to bring up the Apple boot selector is about half a second long, and there is no visual indication to say when.

Re:Same thing as always

[jimicus]

Don't you believe it.

It is quite possible to buy three computers from the same supplier, with the same model number and the same specifications within a month of each other and find that you wind up with three subtly different pieces of hardware.

There are ways to avoid it - most major suppliers will guarantee that certain ranges won't receive a change to the hardware without a corresponding change in the model number - but those ranges are typically quite a bit dearer.

Re:Same thing as always

[jez9999]

The VOIP visitation recorders at work have to have updates turned off because Ubuntu craps itself whenever it tries to update.

If you're looking for stability, why on Earth are you using Ubuntu?? Try Debian squeeze, or CentOS.

Re:Same thing as always

[drinkypoo]

All that's needed is for Nvidia to release the documentation on the components they manufacture, as AMD/ATI did in 2008 (and Intel has always done).

This is probably impossible too. nVidia and Direct3D were basically the same thing in DirectX6, which nVidia helped to create. Had to do with the Xbox, whose internal, original, concept name was DirectX-box. It is likely that nVidia can't release info on their hardware without running afoul of Microsoft.

Re:Same thing as always

[Bert64]

So those are problems that have occurred because the hardware uses binary drivers...

The moral of this story is don't use hardware that requires closed source out of tree drivers.

And were you just installing security updates on ubuntu LTS, or were you actually going to different releases?

I have several ubuntu servers and many other non ubuntu linux boxes, which are regularly updated and never had a problem. I also intentionally avoid hardware which requires proprietary drivers for this reason (among others).

And this is not just a linux problem, i have seen windows boxes and even macs fail after updates because of badly written third party binary drivers. The only difference is that windows users just seem to accept problems, while linux users are vocal about how unacceptable such problems are and want to do something about them.

Re:Same thing as always

[Bert64]

It's not the hardware, i have several linux boxes which use the open radeon drivers including some where the hardware is not supported by the closed drivers anymore. The performance is variable, in some areas its faster than the closed drivers and in other areas slower.. Ofcourse this is a moot point on the older hardware where the closed drivers don't work at all. Stability on the other hand is far better with the open drivers, and any problems that occur i at least have a chance of trying to pinpoint and/or fix the problem.

Re:Same thing as always

[Bert64]

Implementing a stable driver ABI impedes progress... Even MS have had to deprecate their old ABI several times, the most recent of which was vista.

As for writing drivers in a reasonable time frame there are several factors at work here...
One is that a lot of code only needs to be written once, and then adapted or added to for new hardware, once you have a stable working driver for a previous generation card modifying it to support the latest hardware is considerably less work than a from scratch implementation.
The other is that "outsiders" only get access to the hardware once it is released to the public, while internal developers have access to the hardware specs even before any actual hardware exists and can start developing the drivers alongside the hardware.

Ideally, driver development should be done in the open, ie its kicked off by internal employees of the hardware company and available in a working form when the hardware is, but then outside developers can contribute improvements, bugfixes and continued maintenance going forward... These improvements can then be built upon when making the next generation of hardware and so on.

And incidentally, the open AMD drivers are actually very good and improving all the time... They are far more stable than the closed counterparts, they work on hardware which is no longer supported by the closed drivers, they are faster and more reliable when it comes to 2d use, they provide a more seamless experience and allow you to update the rest of the system without risk of them breaking.
The only area where they are currently lacking is 3d performance.. The 3d support is stable, but generally not as fast, so unless you have specific requirements where performance is king (eg gaming) the open drivers are a much better choice. If you have older hardware, chances are you not only don't care about gaming and 3d performance at all but also don't even have the choice of using the closed drivers.

Re:Same thing as always

[macemoneta]

All that's needed is for Nvidia to release the documentation on the components they manufacture, as AMD/ATI did in 2008 (and Intel has always done).

This is probably impossible too. nVidia and Direct3D were basically the same thing in DirectX6, which nVidia helped to create. Had to do with the Xbox, whose internal, original, concept name was DirectX-box. It is likely that nVidia can't release info on their hardware without running afoul of Microsoft.

Assuming that is correct, there will still be functionality not covered by joint patents that is problematic for nouveau. Power management, fan control, re-clocking, Optimus, etc. are areas where implementation details would be useful to shorten the pain of reverse engineering reliable functionality.

Re:Same thing as always

[Bert64]

If you do that, then it will be seen as "good enough" and noone will write drivers to take advantage of new kernels, all drivers will end up limping along.

Re:Same thing as always

[Crosshair84]

"Implementing a stable driver ABI impedes progress"

Not having a stable driver ABI impedes progress. Not having one means you have to take your scarce and valuable developers and put them on fixing busted drivers every 6 months. If you DO have a well planned ABI you can instead put those developers on fixing real issues and improving the product. Linux continues to fall further and further behind Windows and Mac because of this.

Having a stable ABI means you can completely overhaul something and not have it impact anything else. You could completely rewrite something from scratch and as long as you have an ABI, the rest of the system doesn't even see the difference. Having a stable ABI means you can keep backwards compatibility via depreciated ABIs like Microsoft does.

"Even MS have had to deprecate their old ABI several times, the most recent of which was vista."

and yet all that old stuff still works fine. Hmmmm, seems like Microsoft knows how to design ABIs.

Everything else you wrote is irrelevant. OS's with ABIs have stable and reliable drivers. OS's without ABIs have sh*tty drivers that die.

Re:Same thing as always

[Bert64]

On the contrary, one of the biggest problems people had with Vista was a lack of drivers, many drivers made for XP didn't work with it.

The in-kernel drivers on Linux tend to be far less hassle than binary drivers on any platform. Plus they provide all manner of other advantages, for example the vast majority of USB devices supported by x86 linux will also run on ARM, whereas Windows RT will require the hardware vendors to port their drivers, and many won't be willing to do so.
Windows for Alpha, MIPS, IA64 and PPC suffered from a lack of drivers, as did the x64 version for quite some time (and still lacks support for older devices). Linux on all these architectures was able to use the vast majority of existing device drivers, many of which were never intended for use on anything other than x86 but still worked fine.

Driver ABIs have been deprecated many times, not just by Linux... There are all manner of people out there using proprietary hardware with ancient versions of Windows or DOS because the drivers don't work with anything newer.

The fact is that an ABI which seems suitable today, will become obsolete in the future.. It doesn't matter how much design work you do, sooner or later you will need to do something that requires a clean break. If you create an ABI and encourage binary drivers, then those drivers will become useless when the ABI inevitably requires updating. That's why binary drivers are discouraged, if we have source then we *can* update drivers to continue working.

MS also have serious security problems as a result of all the deprecated libraries they keep around...

Unstable drivers are also frequently blamed as the chief cause of windows instability, so MS earns a bad reputation and can't do anything about it because they don't have the source for all these drivers.

And the Linux userland does have a stable ABI and API. Binaries from the earliest versions of Linux (in a.out format no less) will still run on modern systems, although you do have to make sure you have the libraries they require (which many modern distros no longer bother distributing). The API goes back even further, software written for ancient unix systems can be compiled and run on Linux usually with no changes necessary.

What your advocating is that Linux users settle for the same level of hardware support present in proprietary systems... But the fact is, many people use Linux because it offers advantages over those proprietary systems, some of which you are advocating throwing away.

Linux drivers only die when there are no longer any users using them... The current Linux kernel has support for all manner of ancient hardware, and at least the hardware i've tried lately actually still works. For just one example, I booted a fairly recent 2.6.x Linux kernel on an Amiga just last week, and support for all of its hardware was still present and working.

Re:Same thing as always

[arose]

How about you at least try to comprehend what I'm saying instead of reading whatever you expected into it? Those drivers are pretty good after 4 years of development by a small team less than familiar with the specs (as in, compared to in-house driver developers who can draw heavily on their and their peers experience. They are up to speed now, but I'd bet their are massively hurting for manpower and testers, take away nvidia desktop drivers and the current 3D desktop users and quite a few developers will flock them, they'll have no choice! Also compare the progress of Linux, the BSDs and HURD. Userbase counts for a lot. Blender developers focusing on AMD along would probably weed out oodles of bugs.

Short term it will hurt them when the 3D animation houses and supercomputing users have to deal with buggier drivers, and without the huge pool of testing that happens on the desktop they undoubtedly will. Long term, long term they will be even more on their own when it comes to working with Linux/Xorg/Wayland developers, the Radeon drivers (of both kinds) will improve faster, because bug fixes if nothing else and AMD will undoubtedly figure out some way to capitalize on it. I also think you heavily underestimate how much supercomputing user care about being able to fix and adopt things, it's not just the price why Linux would up on all the clusters. Besides, on that end AMD is releasing their OpenCL pipeline, I very much doubt there has to be a high performance 3D driver attached to that to make it a great choice for supercomputing. Just some logic to load it into the GPU, it's not even a graphics card at that point.

Re:Same thing as always

[HiThere]

The problem is, it's not wise to encourage a monopoly, even if only one manufacturer is behaving reasonably. I don't know what the answer it.

Re:Same thing as always

[jthill]

People who want stability often prefer distributions that focus on stability, perhaps you could try one of those.

Re:Same thing as always

[diego.viola]

Stable ABI is bad because then we get flooded of blobs and our system become unstable. I don't want stable ABI, we need specs.

Re:Same thing as always

[diego.viola]

Stable ABIs are bad because then we get blobs from everyone and the system becomes a PITA to debug and/or too unstable. Fuck stable ABIs. We need specifications.

Re:Same thing as always

[diego.viola]

Such nonsense.

Having a stable ABI will get us tons of blobs from everyone and the system will become a PITA to debug and/or too unstable. Fuck stable ABIs. We need specifications.

Re:Same thing as always

[strikethree]

All that's needed is for Nvidia to release the documentation on the components they manufacture, as AMD/ATI did in 2008 (and Intel has always done).

How is the ATI/AMD driver working out now that 4 years have passed? Yeah. So what point is their for Nvidia to do the same?

It is clear that for whatever reason, there are no qualified devs to create drivers for high performance GPUs. So why is everyone getting their panties in a bunch? They keep clamoring for specs and when they get them, they apparently can not do anything useful with them... even after 4 years.

The phrase, "put up or shut up" comes to mind here.

Re:Same thing as always

[rastoboy29]

Well said.

Maybe the time has come to fork Linux.

Re:Same thing as always

[macemoneta]

The open source radeon driver is working flawlessly on my systems. Tear-free video, full Linux functionality, and acceptable 3D performance for light gaming. If you're having trouble with the open source driver, try LinuxQuestions.org.

Re:Same thing as always

[amrs]

So, AMD released a little documentation years ago. Nothing since.

This is from memory but didn't AMD also explain back then that they can't give out much information because they have an agreement with Microsoft to provide a fully encrypted path from a Blu-Ray disc up to the HDMI input of a monitor, in Vista and later? Giving out programming information would violate this agreement. So, they sold out willingly or were blackmailed to do that. After all, Nvidia and AMD are really small players compared to Microsoft or Intel.

As far as NVidia is concerned, I'm very happy with my GTX 670. Linux support is there and works fine, at least for 2D desktop and video acceleration with VDPAU. Only 3D-ish thing I use in Linux is Google's MapsGL in Firefox. Main reason I chose this card is Windows games. Might play some Linux games in the future again, if Valve is serious about Steam on Linux.

