Intel Open Sources Graphics Drivers

PeterBrett writes "Intel's Keith Packard announced earlier today that Intel was open sourcing graphics drivers for their new 965 Express Chipset family graphics controllers. From the announcement: 'Designed to support advanced rendering features in modern graphics APIs, this chipset family includes support for programmable vertex, geometry, and fragment shaders. By open sourcing the drivers for this new technology, Intel enables the open source community to experiment, develop, and contribute to the continuing advancement of open source 3D graphics.' The new drivers, available from the Linux Graphics Drivers from Intel website, are licensed under the GPL for Linux kernel drivers, and MIT license for XOrg 2D & 3D rendering subsystems."

Comments

Wow.

[bobintetley]

This is a great move by Intel - I know which vendor I'll be picking for my next 3D card. I HATE that I only have the choice of Nvidia or ATI's "mystery binary blobs" to play games.

Re:Wow.

[Anonymous Coward]

A hell of a lot is an understatement.

Fact is that Intel is the #1 manufacturer and supplier of graphics chipsets. The vast majority of PCs off the shelf at Best Buy and the like that are marketed towards Joe Average are Intel graphics based, as well as those sold direct via Dell. ATI has made some significant inroads, particularly since AMD-based systems have become more prevelant on retials shelves, but Intel still surpasses them in market share by a significant amount. Fact is that most Intel CPU-based off-the-shelf systems have Intel-based graphics as well.

This whole move is simply to put pressure on AMD/ATI, which Intel is now viewing as a viable threat to their marketshare. It will proabably work too, to a certain degree. The next step will be Intel buying NVidia. No, I have no substanitive argument or proof of such a thing, but it will happen.

Happy now?

[mobby_6kl]

I can't say I particularly care (not using any on-board graphics), but this is a nice move on their part. Also, it would be interesting to see how this affects the performance/features in the long run.

Competition from AMD/ATI?

[thre5her]

Hopefully AMD/ATI will compete by open-sourcing the drivers for their integrated chipsets. Some healthy competition would definitely help the Linux desktop.

Re:Competition from AMD/ATI?

[Mr. Jaggers]

That's a stupid excuse, though. They could always isolate the SGI-laden parts, LGPL the rest, and let the community at least have a fighting chance at replacing what's behind the proprietary API's. I'm not claiming that our homebrew routines would *ever* be better, but I suppose it is within the realm of possibility. Oh, and when I say "always", I do really mean *always*... at any point, even right this minute, they could do so.

The non-licensed parts of the code don't have to compile to be released. Besides, when bugs are traced back into the dark proprietary code, that would also make ATI the good guys and SGI the bad guys. ATI could claim that the licensed part is really fast and awesome and sweet, but proprietary, and that the community is welcome to try and replace it with something fast and awesome and
sweet, but open. Or even something slow and crappy, but rock-solid stable, that plays nice with Xorg and the kernel.

I suppose they might have licensed other companies code and signed away their right to ever release any code they ever write that uses the licensed bits. That would be a collosal blunder, but would partially account for silence on the subject.

I'm fairly certain that the real reason lies not the code ATI has licensed, but the code/tech they've worked hard on and feel they need to keep secret or else lose their edge against nVidia. Of course, it seems that same statement could be made, swapping the names of the two companies, and still be true. In fact, the "trade secret" and "intellectual property" argument is almost certainly the biggest reason for closed-source driver code. Besides, how can a company who is losing money afford to give anything away for free? At least it always seems like the investors and board of directors of tech companies seem to believe that they are perpetually bleeding cash, even when they file record profits with the SEC.

Anyway, that's quite enough ranting and unsubstantiated libel for one post.

Re:Competition from AMD/ATI?

[Cyno]

I agree, and most people don't need anything faster than an Intel 955, even the 855 is good enough for 90% of desktop use minus modern 3D gaming. They play quake just fine. What more does one NEED, honestly? ATI and nVidia better wake up or they may soon find a new real competitor on the block.

I bought Intel graphics with my laptop. At first I wasn't pleased with the performance, but then I got to testing it directly. I can easily get 30 fps in OpenGL for simple geometries. Its really not that bad. They doubled the performance since, and I'm sure their latest stuff is most useable. Can you imagine what they'll come out with next?

I didn't like Intel, but lately they've been attracting my pocketbook more than any other anti-FOSS businesses. As far as I'm concerned if they aren't pro-FOSS by now, they're anti-FOSS. They know just as well as I do what its all about. Microsoft, no matter how much they say they support it, is obviously fighting it tooth and nail behind closed doors.

Re:Competition from AMD/ATI?

[ookaze]

Also, that document is a complete lie. I don't care that it's in the kernel tree. There's lots of wrong stuff in there

BS, the document is not a lie, the document provides an explanation. An explanation can be false, it's still not a lie, just a bad explanation.
And sorry, but I think GKH has way more authority than you on what is right or wrong in this explanation, as he did lots of the drivers in the kernel.

A driver does not have to be in the tree to be stable, running driver, and the driver being in the kernel tree doesn't mean that it is either stable or running

Yeah right. Meanwhile, real life shows us that what you describe is exceptions rather than a rule.
The driver being in the kernel means you can bug the Linux kernel devs to make it work with each new release of the kernel (hence stable).
The driver not being in the kernel means they won't do anything about it, and you have no way of knowing if the driver will work or not.
The basic premise is that the maintainer of a driver would support his driver in the Linux kernel tree.