Re:Same thing as always

[Alex Belits]

And this has nothing to do with updating Ubuntu or any breakage that would (not) happen in the process. If system installs, it updates.

Re:Same thing as always

[Miamicanes]

OK, real-world semi-rhetorical question. What's worse... drivers that are tailored to a specific kernel, but leave end users with a phone that has dysfunctional GPS, camera, 4G, gyro, and whatever else was broken by the new kernel for 5-9 months until the phone's manufacturer grudgingly releases its official update, or drivers that will continue to work with the new version of Android many, many months before the official release ever sees the light of day?

Let's be honest... the Linux ABI breaks because half the kernel maintainers WANT new kernels to break old loadable kernel modules as badly and frequently as possible, and the other half of the kernel's maintainers got tired of fighting with them 10 years ago.

The sad truth is, Qualcomm owns the patents that run America's mobile phone networks, and nVidia makes most of the chips inside the Android phones sold by those same networks. Neither company is going to embrace open drivers anytime soon, and the only thing the current ABI-instability accomplishes is the wholesale DIS-empowerment of end users with Android phones, by keeping us dependent upon our carriers and phone manufacturers for new drivers.

Putting it back into a nVidia context... would you trade 0.2% of your video driver performance if it meant being able to upgrade your kernel at will, instead of having to wait for an official release of a newer one built for your specific kernel?

Re:Same thing as always

[badkarmadayaccount]

Replace standard headers with a compiler generated dispatch table - You can have multiple namespaced/versioned stable ABIs without anything in the kernel noticing.

Re:Same thing as always

[jimicus]

It certainly does if there's a kernel update.

Unlikely, I grant you. But not by any stretch of the imagination impossible.

Re:Same thing as always

[Alex Belits]

It certainly does if there's a kernel update.

Unlikely, I grant you. But not by any stretch of the imagination impossible.

There are very few things impossible. However Ubuntu never crashed after kernel updates.

Easy

[lennier1]

Firstly, deal with the goddamn shitfucking Optimus hardware that's out there!!!!!!

Re:Easy

[jones_supa]

Exactly. The chick's question at Aalto University, which started the whole discussion, was about Optimus. Roll up your sleeves, NVIDIA. :)

Re:Easy

[bemymonkey]

Yes. This is the one that end users will notice immediately...

Re:Easy

[lennier1]

^^

Last year, I bought an Eee PC 1215n as a portable video player and emulator, so I'm all too familiar with the problem myself. Optimus was intended to go easy on the resources, instead it made the hardware nearly unusable (even under Windows, using the official tools).

For fucks sake

[binarylarry]

How about you start supporting the fucking GPUs you are selling people, like this Nvidia Optimus shit!?

Crazy talk, I know, not kicking your paying customers in the nuts. Sounds like something wild and crazy that Apple would try to pull.

Re:For fucks sake

[LurkerXXX]

99.5% of their customers use Windows. I'm hoping they release an Optimus driver, but from a business (all about the money) perspective, I can see why it's not at all a priority for them. The've got no need to support 'the cause' thet Linus wants them to, and making that 0.5% of users happy vs the work involved in making the driver probably isn't a huge motivator.

And no, I don't expect them to release the specs. They don't want to give AMD any help in fuguring out what they are doing, and they don't want to let others know if they are breaking someone elses patents. Doing either of those to make 0.5% of your users happy would be a kind of stupid business decision.

Re:For fucks sake

[arose]

This is an article asking what more nvidia could possibly do, so...

Re:For fucks sake

[Deorus]

How about you start supporting the fucking GPUs you are selling people, like this Nvidia Optimus shit!?

Optimus / synergy is not a GPU; it's a power management implementation that relies on discrete video hardware for 2D and dedicated video hardware for 3D. Essentially it's almost the same deal as the old dedicated 3D cards from 3dfx in the 90s, except this time the switch is seamless, you can run windowed applications, and software can decide which chip to use on a per-application basis because instead of outputting directly to the screen, the dedicated hardware's frame buffer is available for the discrete hardware to read and display.

AMD has a similar implementation called Dynamic Switchable Graphics (the MacBook Pro that I'm using right now uses it to switch between the Intel 3000 integrated graphics processor and the dedicated AMD HD6770 Mobility) which as far as I know is not supported on Linux either, and for some reason nobody seems to be concerned about that... Why?

Re:For fucks sake

[binarylarry]

Go fuck yourself, Optimus is nothing at all like 2D/3D cards of yore. It's two separate cards that both do 2D and 3D. One is just much faster at 3D but uses more power.

The issue is with the *Nvidia GPU* with Optiumus on Linux. You can't get the GPU to initialize where there is no screen; there is no screen because it's hooked up to the intel GPU.

Re:For fucks sake

[Deorus]

Go fuck yourself, Optimus is nothing at all like 2D/3D cards of yore. It's two separate cards that both do 2D and 3D. One is just much faster at 3D but uses more power.

That's essentially what the Voodoo Rush and Voodoo 2 cards were -- separate add-on cards. Optimus itself is not video hardware, as I mentioned.

The issue is with the *Nvidia GPU* with Optiumus on Linux. You can't get the GPU to initialize where there is no screen; there is no screen because it's hooked up to the intel GPU.

Precisely what used to happen in the old days, except that now you need software to access the dedicated card's frame buffer instead of controlling which card sends data to the display. You didn't have a "screen" with a Voodoo Rush or Voodoo 2 either; what happened was that when you used OpenGL for anything, the driver would automatically stop routing video from the main card and initialize the 3D accelerator card.

Optimus

[jaminJay]

Better power management. KMS. More timely releases. Stop whinging that there are problems with Linux/Xorg/whatever preventing you from doing something and work with those up-streams to fix it. Stop making excuses, start releasing code. Extract the proprietry stuff to (a) smaller blob(s) and expose the trivial interconnects and such at the very least. Never say it can't be done: everything's possible. Etc.

Re:Optimus

[Electricity Likes Me]

Better support for the hybrid graphics chips going into laptops would help a lot.

Re:Optimus

[SuricouRaven]

Even for linux, only a very small proportion of the userbase can code anything beyond 'Hello, world.' I can throw a few half-decent utilities out myself, but drivers are far, far beyond my ability.

Re:Optimus

[Osgeld]

dude with new laptop and shit internet whines about filesize, news at 10

Re:Optimus

[Deorus]

NVIDIA's universal driver is a much cleaner and neater solution than anything else on Linux, and makes a lot more sense from an engineering point of view. They load a module into the kernel which purpose is to interact with the proprietary driver, and the proprietary driver does the rest of the magic. Their implementation couldn't be any more self-contained than that, plus as a consumer you benefit from a driver that is thoroughly tested even in niche platforms because it's universal. The lack of elegance is on the Linux side.

Re:Optimus

[drinkypoo]

Lots of us HAVE shitty internet, so it's a valid complaint. I physically cannot download the sources to Android-x86 at my home because I have a "mere" 768kbps and git insists on sending me a lot of shit I don't need when I just want to build sources. Supposedly a 5GB tree, I downloaded over 20GB and gave up. That takes a lot of evenings at this speed. I can get better service, but it's horribly priced and the fastest thing I'm offered is 1 megabit bursting to 3 and they won't even commit to a burst duration.

Re:Optimus

[badkarmadayaccount]

Git is smarter than that read up on your (command line) switches.

he knows

[phantomfive]

. I assume the issue is mainly the lack of open support for the graphics-related parts of our HW,

He knows exactly what people want. The entire point of the question is to make it look like they are doing something without actually doing something.

Re:he knows

[Luckyo]

He also knows that this is not going to happen, and is looking for a middle ground. Fanatical "open source or GTFO" movement isn't look pretty in this either.

Re:he knows

[VortexCortex]

Fanatical "open source or GTFO" movement isn't look pretty in this either.

To be fair, neither is the proprietary planned obsolescence movement.

Re:he knows

[wrook]

But it's not a middle ground. It's not related.

Let's say I'm trying to get to the bus station. You live in the area and so I ask you how to get to the bus station. You say, "I don't really want to tell you that." I get pissed off and complain loudly that even though you know the way to the bus station, you won't tell me. People start saying, "Boy, that guy's not friendly at all, is he?"

So in order to avoid the bad reputation you say, "Well, I still don't want to tell you where the bus station is, but what if I offered to cut your grass. Would that help matters?"

Of course I would love for you to cut my grass, but it isn't at all related to the issue of the bus station. It's not a middle ground, it's a red herring.

Re:he knows

[arose]

You sure showed that fanatical strawman where it's at.

Re:he knows

[drinkypoo]

That's a shitty analogy. Don't feel bad; most people make shitty analogies when a car gets anywhere near the conversation.

Here's a better automotive analogy to the current situation: It's like you're trying to get somewhere on a bus, because you don't own a car, don't want to own a car, you think cars suck. And you ask which bus you need to get on to get where you're going, and the person you're talking to says "Well, I just get in my car and I drive up this way and down that way and I'm there." You don't want to use a car, you want to ride the bus, but the person you're talking to isn't going to tell you about that, because (as it turns out) they work for a car company and their business is in sellin' cars.

It's still not perfect, of course, because it's an analogy. For some reason, even though [for the most part] we have fairly functional minds and are technically capable of discussing most issues without bullshit abstraction. The truth is that nVidia probably can't give away the information that's being requested of them because of their relationship with Microsoft. They can only tell you how to use their car to get there, or suggest you ask someone else with another vehiclular strategy.

Here is some idea

[dark-templer]

Help open source community write a generic high level opengl/whatever they need This layer will be hardware indepenedant, then at least write something in Galliume3D interface for nvidia cards and maintain it. Do it anyway, if we use nVidia cards and want to switch to Wayland everything works alright. Do it in a way, my laptop backlights work! Do it in a way that is easy for sysadmins to maitain. Finally, if you want to keep the same source on every platform Galume3D is much better. If you think it misses feature, improve it.

I'm sticking with Radeon

[ArcRiley]

Very happy with my Radeon chipsets, don't have to worry about kernel version incompatibility or graphics lockups from some bug we can't even begin to fix. I've had both problems with nVidia hardware.

No PR move from nVidia is going to change my mind beside opening the specs.

Re:I'm sticking with Radeon

[Barsteward]

I'm not. I've got an old laptop with a Radeon and i can't update opnesuse passed 11.4 because 12.1 will not load the GUI.

Re:I'm sticking with Radeon

[antdude]

How's the accelerated video features in the newer ATI/AMD Radeon video cards in Linux? I like NVIDIA's closed binary drivers because of this. However, ATI/AMD closed drivers weren't that good. :(

Alright, I'll bite...

[Narcocide]

Frankly the GPU support issues are going to be the big key issue no matter what. Whether you think its wrong/stupid/immoral/wasteful/anti-american or whatever to use Linux as a gaming platform or for any other sort of performance-critical real-time 3d rendering *some* Linux users will still disagree with you and do it anyway. Whether flaws in your business plan and/or shady under-the-table anti-competitive agreements made years ago with certain big software vendors preclude you from giving Linux full support or if its really some legitimate logistics problem *some* Linux users will still disbelieve the excuses and not forgive you for it. Some of those Linux users still consider themselves your paying customers. You will never truly live this down, but don't worry; judging by the word on the street these days neither will any of your competitors.

However, if you want to set up a smokescreen that hides the fact that *someone* in the higher end of Nvidia's chain of command is openly prejudiced against open source software (or just made a shit ton of cash on Microsoft stock perhaps and refuses to believe therefore that Linux is anything other than a waste of the company's time) you could at least consider making an attempt at distro-native packages for your driver that show evidence that you are capable of and willing to put at least half as much effort into working within the rules and parameters of various distro's packaging systems and with their tools properly as you've already put into that big self-extracting NVIDIA-Linux-x86-whatever.run bash script monstrosity to make it capable of the far more arduous task of cleaning up after itself enough to get a working GLX after it has fully subverted and broken said packaging system by using very crude non-native tools and methods that are not dependency-aware.

Re:Alright, I'll bite...

[jvillain]

What if Microsoft was to ... oh I don't know ... make a radical change to their OS and ... say no one liked it and didn't go trudging out to buy new hardware with Windows installed. I suspect the companies that would take the biggest hit in that case would be the ones with their lips farthest up Balmers butt. The ones that would do best would be the ones that had already started to cater to the other users out there. Just sayin' .

In a nutshell: no

[Anonymous Coward]

"I can't operate your cancer, but would it improve my image as a doctor if I cured your limp?"

Re: Some help is better than none.

[symbolset]

Not really. What is needed is hardware that can be documented. Nice of nVidia to confess they are "not it". Spares having to consider them.

nvidia same as adobe flash

[digitect]

Nvidia has the same problem Adobe has with Flash: Closed source equals instability.

Between Fedora 14 and 17, I have never experienced so much system instability in Linux, and I've been a user since RH 5.1. My X is now guaranteed to lock in 5 minutes either by watching a Flash video or doing a yum install kmod-nvidia.

The year of the desktop, yeah, right.

Re:nvidia same as adobe flash

[jvillain]

Well 64 bit flash on Linux is now incredibly stable since they dropped it. It's like it isn't even there.

Re:nvidia same as adobe flash

[SuricouRaven]

This is one of the many reasons I look forward to the day when flash can finally be taken out back and shot.

Re:nvidia same as adobe flash

[tepples]

Has 32-bit Flash in a 32-bit browser on a 64-bit OS stopped working as well?

Re:Of course

[ethan961]

Yes, it's their own fault that they're hindering the progress of the community. But corporations are corporations - the way I see it, we're lucky they're offering help at all in response to it, even if only to save their image. Truthfully I haven't bought Nvidia in years, it's been AMD/ATI for me for the last decade. Just remember that our wallets are the most powerful tool we have.

Summary of Previous Discussion

[peanutious]

Here's a summary of some of the most insightful discussions posted on slashdot when this discussion came up last week:

nVidia Issues:
*Proprietary drivers that don’t always survive kernel upgrades. So people who rely on nVidia's proprietary binary drivers can't always update their kernel or they lose their graphics until nVidia puts out an update. (from UnknowingFool) nVidia only provide a binary blob driver which makes bug fixing for it dependent on Nvidia's whims. (from AC)
*open source drivers – nVidia refuses to provide specs and API's for their hardware which make writing open drivers much more difficult and time-consuming because of having to reverse-engineer everything to get a workable driver. (from AC) As a result, open source drivers are unable to use full card functionality like full 3D acceleration (from UnknowingFool)

Summary of graphic chip vendor support (from Lonewolf666):
*AMD provides specifications and a small developer team that actually works on open source drivers.
*Intel provides open source drivers.
*NVIDIA makes good binary drivers, but those have problems when a new kernel version comes out with changed interfaces: Only NVIDIA can adapt them, and until they get around to it, NVIDIA may not work with the latest kernel version.

From rajafarian: If the kernel maintainers have a question about the hardware, they can't ask NVIDIA they have to test and reverse engineer to find the answer whereas with other companies, they may get an answer directly from the manufacturer. Get it? "...NVIDIA just made the damn drivers. Now that is not good enough." Not from a kernel maintainer's or Stallman's point of view, I'm pretty sure.

From jmorris42 : Name another major chip vendor who hasn't figured out that getting into the Linux kernel is a required checkoff for market success. Doubly so for any product used in the enterprise vs the fanboi market. NVidia's CUDA is about the entire list these days, the last major holdout.

From basscomm: Windows users who have SLI and multiple monitors have been able to enable SLI and use both of their monitors at the same time since about 2008. But under Linux, no dice. So if I had two monitors (which I do), and two Nvidia GPUs in SLI mode (which I do), and I wanted to run some 3D app that took advantage of SLI, I would have to: reconfigure X to disable my second monitor and enable SLI, restart X, play the game/use the app I wanted, when I was done I would have to reconfigure X again to enable my second monitor and disable SLI, restart X again, and reopen all my apps. Hardly ideal.

Given all of this discussion, here are a few ways nVidia could work better with the community:
*Open Source drivers - 1) provide specs 2) provide developer team that works on the OS drivers 3) provide rep to interface with the OS community 4) provide enough detail to get 3D working well
*Proprietary drivers - 1) monitor upcoming kernel builds and proactively update drivers before the next kernel release or 2) have a dedicated nVidia contact to work on updating drivers ASAP when notified that an upcoming kernel build breaks them
*Overall - enhannce SLI and multiple monitor support,

Re:Summary of Previous Discussion

[kthreadd]

Sounds like the situation could improve if Linux exposed a more stable API toward drivers.

Re:Summary of Previous Discussion

[bersl2]

That's not the Linux driver model, though. The model is that drivers are expected to be at least working towards integration into the kernel tree, so that the kernel devs can easily know if they're breaking something and maybe even fix or help fix it. Failure to do this results in long-term pain for everyone, while being in the tree or working towards being in the tree keeps everyone communicating on a regular basis and working together.

In other words, developing hardware for Linux actually requires communicating with people who are much closer to the end user than the typical direct customer of a chip design company. I know, it's a major paradigm shift for companies who (understandably) can't contemplate money except if someone plans to give it directly to them, never mind the needs of those who participate in the actual creation of demand for what they sell.

Re:Summary of Previous Discussion

[jones_supa]

You are absolutely correct. Currently the stuff that works best is the (OSS) drivers that can be continuously updated to match a recent kernel API and/or X.org ABI. In Windows world, you can create a driver which pretty much works the whole lifetime of a major OS version. Additional points for having a unified way for installing third party drivers instead of a vendor-written shell script. :)

Re:Summary of Previous Discussion

[rtfa-troll]

Sounds like the situation could improve if Linux exposed a more stable API toward drivers.

Allowing and partially supporting hidden APIs is a mistake that has been often repeated in the computing industry. This is something where I think the kernel developers should learn from Apple. Apple, during old MacOS days used to deliberately, repeatedly and gratuitously change their hidden API calls even with minor point releases. This meant that software which used the API in ways that it shouldn't (in this case, by providing a binary blob) would break often and quickly. Users would be unhappy with their software and vendors would be forced to change. Software developers learned to read and respect the documentation. When they rewrote their O/S and even migrated to completely different architectures this meant that, merely by reimplementing the well defined published interfaces, almost all Apple third party software vendor's software was compatible.

I think that somewhat perversely, the actual mistake is that the unsupported linux APIs are too stable. This is the same mistake as Microsoft has often made and is something where the Linux Kernel developers and other free software vendors could learn from Apple. Microsoft's main benefit is plausible deniability when they want to have hidden interfaces for their own software. Free software doesn't need or benefit from such underhand tactics. There's no need to repeat Microsoft's mistake.

Re:Summary of Previous Discussion

[GNious]

This is the part I don't understand:

Why is Linux unable to provide a stable API, at least over a given period?

Linux 3.x has been released. Why could it not have a set of API/ABI calls that are guaranteed to be stable for the duration of the v3.x kernel?
Example
    SomeFunction() - Default function, does something, but not 100% guaranteed to behave the same way always.
    SomeFunction_v3() - Largely same function, but guaranteed stable for Linux v3.x

Manufacturers like nVidia can utilize the _v3() calls if they prefer, and state that they support/require Linux v3.x kernels. A user then knows to stick with Linux v3.x until a new API and new drivers are released.
nVidia et al can also choose to have a build that uses the regular calls, so users on a different kernel can at least try to get the drivers working on a pre-release. Such a build would likely not be fully supported, but being available would mean that tinkering/deviating is possible.

As for Linux having to maintain the stable calls, this might be an advantage, not just in terms of getting more/stabler drivers. It should force some discipline into the kernel-developers, that at least in the past has been missing (I don't mess with kernels any longer, but around 2.4/2.6 there were some shitty kernels released occasionally.) Naturally, the Kernel would need a "Non-Free" option to enable any ABI to be used by binary blobs - RMS compliance and all that.
Note: It might also be possible to get ATI/AMD/nVidia/BroadComm etc to help with developing the stable interfaces, if they can see that this is in their interest. Would help to get them into the kernel and tinkering.

Overall, I think it is a win/win situation if a time-limited API/ABI can be available for those manufacturers, who refuse/cannot participate directly in kernel development, and one Linus and Friends should reconsider.

Note: I can appreciate the fear that if further support for Binary Blobs is added, manufacturers would simply skip on developing drivers inside the kernel. Only solution, that I can see, is to set up a team to help facilitate proper driver-development, which could make in-kernel development cheaper for the HW companies, than doing Binary Blobs on their own.

Re:Summary of Previous Discussion

[Pentium100]

Completely agree - just look at the situation with Windows. Drivers made for Windows 2000 work on XP, most (not video though) work on vista and 7 (and probably 8). Some drivers even work on Windows NT4. Updates do not break compatibility with drivers, though some newer drivers may not be compatible with old service packs, but that's OK.

The goal of open source everything is admirable, but it is impractical - the company may have signed an NDA regarding part of the hardware, so it cannot release the specs or whatever. Yes, everything open source would be ideal, but I'll just stick with reality - use stuff that works, not caring about the ideology.

Re:Summary of Previous Discussion

[diego.viola]

Do you want to be flooded with binary blobs and make Linux more unstable and a PITA to debug? That's what a stable will do to Linux.

FUCK STABLE ABI.

WE NEED FUCKING SPECIFICATIONS.

Re:Summary of Previous Discussion

[diego.viola]

*That's what a stable ABI will do to Linux.

Re:Summary of Previous Discussion

[diego.viola]

Fuck off.

Re:Summary of Previous Discussion

[diego.viola]

A stable ABI won't do us good, do you want Linux to be flooded with blobs that make the system a PITA to debug and/or cause stability issues, that's what a stable ABI will give to Linux.

We need specifications, not a stable ABI and/or blobs.

How many fucking times do we need to repeat this?

Re:Summary of Previous Discussion

[hobarrera]

Linux is a kernel, Windows is an OS.
Naturally, Linux doesn't have a user-friendly way of installing drivers because it's just a kernel. Installing stuff isn't the kernel's job.

Re:Summary of Previous Discussion

[jones_supa]

That is just unimportant nitpicking. :)

Nothing Really

[chadruva]

Nothing, really!

I can complain a lot about their lack of support of modern X stack, xrandr, etc. But my next card will be nVidia, unless performance is not an issue at all (in which case integrated Intel is more than perfect).

Unless AMD put its stuff together and starts either releasing quality closed drivers or pumping some extra effort (money, more docs, help) onto OSS drivers, I don't see any threat to nVidia in the linux perfomance graphics segment.

I do not think Intel will try to get head to head on the performance GPU market and thus they do not represent a serious threat to nVidia in the linux perfomance graphics segment too.

Same thing as always: DRM.

[Ostracus]

It's unnecessary, and likely impossible, for Nvidia to open source its proprietary driver, due to licensed software they don't own (they have stated in the past).

It also can break DRM.

Re:Same thing as always: DRM.

[gl4ss]

It's unnecessary, and likely impossible, for Nvidia to open source its proprietary driver, due to licensed software they don't own (they have stated in the past).

It also can break DRM.

which brings up the question, is there a way to legally play a bluray(with hdcp and all that jazz) in linux? if you could make a driver that lied about the hdcp state, then that would be a breach(it's irrelevant to this that bluray drm is already effectively broken)

Re:Ugh, this makes me Stallman.

[rubycodez]

crown jewels? the company makes graphics cards, no one is buying drivers without the card. Most software companies sell support.

you closed sourcers are so ignornant on what makes money.....

I hate to interrupt you slashdot...

[mynis01]

...but here's some things I wish NVIDIA would add to their closed source driver: 1)Allow us to just shut off the nvidia card on optimus laptops and use the Intel one for video display. That would at least make the laptops with optimus usable. Brownie points for letting us run CUDA/vdpau stuff on the nvidia card without using it as a display adapter. 2)Re-enable coolbits/nvclock. My GTX 670 GPU sits to about 80-82C while using BOINC with it, and it would be nice to be able to underclock it inside linux, or at least speed up the fan. 3)Make it so the proprietary driver doesn't make my TTYs all ugly!

Read slashdot...

[XB-70]

...and actually respond to the comments. Start making us feel like we're part of the development loop. Let us know when you run into a problem and why something is difficult or an obstacle. If we know and understand the problems you're facing, we'll be far more likely to have a positive attitude towards your product line and cut you some slack. If you ignore us and create a huge, anti-feedback corporate wall, we'll feel shunned and ignored and respond with negative posts (whether we're accurate or not about what's really going on).

Re:Read slashdot...

[jones_supa]

Is the NVIDIA guy aware of this thread?

Re:Of course

[Anonymous Coward]

Absolutely!

However, AMD/ATI is a PR stunt. The drivers just wrap non-free software and can't be utilised at all on truly free software platforms. Intel is the way to go. While you can't buy an Intel card explicitly you can utilise boards and/or laptops without nVidia/ATI and then use an Intel CPU with integrated graphics.

http://www.thinkpenguin.com/ offers absolutely the best hardware for free software users. There are no proprietary drivers or firmware required and even the free software endorsed Trisquel distribution is supported. That isn't just some hardware. It's everything. An impressive feet given the selection of hardware available.

Actually. ThinkPenguin has the largest catalog of GNU/Linux hardware by far. There really isn't a comparable offering anywhere else.

The Good, the Bad, and the Ugly

[DrJimbo]

Years ago I got good support from Nvidia. I had a problem and sent a bug report and got a response late on a Saturday night (West Coast) that involved a kernel patch to fix a bug in the kernel that was causing the problem. I was an Nvidia fanboy for many years after that. I helped dozens of people get things like Twinview working.

In recent years, working with Nivida has been very frustrating and I can no longer recommend them for Linux systems. For example some interaction between closed-source Flash and the closed-source Nvidia drivers turns people blue in Youtube videos but not other sites.

There has also been a heart breaking struggle to remove video "tearing" (vsync problem) when watching dvds and blurays. The last time I checked I needed to use the GL video to remove tearing when watching dvds but I still have some tearing when watching blurays which is kind of heart breaking. At the very least Nvidia should have a sticky post in the Linux forum explaining all the hoops one must jump through to try to get rid of video tearing. Also, having the sticky posts show up on all pages, not just the first page is a big PITA. It wastes my time and attention. It is disrepectful.

I don't understand why video tearing is such a recurring problem. Are these environment variables still needed?

export VDPAU_NVIDIA_SYNC_DISPLAY_DEVICE="DFP-0"
export __VDPAU_NVIDIA_SYNC_DISPLAY_DEVICE="DFP-0"
export GL_SYNC_TO_VBLANK=1
export __GL_SYNC_TO_VBLANK=1
export GL_SYNC_DISPLAY_DEVICE="DFP-0"
export __GL_SYNC_DISPLAY_DEVICE="DFP-0"

When an nvidia driver update causes tearing to start again, or worse, if it causes X (or the entire machine) to crash, it feels like you are telling me "F- you!"

I get it that opening up your drivers is not an option. The problem is that this decision causes a lot of breakage and you do not make it easy to fix this breakage. Please just make it easy for me to get your drivers to work. Is Twinview plus non-tearing video playback really too much to ask for? Also, what about the problem with non-tearing and composite? Has that ever been fixed? If not, maybe that's something you could help with.

Years ago I felt like I was getting support. Nowadays I'm not feeling the love.

Re:The Good, the Bad, and the Ugly

[mynis01]

I don't use any of those environment variables. Just use mplayer with vdpau or vlc with glx output and the problems disappear (as long as I disable cairo-compmgr). I run two 1080p devices with different refresh rates stacked on top of each other in twinview and I don't have vsync issues. This is a bash script I use to disable compositioning when mplayer starts (replace cairo-compmgr with your composition manager of choice): '\'' if [ -z "$(pgrep cairo-compmgr)" ] then $* else killall cairo-compmgr $* cairo-compmgr & fi '\'' Lets say we called that script "no-comp.sh". Then you make another script to bind a hotkey to that launches your program that you want to use without compositioning: '\'' bash no-comp.sh smplayer2 '\'' As soon as you close smplayer2, your composition manager will start right back up.

Re:The Good, the Bad, and the Ugly

[diego.viola]

Fuck nvidia.

Stop developing a closed kernel module...

[iCEBaLM]

Move everything to userspace, and use an existing driver, or a very small open driver, to access the card. There should be no reason, even if they're crying "trade secrets" for this to happen. If nouveau can do it, and they're not crying trade secrets over it, then nvidia proper can do it too.

Re:Stop developing a closed kernel module...

[kthreadd]

Move everything to userspace, and use an existing driver, or a very small open driver, to access the card. There should be no reason, even if they're crying "trade secrets" for this to happen. If nouveau can do it, and they're not crying trade secrets over it, then nvidia proper can do it too.

Everyone is free to keep secrets, even you. If you have a problem with their actions then use other chipsets.

Re:Stop developing a closed kernel module...

[Junta]

If nouveau can do it, and they're not crying trade secrets over it, then nvidia proper can do it too.

If nouveau worked just as well as proprietary drivers, then no one would be bothered. The fact is nVidia is doing *something* that makes them much much much faster than nouveau, and it's difficult to speculate what can and can not be done without destroying performance.

Re:Stop developing a closed kernel module...

[diego.viola]

That something will eventually be found with or without nvidia's help. I'm sure.

What's wrong with your update system or ubuntu?

[IBitOBear]

I run gentoo on a myraid of boxes and update quite regularly including bleeding edge kernel and ~ARCH ("not yet gnerally released for this this archetecture") packages, which is as bleeding edge as you can get, and I havn't had a update failure outside of the X config (which I have been churning a lot on purpose) in forever.

If your distro is crap, maybe you should change your distro.

If your admin is crap, maybe you should change your admin.

I would characterize your characterization of patch stability under linux as "uncommonly unlikely, or procedurely impure" just on blind weight-of-anecdote compared to everyone eles I know.

Then I did drop Ubuntu because they started to bring in non-open tidbits (and freaking mono) so maybe you are choking on some binary bull (or someone else's platform envy). /doh.

Re:What's wrong with your update system or ubuntu?

[Crosshair84]

"If your distro is crap, maybe you should change your distro."

We are, we are transitioning to Windows.

"If your admin is crap, maybe you should change your admin."

Our admin has real work to do, he's not sitting around playing Halo. Wasting a couple days fixing Linux monkey code every 6 months gets very expensive, especially when I have to drive to the site because the NIC drivers crapped themselves. He needs to continue development of new features with our programmer to keep us ahead of our competitors.

er

[IBitOBear]

So that's two "shitstorms"...

Oddly enough that isn't compelling in its depth of analysis, and in both cases you can point to closed vendor hardware....

hrm... patern?

Re:er

[im_thatoneguy]

in both cases you can point to closed vendor hardware....

hrm... patern?

Dell and Nvidia? Yeah, the pattern sounds like he was using real computers that an average user would encounter.

God forbid someone try running normal computers that aren't religiously pure.

Re:er

[Alex Belits]

It doesn't.

Well if you can't tell us what is -in- the blob

[IBitOBear]

Try de-obfuscating the interface to the blob itself. Just come out and say "our blob is something we are all stuck with, therefore, this is how you will now and alwyas operate the blob."

If you cannot, for some legitimate reason, release the API for your hardware, then release the API for your stoftware interface as a well defined and "warrented" interface.

Then make that interface your -hardware- -interface- by pushing that back into the GPU or a "front end" processor.

In short, you claim you have a mess you cannot share with us for all sorts of odd reasons...? Then put your mess inside your own junk and give us an interface we can then use. You test and warrant your half and we'll test and warrent ours. This is far superior to this thing where you calim to need to come into our yard and move the furnishing to match some fung shui most holy and secret.

And if your blob interfce is too broken to play well, then its to broken to expect us to want and need so its kinda moot that you gave it to us anyway.

This is not that hard. Not that you are that unique. Ever try to get an interface document for a Brother label writer etc? All this "my ball" crap is killing innovation.

Re:Well if you can't tell us what is -in- the blob

[Junta]

I would assume that could/would drive cost up. Suddenly you need infrastructure on the device to support a piece of code that isn't a problem for 97% of your customers that adds a few dollars to the price, which kills margins on the integrated offerings.

Besides, I presume part of the whole need of the driver is to manage how data flows through the whole system, not just what happens inside the GPU itself.

Re:LOL slashdot

[X0563511]

News for nerds, random bullshit posts by anonymous assholes.

Re:Of course

[ppanon]

However, AMD/ATI is a PR stunt. The drivers just wrap non-free software and can't be utilised at all on truly free software platforms.

Seriously? What do you think this is about? What's the licence on this that makes you think it's non-free? You seriously don't think this licence cuts it?

Re:Of course

[ArcherB]

It's everything. An impressive feet given the selection of hardware available.

Must be a Gnome user...

Re:stable driver ABI

[unixisc]

Does Nvidia's driver exist for the BSDs, which do have a driver ABI?

Re:Ugh, this makes me Stallman.

[kthreadd]

crown jewels? the company makes graphics cards, no one is buying drivers without the card. Most software companies sell support.

you closed sourcers are so ignornant on what makes money.....

How do you know that? Nvidia may have good reasons for their actions, that this is how they maximize their profit. The always unknown step two.

Instead of being upset about companies which rightfully keeps its secrets closed we should thank those who don't, such as Intel.

Re:Ugh, this makes me Stallman.

[bky1701]

So NVIDA's product is the driver, and the card is just an extra? Ok, they can go out of business. (Or not, since it isn't, and you're full of shit.)

Perhaps not just Linus' comments

[SpaghettiPattern]

Perhaps in this case Linus' comments were secondary to his vim and tertiary to the gesture -which would have been rude in other circumstances and apparently perfectly appropriate in this case- in his statement.

Re:How About...

[kthreadd]

Or, how about buying competitive open alternatives from other vendors?

Re:stable driver ABI

[kthreadd]

As far as I know Nvidia support their driver on FreeBSD.

Well, they lost me.

[lattyware]

It's funny, when I used to buy graphics cards, I bought nVidia because the driver was that much better - TwinView meant good multi-monitor support, etc... Now all that has changed. I replaced two nVidia cards with an AMD card, because AMD support their Eyefinity tech even on low-end cards, and my triple monitor setup works perfectly with 3D acceleration on the open drivers, which means I get KMS and no binary blobs. By comparison, I had to use xinerama to get triple monitors with the nVidia driver, no KMS, no 3D acceleration.

It's obvious what we want, open source drivers. If they can't handle that, documentation, but we all know they'll spout the same stuff about NDAs and 3rd party stuff they are licensing and can't reveal, and tell us binary blobs are the only way - well, I don't buy it. Intel manage it well, and AMD are definitely getting there. Binary blobs just don't give the experience any more, that's the reality.

Re:Well, they lost me.

[moderators_are_w*nke]

I find the Nvidia drivers to be excellent from a features perspective. 3D acceleration is excellent.

Re:Well, they lost me.

[Lisias]

Yes, NVIDIA has the right to keep all their drivers closed source.

But if they want to sell CHIPs to Linux powered machines, it's better to open the drivers or face the risk of loosing this market to the ones that do it.

You know, they do not HAVE to sell chips to Linux users if they don't want to open their drivers, right?

Re:Well, they lost me.

[diego.viola]

They have all the right they want, but others are being more open than them, and if they want to remain competitive for the next years to come, they better do something about it.

Re:Well, they lost me.

[diego.viola]

While it sucks in other areas, no proper XRandR, no KMS, no Wayland, no optimus support, poor suspend/resume/ACPI, poor integration with the rest of the system (user-space), etc.

Asking for the empty set?

[jejones]

From TFE (email): "Within the constraints I have, what should I and perhaps other NVIDIA employees be contributing to in the kernel?"

I suspect those constraints essentially preclude what would really be useful, so what's the point?

Re:Asking for the empty set?

[Pastis]

I suspect one should maybe let kernel developers answer those technical questions themselves ?

For example Ted Tso's answers points out to some things that fall in the scope of the feasible given Nvidia's developer constraints:

http://lists.linux-foundation.org/pipermail/ksummit-2012-discuss/2012-June/000305.html

And Mathew Garret's tips for splitting to open kernel/closed userspace architecture as also valid:

http://lists.linux-foundation.org/pipermail/ksummit-2012-discuss/2012-June/000336.html

Re:Of course

[fuzzyfuzzyfungus]

AC may be referring to the firmware blobs without which newer AMD chips are good for little more than spartan 2d... That said, you really end up picking your poison with firmware. Some vendors include enough flash to store the blob, some demand that the kernel hand them the blob. There are exceptional cases, where the firmware is OSS, or where the vendor is a real asshole and forbids the blob to be distributed(for no obvious reason, since somebody always whips up a script that grabs the windows driver from the vendor site and gouges the firmware out of that.) In this unfortunate vein, it is probably worth noting that Intel is better about not making the firmware stuff visible, but AMD has historically been overwhelmingly nicer about coreboot vs. BIOS.

Small things for me

[Knuckles]

Get rid of the nvidia-settings application and provide something that resembles modern support for multimonitor setup, preferably by tying into the desktop environments' infrastructures. And support KMS.

Re:Small things for me

[drinkypoo]

Get rid of the nvidia-settings application and provide something that resembles modern support for multimonitor setup, preferably by tying into the desktop environments' infrastructures. And support KMS.

Dingdingding. nvidia-settings sucks. Only the text-mode component should stick around, and it should partly interface to XRandR and partly should provide the command-line conf file editing functionality. The GUI part of it ought to go away entirely.

Console Framebuffer

[DMJC]

For years there has always been the problem of the closed source drivers not supporting the console framebuffer properly. I think if NVIDIA had a proper console framebuffer it would go a long way to making the linux experience smoother. Honestly that's always been my biggest gripe with the drivers. Under X11 the drivers support pretty much everything I want to use. But as soon as I hit Ctrl-Alt-F1 everything turns to shit. Also, I've noticed that the SLI performance boost doesn't appear to exist at least on 460gtx hardware. I have dual GPUs but on linux there doesn't appear to be a noticeable improvement in wine or even opengl linux applications with SLI enabled.

Re:Of course

[bhcompy]

I take it you hate gaming for Linux, since Intel is worse than Matrox for gaming.

Re:Beware of dogs! Beware of the mutilators!

[equex]

lol get some help

Re:"Just specs"?

[Gordonjcp]

If Linux had a platform share of 50% of the hardware market, they'd have sufficient grounds for biting the bullet and betting on staying ahead of the competition by being the best at doing the hardware part of their architecture.

The thing is, they could open their specs, wait a couple of months until open-source driver support catches up or even exceeds their closed-source drivers, and take pretty close to 100% of the Linux market.

Now, because Linux has at that point got as good or better accelerated video support than Windows, it will overtake Windows even faster than it's already doing for things like home theatre PCs. Valve will be happy, since their Linux-based console will be able to rock the cheap, powerful NVidia chips, and they will port more games to Linux since it's a pish easy job because of the improved 3D acceleration.

Yes, that's a bit idealistic.

it's very simple

[Xtifr]

Any or all contributions will be welcomed, and will improve my image of the company. Improvements that don't involve opening up the HW mean still no sales here, but I will have a better image of the company. :)

Re:it's very simple

[Xtifr]

If they don't care, why did they ask? Note that the question was about Linux support, so the opinion of Microsoft shills is probably of much less interest to them than mine.

Anyway, the thread is filled with people saying that nothing but opening their hardware will do. That's just silly. Why wouldn't their contributions be welcomed, and improve my/our image of the company? Contrariwise, for the large numbers of us who don't buy NVidia cards because of the non-free drivers, why would a lack of change in that department affect our willingness to buy their cards?

Re: Some help is better than none.

[Tough Love]

Not really. What is needed is hardware that can be documented. Nice of nVidia to confess they are "not it". Spares having to consider them.

I've been NVidia-free for quite some time, and happier for it.

Re: Some help is better than none.

[Tough Love]

for whatever reason, if their higher ups don't want to be telling the world how their drivers work, and so they won't.

That's nothing a good firing wouldn't fix.

If the engineer knows it will never happen

[Anonymous Coward]

and we ought to just accept it, then he should know a better perception will never happen and just accept it and the consequent loss of sales.

The REALLY DUMB thing about both the engineer (and NVidia) and you is that you don't seem to even consider "Hey, lets tell them why it can't ever be done.".

Except that was tried a long time ago, where NVidia said something like "SGI owns some of the stuff we use and they won't let us release it". Well, given specific goals, the FOSS people went to SGI and asked "Can NVidia release the stuff they got from you?" SGI answered "We cannot conceive of anything that NVidia has off us that we wouldn't be happy with them releasing.".

Now, given that, did NVidia release specs?

No.

And, having been caught out in a porkie-pie, they have refused since then to do other than let their fanbois (and I use their cards too) say "Oh, maybe they can't for contract reasons" because, not being party to any secret NDA, the fanbois cannot supply any information on who/what or why and therefore it can't be rectified.

Re:If the engineer knows it will never happen

[Anonymous Coward]

Except that was tried a long time ago, where NVidia said something like "SGI owns some of the stuff we use and they won't let us release it". Well, given specific goals, the FOSS people went to SGI and asked "Can NVidia release the stuff they got from you?" SGI answered "We cannot conceive of anything that NVidia has off us that we wouldn't be happy with them releasing.".

First, this is common... company A says "I cannot do X because I have an agreement with company B" and then some person allegedly from company B (but who may not be in a position to know, or who knows but just feels like lying) says "nah, we're not the bad guys here!". If you have not seen ALL contracts between the two companies, then you have no way of knowing the truth... however IF such an agreement exists, Nvidia will never breach it even if goaded into it by statements from the other vendor, because then the other vendor could sue them.

second, AFAIK Nvidia has never said that the ONLY reason for the closed nature of their stuff was one agreement with SGI. In fact, I once worked at a company then got into talks with Nvidia and we attempted to negotiate for certain info and I was told (this is third-hand info) that our people were told that Nviida could not release what we wanted even under NDA because they had contracts with multiple IP owners. I have no Idea of how true that is other than that I trusted the person I heard it from at the time... but my point is that nobody other than Nvidia and anybody they have inked deals with truly KNOW what is involved. One thing is certainly true however: Nvidia gains no sales by holding the info private and loses sales by not opening it up, so the odds are that they truly are unable to open it up and no public statements by anybody else will change that. In fact, Nvidia may well be contractually forbidden to even publicly divulge the identity and/or existence of some contracts and partners.

The real problem here, however, is NOT Nvidia...

Nvidia makes hardware. The hardware has a hardware interface (but not an "API" as some mistakenly claim). The hardware must be compatible with the host PC, which has a clearly published hardware spec. Clearly Nvidia hardware interfaces well with the host machine because both sets of hardware conform properly to the hardware "binary/digital" spec.

Nvidia makes a software driver. On Windows, there is an ABI spec that Nvidia can adhere to so their driver will play well with Windows. On Linux, things are either so poorly implemented that the driver interfaces are in constant flux and no ABI is possible, OR the architecture itself is so incomplete that no ABI has been completed, so Nvidia goes out of their way to make a driver wrapper that can be re-built at install time to adapt their solid and high-performance code and hardware to the unsettled platform. The simple solution is for Linux to finally mature to the point that it has a stable, well-documented ABI. Refusing to do this is immature/political not practical. The refusal to have a stable ABI is not just a pain re Nvidia, it is the reason why any minor update of linux can trigger a tidal wave of package updates and result in an unstable system. Adding another unstable pile of incomplete "open" code like nouveau does nothing to advance linux into a modern mature state... it's a not-work-around for a broken/missing ABI and all the anti-Nvidia FUD on earth will not make the nouveau driver fully-functional. With a proper ABI, old hardware could continue to be functional on future Linux releases even when a maintainer for the device's driver code cannot be found. Without a proper ABI, such hardware will inevitably be dropped as maintainers lose interest.

Software patent troll problem

[dbIII]

I think some of the SGI people that were burnt very badly by software patent trolls ended up at Nvidia. Currently there is nothing to stop it all happening again if Nvidia describe their methods in enough detail to let opportunists go after them in court. It appears they don't think they can afford to open up the specs while broad software patents that should never have been granted still cover everything obvious in computer graphics.
That appears to be the "big deal".

Wayland?

[ChunderDownunder]

Retrofit your monolithic binary blob into a wayland capable architecture and you'll have addressed many concerns such as KMS, EGL, etc.

Supporting it on Tegra would be a step towards running standard accelerated Linux on an ARM phone or tablet and you'll be the slashdot nerd's platform of choice ahead of Adreno, Mali, PowerVR, VideoCore etc.

Would you use a 1992 GPU?

[tepples]

Patents last how long?

Patents last 20 years after filing or 17 years after grant, whichever is longer. Modern PCs don't even have slots for video cards made in 1992.

Re:Of course

[nashv]

It is not nVidia's job to do ANYTHING for X or Y operating system. If they make crappy drivers for a certain OS, they lose market on that OS. Maybe they don't care about that market.

Good news. . .

[JSBiff]

Hey, good news. China is changing.

Now, they tell you up-front that in order to sell a small number of units in China, you must hand over all your technical documentation so that they can start making and selling them after the initial sale.

Re:Optimus Support!

[Deorus]

So, will you buy the even less supported Dynamic Switchable Graphics solution from AMD or are you giving up on dedicated 3D acceleration for good?

No more nvidia for me

[trevelyon]

After my optimus experience I will simply not buy another closed source NVIDIA graphics solution. In the end I much like others and the internet in general route around broken parts. I will be curious to see how kernel module signing some distros are doing now for secure boot will effect NVIDIA proprietary drivers.

Re:No more nvidia for me

[Junta]

Challenge being, what Optimus like technology actually works in Linux? I'm unaware of anything off hand. When you have nothing to compare with, what are you supposed to do? Perhaps the life lesson is to look up the product before purchase regardless of vendor to understand what is going to happen.

On the other hand, of the technology that *has* viable alternatives, how does nVidia stack up, capability wise, to the competition? In terms of accelerated 3D, AMD and nVidia proprietary drivers are both very capable, and none of the open source (including Intel GPU) comes close. In the case of Intel, you have to limit the comparison to similarly weak AMD/nVidia hardware, but even then it still lags significantly. One oddity though. With Mythfrontend, any attempt to use GL painter or renderer fails miserably with AMD. This either indicates an AMD bug or a MythTV bug, but the latter is still AMD's problem since clearly the developers do not target their platform.

In terms of video decode acceleration, nVidia VDPAU more or less stands alone. VA-API comes in second, but even with a brand new Ivy Bridge GPU, the experience is far more poor than VDPAU on a low end nVidia GPU that's 4 years old in projects that support both. VDPAU also has more project support. XvBA doesn't even warrant a mention, between the crappy capabilities and only one experimental branch of a major multimedia project bothering.

Video playback tear-free is a massive issue for everyone except nVidia. They are the only ones that do tear-free XV (though usually not needed due to vdpau). They also are the only ones I've tried where OpenGL renderers have worked consistently well.

In terms of kernel module signing, it looked like only RedHat was planning to play that game, and Ubuntu was pursuing a more minimalist approach where the signing requirement is removed as soon as possible. Even when they do 'play the game', I wager it will be like SELinux, strongly encouraged to be on but an optional capability that can be disabled. It probably will disabled first thing by a ton of users because it's too much effort to deal with (much like SELinux). Lack of stable driver ABIs in Linux make signed module requirement in some cases just too limiting to possibly deal with.

Re:No more nvidia for me

[trevelyon]

Challenge being, what Optimus like technology actually works in Linux? I'm unaware of anything off hand. When you have nothing to compare with, what are you supposed to do?

That is of course assuming I wanted an optimus solution. I didn't. At the time optimus was pretty much the only nvidia option if you didn't want a 14lbs gaming rig but wanted decent memory and a i7. I've used nvidia graphics for many years now because they above any other just worked. They don't anymore and the company came out and basically said "if you have optimus on Linux you're on your own". Well I heard that message loud and clear and I won't be going that route in the future.

Perhaps the life lesson is to look up the product before purchase regardless of vendor to understand what is going to happen.

That's what I thought. I checked for reviews or comments and scoured the internet for days trying to find a model that met my not so odd requirements. I've been running Linux for just over 10 years now and know the need to check hardware specs. It turns out most vendors provide even less detailed information (HP wouldn't even tell me what wireless chipset I would get when I called and asked). With all my requirements I could either go with a monster gaming rig (heavy) with an nvidia fixed graphics card or an optimus solution. I ended up after all the research selecting this because it supposedly had a hardware switch. Turns out it was a software switch and useless in Linux despite the vendor claims.

On the other hand, of the technology that *has* viable alternatives, how does nVidia stack up, capability wise, to the competition? In terms of accelerated 3D, AMD and nVidia proprietary drivers are both very capable, and none of the open source (including Intel GPU) comes close. In the case of Intel, you have to limit the comparison to similarly weak AMD/nVidia hardware, but even then it still lags significantly. One oddity though. With Mythfrontend, any attempt to use GL painter or renderer fails miserably with AMD. This either indicates an AMD bug or a MythTV bug, but the latter is still AMD's problem since clearly the developers do not target their platform.

In terms of video decode acceleration, nVidia VDPAU more or less stands alone. VA-API comes in second, but even with a brand new Ivy Bridge GPU, the experience is far more poor than VDPAU on a low end nVidia GPU that's 4 years old in projects that support both. VDPAU also has more project support. XvBA doesn't even warrant a mention, between the crappy capabilities and only one experimental branch of a major multimedia project bothering.

Video playback tear-free is a massive issue for everyone except nVidia. They are the only ones that do tear-free XV (though usually not needed due to vdpau). They also are the only ones I've tried where OpenGL renderers have worked consistently well.

I thought that too but now that I am basically running on intel graphics all the time and found I can live without the nvidia 3D. It or something causes problems in the only app I really used it for. The intel video works fine for all the video I watch (xvid, web, flash, even h.264). The 3D I wanted was to play an old game but installing the ia32libs on this graphics config causes me to crash to desktop. I haven't had to deal with that kind of instability since I left windows a long time back so ia32libs got removed which means I no longer need the 3D from the nvidia proc at all. In the end it was NVIDIA who taught me just how disposable they are. My future systems will go straight intel. I hope that works better. Even AMD's optimus like technology lets you switch between the graphics cards at boot, nvidia doesn't give that option.

Your mileage may vary but for me NVIDIA went from a sure thing to out of the picture with their whole handling of optimus. Potential kernel module signing just makes it even more likely to "find" more problems using their drivers in the future.

Re:No more nvidia for me

[diego.viola]

Just go with Intel.

Re:Of course

[JamesP]

Oh please, it's firmware whining again

Dear open source zealots, no one is obliged to add more circuits to a computer just because you can't be bothered to give the embedded card some data.

ALL computers depends on some amount of closed source data, deal with it.

There are exceptional cases, where the firmware is OSS, or where the vendor is a real asshole and forbids the blob to be distributed

Absolutely. Dear manufacturers, if you complain about distribution of fw at the same time anyone can download your "windows driver" off your website, You Are An Asshole

Re:Of course

[Anonymous Coward]

Have you even looked at the firmware in question? It really does things as "implement fencing point support" low level crap, and it is damn SMALL. It is more akin to cpu microcode than, e.g., NIC firmware (which is often actually machine-code for ARM/MIPS cores implementing higher level details that indeed would benefit from source code and modification), in that it implements stuff that is so low-level, it is not only simple enough to not be buggy, it is also of no general interest whatsoever as far as modifying it goes.

Yeah, it would be nice if it were open, but it doesn't matter very much if it is given a license that doesn't cause it to end up in Debian's non-free repo except for the lack of source (or you can actually document that it has no source, being a bitfield based microcode or something).

Do what works in other companies

[sqrammi]

They could pull an Atheros and hire several of the core open-source driver developers to their own team. Linux support for Atheros chipsets has been phenomenal ever since.

First limitation...

[Junta]

I haven't figured out a way with just the OSS tooling to correct for overscanning on my TV.

Re:First limitation...

[drinkypoo]

I haven't figured out a way with just the OSS tooling to correct for overscanning on my TV.

Sounds like a perfect place for nVidia to make a contribution then, doesn't it?

SUPPORT OPTIMUS

[brunes69]

The biggest problem with NVidia drivers right now is lack of Optimus support.

With every new laptop that has an NVidia chipset using this feature, this is not acceptable IMO, because it puts Linux users in a horrible position of having to choose between horrible battery life or horrible performance.

And yes, I have been trying out project bumblebee. But really, it is a giant hack and not a workable solution long term IMO, because it is the opposite of "seamless"

Re:Bug DB

[moderators_are_w*nke]

That's a really good idea.

Play for power from the Linux kernel team

[moderators_are_w*nke]

This is a power play from the linux kernel team. They don't like that someone else has control over a large part of their system and they want the source so that they can control it. As far as I can tell they are the only people with a real, practical issue with the status quo in that they're the ones using exprimental kernels that Nvidia don't support yet and obviously Nvidia drivers being under their control would make their lives easier.

I'm with Nvidia on this one though, it's their hardware and they have the right to maintain control of thier drivers if they choose to do so. Of course, as others have pointed out, we have the right not buy that hardware if we choose and those with an ideological issue with Nvidia's position will probably choose to do that but this clearly doesn't worry Nvidia too much.

Personally, I'm satisfied with Nvidia's model. Seems to work pretty well for me.

Re:Of course

[gbjbaanb]

you don't quite understand the problem here - the HW manufacturer can offer a single Windows binary (or 2 or 3 at worst). But with Linux, they have to recompile a version for every (well...) version of the kernel that's out there. If the driver system had a stable ABI that never changed, then the HW manufacturers could deliver a single binary built "for linux" and would expect it to work. Currently, they have no such guarantee.

You wonder why Linux on the desktop never took off? There's no business benefit to the manufacturers to support it. Sure, its technically better, but that means diddly squat in the real world especially when the answer is to recompile from source. You're never going to get Nvidia's source code, so accept that and start to deal with it with a technical solution rather than an idealistic one.

Re:Of course

[Anonymous Coward]

Only RMS and his rabid zealots give a fuck about that. Normal people have real work to get done.

And what do you do when a proprietary vendor changes something in their code that breaks your application? If it's closed, you're unable to fix it.

How many hours of "real work" do you lose while you try to resolve the issue or find a workaround?

You can call RMS and his followers as "zealots," but ideas have practical consequences. In RMS's case it was not being able to fully utilize a printer that prompted him to start the GNU project. That was a tangible instance of closed source "ideology" preventing people from getting their real work done efficiently.

Re:Ugh, this makes me Stallman.

[gbjbaanb]

I'd say you are.. apparently the drivers contain much of the magic that makes their cards word well, so opening up the source to the drivers would hand all their competitors a ton of advantage. Sure, I'd like to see that ;) but it isn't going to happen.

So start thinking how to address the issue that allows Nvidia to release a single binary that works across all Linux platforms and versions, then I can download the driver from their website, bundled CD, or have it packaged with their own repo and you'd get super fast graphics without any hassle.

Re: Some help is better than none.

[Crosshair84]

What is needed is an operating system that doesn't sh*t itself on update so nVidia can write a single Linux driver as they do with Windows/Mac and call it a day.

Real Time Linux

[trasgu]

My current peeve... how about a driver that works nicely with Real Time Linux. The word is that proprietary NVIDIA drivers are death to Real Time OS. I would purchase a NVIDIA card and use proprietary drivers if they had low latency, low jitter, and no memory conflict issues. Until then, I will try AMD/ATI and/or use KML drivers. Real Time Linux is used for device controllers, robotics, and other timing-sensitive applications. I am not a Freeware fanatic, I just want something that works.

Re:Of course

[gstrickler]

The lack of binary compatibility is a factor limiting linux mainstream usage, however, it does not need to be a factor for the manufacturers. The OSS community will happily build/repackage compatible binaries if the manufacturer will allow them to distribute things such as firmware blobs. No one is asking the manufacturers to do more work, they're asking the manufacturers to alter their policies on documentation and/or distribution so that the OSS community can do the work.

Re:Of course

[KZigurs]

Linux still uses X. Regardless of how far away it's abstracted with dirty ugly hacks, it's still X.

So - for desktop use - no, it's NOT technically better. And it has little to do with the drivers not being open, but rather the religious insistence of nutters to keep the useful, flashy, graphics, you know - desktop, bits out of where they could work.

Re:Of course

[bill_mcgonigle]

If the driver system had a stable ABI that never changed, then the HW manufacturers could deliver a single binary built "for linux" and would expect it to work. Currently, they have no such guarantee.

Is there no way to write some middleware that would at least allow the same operations to succeed across kernel releases? I mean, let the kernel guys do what they want, but chance the interface's back end to meet the kernel and keep the front-end stable.

Re:Of course

[fostware]

I'm pretty sure they run on OS X just fine...

A Framebuffer Would Be Nice...

[ilikenwf]

Seriously, if we have to use the blob, it may as well have a framebuffer like every other driver that exists. I have to use it for VMWare, otherwise it'd be Nouveau.

Re:"Just specs"?

[Sir_Sri]

If Linux had a platform share of 50% of the hardware market

That's a very interesting assertion... See linux has like 1% of the desktop market, and the server market is irrelevant in this discussion. But the GPU computing market is not. And nVIDIA and AMD are battling it out for systems with tens of thousands of high end GPU's in them, and those systems all run linux, or a variant on it.

It would not surprise me in the slightest if nVIDIA does 30 or 40% of it's high value business on linux as part of GPU computing clusters. It doesn't take a lot of customers buying 100 000 GPU's at a time to suddenly have a big market, and a big percent of total sales (of cards anyone cares about).

Re:Of course

[gbjbaanb]

you mean something like NDISWrapper that lets you run Windows binary drivers for network devices. We need something similar for Windows graphics drivers... well, if it won't be done natively, then yes, this is probably exactly what is needed/

Re:stable driver ABI

[cpghost]

Using NVIDIA's FreeBSD driver on FreeBSD/amd64 9.0-STABLE. Works okay, most of the time. Better than on Linux. Thanks for that, guys. My only real gripe with NVIDIA is their lack of FreeBSD CUDA/OpenCL support. That's REALLY holding FreeBSD back in the HPC domain.

Re:LOL slashdot

[kelemvor4]

Slashdot needs to lose the "news for nerds" tagline and adopt "reddit news from yesterday, today." instead.

Modded down for that? I thought it should be +5 insightful.

Re:Of course

[Anonymous Coward]

They could just get thier driver into the kernel tree. A fixed ABI would not encorage them to not do that.

No, please! There's already too much transient stuff in the kernel. I think the developers should be asking themselves why they feel they need to keep tinkering with the kernel. Lending some stability to the ABI would be a godsend.

Simple!

[Lisias]

Answer our questions!

Re:Of course

[Raenex]

Have you even looked at the firmware in question? It really does things as "implement fencing point support" low level crap, and it is damn SMALL.

Code is code, and even code that compiles down to a few kB at some point often requires a patch. It's quite common for new versions of firmware to be released that fix problems.

Yeah, it would be nice if it were open, but it doesn't matter very much if it is given a license that doesn't cause it to end up in Debian's non-free repo except for the lack of source

No source means it goes into non-free. There's no license that will change that. From the Debian Free Software Guidelines: "The program must include source code, and must allow distribution in source code as well as compiled form."

Re:Of course

[Raenex]

Oh please, it's firmware whining again

Oh please, it's another apologist for closed-source exceptions. Imagine that, people dedicated to open source want source instead of being required to distribute binary blobs.

no one is obliged to add more circuits to a computer just because you can't be bothered to give the embedded card some data.

Nobody is asking such. Just release the source instead of a binary blob. If it truly is data without corresponding source, then there is no problem. Of course, you know there's source there.

Re:"Just specs"?

[Sir_Sri]

Yep.

But computing cluster guys might be quite content with a proprietary driver that gets the job done, where the open source community isn't pleased with this solution.

Although having been in the GPU computing business relatively recently. You do actually want regular driver updates, or at least did at one point, since you want to take advantage of language features and performance improvements.

Re:Of course

[Mabhatter]

That is a fundamental Phillosophy problem.

I work in Midrange stuff and I understand exactly why the Kernel devs are so hard line. In our MRP system, we have a licensed copy of the source code. That code has run on CISC, RISC, and POWER IBM hardware with little input from the manufacturer in 15+ years. 90% of the time the code just needs to be recompiled under the new compiler version. SOMETIMES it as to be touched to change depreciated data types, etc... The box uses Object code, but some things still need the source TOUCHED and without that ability we would have been dead I the water so many times and shut our business down.

Fundamentally, it is a "I'll show mine, you show me yours" thing. Nvidia can download any of the Linux pices THEY need to make drivers work... It's FREE! It becomes "just too bad" that they can't share back. Kernel Devs want to work in Source Code or they will get cornered by the guys like TiVo-ize and don't give the MEANINGFUL parts back to the Devs.

An alternative

[Casandro]

Put a microprocessor on the card which accepts OpenGL commands (or something similar) from the computer and does whatever proprietary stuff you would usually do on the card. Then publish and open the interface between the host system and that controller, so every graphics card will look "the same" to the host system. Get the complexity out of the driver and into the "hardware".

The whole idea of closed source software is ridiculous anyhow. Don't you think your competitors haven't already analysed it in detail just like they reverse engineered your chips for years?

Re:An alternative

[lpq]

Um:

OpenGL Driver Support
Windows driver version 280.47 and Linux drivers version 280.10.01.04 provide full support for OpenGL 4.2 and GLSL 4.20 on capable hardware. This driver also supports several new OpenGL extensions for both 4.2-capable GPUs and older GPUs. The driver download links are at the bottom of this page.
(http://developer.nvidia.com/opengl-driver)....

Apparently, OpenGL support isn't enough...

 

Re:An alternative

[Casandro]

You misunderstood. The OpenGL support should move from the driver to the hardware so you won't need a proprietary driver any more. Essentially every graphics card would have (more or less) the same interface to the operating system. No hardware dependent drivers would be needed then.

Re:An alternative

[lpq]

But the openGL standard is constantly being updated... so ...you'd have to constantly ship new Firmware updates to stay up with the standard --
They would end up being 'binary blobs' as well -- so the problem would still be the same I'd think...

Re:An alternative

[Casandro]

Well the binary blobs would then be only firmware, but nothing which runs on your main CPU. This solves a lot of the problems. For example you get a security barrier between the proprietary code and your kernel.

I mean the obvious solution would be open source code, but nVidia doesn't want that.

Re:An alternative

[lpq]

What do you think your ROM BIOS is?

It's Firmware -- and it most certainly runs on your machine.

If they implemented a callable interface in Firmware, it would be a native

code or callstack -- but worse -- they have to deliver results back to user space and transfer large amounts of data over the bus -- that means DMA access to user, and likely kernel space -- they could still write all over memory and would be no increase in security unless the code was inspectable -- in which case you are back where you started.

Re:An alternative

[Casandro]

Well the ROM BIOS is executed by the main CPU. Of course if you want to have binary firmware on the PCI-Bus you need to have a way to restrict the access rights of the hardware. Again that's a lot harder than simply releasing the specifications for the hardware.

Re:An alternative

[lpq]

Absolutely -- that's why kernel maintainers want open source -- so they can inspect everything it does.

As it stands now, you have a driver that needs unconstrained privileged access in order for it to perform optimally. Thus, if it is not open source, one can't easily tell what it does with each new release -- does it contact the "TPM" for information about how secure the OS is? Does it use that information to introduce lower visual quality on your monitor? Did they only reduce it when playing restricted output? How do we know? This stuff is hard to get right all the time (for some, it's hard to get right even most of the time! ;-)).... Does it try to call network routines to validate the driver? or check for updates?...Does it try to use peer-to-peer networking ala MS style in order to circumvent firewalls w/tunneled IPV6? Etc etc etc.

I don't know why but Nvidia feels they have to keep their source proprietary -- it must really be awful in some way for them to have to hide it. If it is a proprietary and unique algorithm, patent it!... (or maybe they are using someone else's patented stuff and they don't want that known?)...

It's just awfully squirrelly for HW-Dev manufacturer to claim a need to keep their source to their driver "proprietary".

In a similar way, I ran into a weird situation with the Spyder 3 Color measuring HW -- you can't use the software w/o the device, so it's a bit like it already has a built-in HW dongle -- w/the exception that you can configure
the software, then unplug the device and the SW will still run -- but you won't be able to recalibrate -- something you need to do every few months, at least.... But now they added in a full "call-home" licensing suite to the software, so you have to activate it -- even though you can't buy it or copy it without the device... Just seemed a bit paranoid to me. It seems their licensing mechanism was already guaranteed with the need for the HW for the device to function. So why a need for proprietary SW licensing?

Situations like that make you suspicious about how truthful and forthcoming they are being with you -- and if they do it there, then we know they don't have a problem putting in shady stuff when their judgment says so... so without an OpenSrc driver, you just don't know what they are doing...and many Proprietary vendors (Sony, multiple times, et al), have shown us that given the opportunity, they will put rootkits on our systems and will disable our hardware at their control (like all the MS-Media-Center users who found a popular show didn't record that week because the broadcaster (NBC of the previous MS-NBC alliance), tested the 'do not record flag' (accidently, of course!) only during this one show -- as a metric to see how well the new blocking technology worked...

Sweet.

If anything, Corporations have shown us they are not to be trusted.

Re:How About...

[diego.viola]

My next GPU will be Intel, when I change my motherboard.

Re:Open it up a bit more than today..

[diego.viola]

They don't need to "create a new driver", nouveau already exists and it's pretty good, they just need to release specifications and/or start helping nouveau.

lol binary blobs

[Capt.Derp]

Release a seperate product that's completely open and give the open hardware specs to bsd and linux devs to work with, or open up all old products. Let the O/S developers drive the device no binary blobs. If you don't one of the Taiwan chip manufacturers will and you'll have lost out to hundreds of millions of future Android handsets. Then you can continue to swindle the computer gaming market with your ridiculously inflated windows closed source drive bubble and help future proof the company when that market inevitably disappears in a decade or sooner.

Re:Of course

[gbjbaanb]

unfortunately its not a Nvidia v Linux thing, Nvidia *doesn't* have to release anything for Linux. They can ignore the platform completely, which in turn means consumers ignore the platform too (because it has shitty graphics) and then Nvidia has no reason to work with Linux because it has no customers demanding Nvidia support...

but its not just Nvidia, if there was a stable ABI, any hardware manufacturer could bundle a CD with Linux driver software (and a little "works with" penguin icon) with their kit and anyone buying it would know it worked with their Linux install. Today, that's not the case, so the manufacturers standard mechanism of releasing software to consumers fails. They might know to rerlease the source and let the Linux guys recompile it and (hopefully) fix it for new versions, but that's not the way its done (and it puts an increasingly large burden on the distro compilers, wouldn't it be better if all these driver modules didn't need touching between versions?)

I like to think of this problem as a classic business v technical one, where the techs are not providing the business what it needs. Unfortunately, I think it really is a big part of the reason Linux is not more popular as a consumer OS.

Do nothing is the key.

[Stumbles]

This is just marketing bullshit from NVIDIA. IF they were really concerned or interested this announcement would never have been issued; instead all they needed to do was listen to their own developers and the kernel folks.

Re:Of course

[logicassasin]

Only RMS and his rabid zealots give a fuck about that. Normal people have real work to get done.

And what do you do when a proprietary vendor changes something in their code that breaks your application? If it's closed, you're unable to fix it.

How many hours of "real work" do you lose while you try to resolve the issue or find a workaround?

Dunno, but I know I've lost many hours trying to get a few FOSS drivers to work in the first place.

S3 Texture Compression is one problem

[logicassasin]

Many moons ago, S3 had one amazing piece of tech in their extremely underwhelming Savage4 and 2000 video cards: S3TC. With it, a Savage4 could muster some of the most gorgeous 3D images of it's time, particularly in Unreal Tournament when the 2nd disc full of S3TC specific textures was installed. While still slow as a constipated turtle shitting molasses, visually a Savage4 was far ahead of the TNT2's and GeForce256's of the day on a game that supported S3TC via S3's MeTal API.

Nvidia licensed S3TC (as did many other companies) for their products many years ago, and it's still proprietary. This is but one hurdle facing a completely FOSS driver.

Dealing

[DrYak]

And by "dealing" we mean "using the same software facilities used by anyone else and which are in the kernel for ages", not "trying to reinvent the wheel each time, so you can re-use the same wheel as in Windows instead of playing nice with everyone else".

THIS!

[AlexiaDeath]

XRandR (REALLY-REALLY THIS! I cant use my colorhug without cripling my nvidia driver and switching to nouveau because of XRandR fail) and if you dont want to make your driver work with old cards, help NOUVEAU support them properly with just the knowhow, mainenance will be free.

Software facilities

[DrYak]

*NVIDIA makes good binary drivers, but those have problems when a new kernel version comes out with changed interfaces: Only NVIDIA can adapt them, and until they get around to it, NVIDIA may not work with the latest kernel version.

In addition to that, Nvidia tends to do things in their own way in order to share code between the various OSes.
And some feature can't be implemented in the Windows-y way on Linux.
On the other hand, the linux kernel provides plenty of software facillities, thus increasing code share between drivers. Nvidia tends to ignore these. (For example, Linux has standard interface for handling multiple monitors, changing the resolution, and so on: XRandr. Until now (version 302) all that the nvidia drivers could do was only their own TwinView, which is proprietary, has nothing to do with XRandr, doesn't play nicely with other Linux software. But is the same technology as what they do on Windows - Meanwhile Xrandr has been available on ATI for ages, both on the proprietary and opensource drivers).

The Optimus situation boils down to that.
- A new generation of optimus laptop has an embed Intel gpu which is always on, and which is permanently connected to the display. It has also a discrete Nvidia GPU. Unlike older dual GPU laptops, it lacks any electronics to switch the output from one GPU to the other. The Nvidia GPU will never see a monitor connected to it, it stays always headless. Instead the Nvidia GPU has to render its 3D ouput on the frame buffer of the Intel GPU , which will display it.
- The Linux operating systems has several software facilities which could make such an "shared framebuffer" architecture workable (both as in kernel code to route ouput, and a modular Gallium3D driver stack) - in fact, current experimental implementation of such software works to do crazy stuff. Like hot plugging an USB attached display adapter and have the computer's high performance GPU off loads the 3D rendering and then stream the graphics to the external 2D LCD.
- But the Nvidia people don't use these because they differ massively with how they deal with these things in their common OS-shared code. End result. no Optimus for you, which can in some circumstance even mean no graphic output for you at all.

Nvidia's answer? Go see Bumblebee.
(Bumblebee is a hack by a small 4-man team [kudos to them!]. Basically it will start on-demand a separate X server on the Nvidia hardware and use VirtualGL to copy back the visuals on the main X Server running on the intel. Users can select which software to run on the real Intel X server, and which software to remote to the Nvidia X server. It can solve the optimus problem in a few situations. But that is still a hack. And that's because Nvidia has a closed driver that can't be fixed, so volunteer have to make such hacks aroudn the unfixable problems)

And as far as I've heard, the Tegra situation isn't any better. Nvidia keeps doing things their way.

*Proprietary drivers - 1) monitor upcoming kernel builds and proactively update drivers before the next kernel release or 2) have a dedicated nVidia contact to work on updating drivers ASAP when notified that an upcoming kernel build breaks them

Well that would be the strict minimum. It would be even better if Nvidia could at least try to play nicely along with what's done by everybody else.

We have a real clash of mentalities.

On the linux side, we have developpers which like consolidation, code sharing and so one. If some function has to be performed by more than 2 modules, the developers dream is to have that function move out into a common module with a nice API to export such functionnality to anyone who can need it. (for exemple, in storage, the long term is to consolidate all the various data redundacy functions (MD linux software raid, vendors sfake raid, LVM stripping, DRBD network redundancy) into the same facility - currently LVM and Fakeraid alrea

Won't work for everything

[DrYak]

Depending on how much functionnality you want from the card, that might not work.
Several functions (the Optimus situation being a prime example) require a lot of collaboration between lots of component.

The very small open drivers, begin to just be a small interface for passing commands to the hardware. But then you start to need initialisations in the kernel (KMS - Kernel mode switching) for various good reasons (suspend/resume, error message, graphical console). You end up with the 2D support being opensource and the closed source component being only the OpenGL driver.

But then (specially for SLI, Optimus, etc.) you need to be somewhat able to share states accross drivers and graphic cards: you need to use the in-kernel memory management facilities that the other use, you nee d to use the modular architecture of Gallium3D, so the opensource part gets even bigger and spans user space too, while the closed source part is even smaller.

At that point the closed part is nothing more than a thin layer which exports basic hardware function blocks (shaders, etc.) while everything else is handled in the common part (memory management is handled by the kernel, openGL and other APIs are handled by Gallium3D, etc.)

And then, the closed source model start to lose its advantages: you won't be able to share much between OSes (Gallium3D modular architecture hasn't much in common with WDDM). You won't be able to exploit all your small private secret über optimisation because you have to play nicely with API state trackers in Gallium3D and play nicely with all other drivers cards.

That is something which was understood by Intel : They moved completely to the opensource stack and asked Tungsten to write their driver for linux as opensource. It uses KMS, it's been on the fore-front of memory management facilities (TTM, GEM, etc.), Gallium3D was developped *by* people at Tungsten (although, curiously Intel themselves only started using it recently for the latest hardware, the rest still uses the older stack in Mesa).

That is also something wich was understoof by ATI: They offer both a Linux port of Catalysts (which might not be as stable and great as Nvidia's blob. But on the other on the other side plays nicely with some standard in Linux like Xrandr), and they support opensource drivers (release docs, pay developers). We've reached the point where for old hardware (up to r500 / Radeon X1nnn) the opensource drivers are the best solution for stability AND performance. And newer hardware (R600 and up / Radeon HD nnnn) is more or less functioning, with very good stability and with acceptable performance for everyday work.

Nvidia on their side prefer clinging on their advantage of closed source (shared code accross OSes, über optimisations) which make them win on ease of development, but make them lose on the cooperation with the rest of the ecosystem: their driver doesn't play that nicely with kernel updates. their driver doesn't play nicely with other driver/hardware and some feature just can't be make to work (Optimus). their driver doesn't play nicely with the current standard and some feature work bad with some software (TwinView vs. Xrandr). and their monolithic drivers make kernel developpers want to cry (from what i've heard the Tegra situation is worse).

Intel GPU != server

[DrYak]

and the only reason Intel drivers work is because of the server market and because chips don't change often.

Well, you're wrong. Foremost, Intel GPU are at most deployed on desktops, server tend to use either even older GPUs or use GPUs with embed VNC server for remote KVM.

But Intel has been quite active on the GPU in kernel. They have helped create or sponsored (through paying Tungsten graphics to write opensource driver) a lot of development.
Kernel mode setting has appeared thanks to joint collaboration between all (non-Nvidia) developpers and that helps a lot for example for suspend/resume to work.
Intel (through Tungsten) has helped in-kernel gpu memory management (TTM, GEM, etc.)
Tungsten developpers (paid by Intel, then by VMware) have created the modular Gallium3D stack (which use separate font-end for the various API like OpenGL, openCL, etc. there's even a dx10/11 frontend. And a separate back-end to interface the hardware). The opensource ATI driver use it, Nouveau use it, newer Intel hardware (and alternative driver for older intel hardware, in part sponsored by Google) use it.

Yup there's a lot of development being done on the Intel front. They are not just sitting around with the same old 10yo driver and juste adding new PCI IDs from time to time.

It's not rocket science, writing graphics drivers are HARD and Linus continually poking his pecker into the kernel means they continually break and have to be tinkered with to work again. {...} It needs to be repeated so it sinks in, Linux needs a reliable ABI so ATI and others can just slap a Linux driver onto the CD that comes with the card and call it a day.

Nobody said that graphic drivers are easy. Nobody said that once AMD published specs, perfect drivers will insta-magically show up on the next day. (Although by now, the opensource AMD drivers are the best thing for Radeon X cards, and work well enough (not the best perfomance, but work stable) for most Radeon hd except the latest generation).

And if "linus is poking his pecker" (well actually not. He doesn't do much GPU development. the "pecker poking" in the GPU world is currently done by people on AMD's, Intel's, VMWare's, Tungsten's payroll) that because *technology is changing*. keeping a stable ABI would also mean that we stick to old paradigm and old technologies. People aren't just breaking Linux for the sake of breaking it. Their are breaking it because old cruft needs to be removed, new capabilites needs to be added. Disparate stuff needs to be consolidated in common. Linux is a moving target, because technology is moving on.

(And in fact, it's not just Linux. Microsoft has also needed to break compatibility in several point in time in their Windows family. You don't notice it as much, because overall Windows development tends to move at the speed of a glacier, specially in the post XP era. But for exemple the graphic driver infrastucture between XP and Vista changer completly. For exactly the same reasons as it is also changing in Linux: bring in modularity)

If Linux sticks with a stable API/ABI, it will be stopped in it's development. (You like how there isn't any problem on laptops by suspend resume with opensource drivers? Well that's because KMS was introduced, remove the needs for a crazy ping-pong between kernel and userland x server for state saving and re-initialising).

It's because Nvidia doesn't use the newest feature of the kernel that Optimus isn't working.

Also binary-drivers-on-CD have the problem that, when there is a big change needed in the kernel and in the API/ABI, drivers won't get updated.
- On linux, substem team can make sure that if their subsystem changes somewhat, all their drivers can be adapted to work with the newer infrastructure.
- On windows, not every single manufacturer upgraded every piece of hardware from XP to Vista.

To be able to keep up with such a moving target, hardware manufacturer need to :
- either actively tra