And I should know, as I have written multiple closed-source Linux device drivers, two of which have open-source versions in the kernel that have at various times either not worked, or worked poorly, and both of which perform signifigantly worse than the closed version

Now I wonder how you can have the guts to write that. So you basically admit that you do closed drivers that have equivalent Free Software ones, though they used to be worse in Free Software version. And then you complain about the unstable API document and want to be taken seriously ?
But you know what, I'd rather praise the guys who made the FOSS drivers. Of course they were worse at first, but now, we have correct free drivers, and that's way better than being stuck with hope of endless support for the closed ones. Support that you say is better, but we have no way to know if that's even true.

Go actually read that document. The argument it makes is that a stable kernel/driver API is a bad idea because the kernel/driver API is unstable. It's a circular argument

BS, where is the circular argument ? It explains quite well why the kernel/driver API is unstable, and no, it's not because "stable kernel/driver API is a bad idea", which would make a circular argument. Go actually read the document.

One, there isn't enough agreement amongst the diversity of kernel developers to ever come up with a stable API

BS, the main reason is discussed in the document, and history has shown the document is right.

two, there is no dicipline amongst the people in charge to maintain that stability even if a consensus was reached, and three, there are some who would like to keep the interface unstable merely to keep this argument for open source drivers valid

What you say is just a pure troll. The discipline is to make the kernel work better no matter what, and to not get stuck by in-kernel stable interface, which you see as an issue. You just can't accept that, that's your main issue.

Re:Competition from AMD/ATI?

[ivan256]

Yeah right. Meanwhile, real life shows us that what you describe is exceptions rather than a rule.

That's only true if you use popular hardware.

The driver being in the kernel means you can bug the Linux kernel devs to make it work with each new release of the kernel (hence stable).
The driver not being in the kernel means they won't do anything about it, and you have no way of knowing if the driver will work or not.
The basic premise is that the maintainer of a driver would support his driver in the Linux kernel tree.


In the real world, all the smart users have vendor support to take care of this issue for them. As linux popularity grows, the number of people using a non-vendor kernel shrinks. A tiny minority of linux boxes run Linus' tree.

The discipline is to make the kernel work better no matter what, and to not get stuck by in-kernel stable interface

Not only is that the lie, but it is provably wrong. Look how many operating systems there are out there that are successful and have excelent driver support even though they don't have the flexability of changing the driver API every release and the drivers aren't built with the kernel. I can count seven that are on the market right now. If this was really about making the kernel better no matter what, they would add a stable interface, because that would be an improvement. Drivers in the tree could continue to exploit new features and wouldn't be stuck using the old interface as the document implies. There is a history of this happening in the real world, and even in the linux kernel, as in the past, the Adaptec SCSI driver was shared code between Linux and BSD, with a compatability layer inserted.

I can appreciate the goal of being good over being stable and or popular, but there are some parts of 'good' that only come after you are stable and popular.

And this is still moot in relation to this discussion, as graphics drivers are written for X an not the kernel.

Now I wonder how you can have the guts to write that. So you basically admit that you do closed drivers that have equivalent Free Software ones, though they used to be worse in Free Software version. And then you complain about the unstable API document and want to be taken seriously ? But you know what, I'd rather praise the guys who made the FOSS drivers.

Yes. And I don't see why that's surprising to you, unless you buy into Stallman's 'all software should be free' trash. I didn't say I don't have any open code. I simply don't have the luxury of being generous when it comes to this stuff most of the time. I have to make a living, and if that means taking a contract that is uder NDA in exchange for a paycheck, so be it. Pull a 2.4 tree and look in the ppc branch at the galileo code. Check the mailing lists for alpha, ppc, and just plain linux-kernel. When my contracts allow, the code goes into the tree.

It hardly matters anyway, because the only part of the system that really needs to be open is the platform, and none of this changes my main argument, which is that it's not the status of the source license that is causing the driver issues with linux graphics.

first reaction:

[mihalis]

Fantastic. Great work Intel. This puts your products in a different, more positive light for me personally. This could be really good for X11. I worked with it for about 10 years and have been very despondent about its chance in a world of proprietary drivers from ATI and NVIDIA being the only way to use modern graphics hardware. Maybe there's a chance for open source desktop after all.

Linux Laptops!

[db32]

Ok here is the thing...ATI and nvidia can be a bit of a pain...but on a desktop you buy one or the other and you plug it in and go. Laptops on the other hand your selection is FAR more limited and you have to juggle hardware, and more often than not, something just won't work right or well. This makes the Intel integrated laptops even more attractive now instead of the ATI/nvidia ones. I really hope they go backwards with this to and open their recent chipsets up completely as well.

Re:Maybe we'll start seeing Intel graphics clones.

[Tumbleweed]

I'd be willing to bet the REAL reason they don't open their drivers is because they're using stuff they know is the intellectual property of others. Just a guess, though; I have no real information on this, but I'd be very surprised if they can't dig into each other's hardware under a microscope to figure out what the other guy is doing, and reverse engineer each other's drivers. These are some very smart folks we're talking about here.

Kudos to Intel!

[Anonymous Coward]

This is good news. Open Source won't fix a bad product (hello Netscape), but you can have an army of eager (unpaid!) geeks happily extending your product. The idiocy of companies that hold their driver source proprietary is beyond belief; Does nVidia and ARI really seriously believe it gives them an advantage? Hardly. nVidia's drivers are buggy and crash prone. I am sick of my nVidia card hanging, and the saps at nVidia's support merely send you an automated email "Have you installed the latest driver." Yes, and it also crashes. If I had the source, I could fire up MSDEV. But I don't.

Intel made an earlier foray into 3D with the i740 which didn't do that well in the marketplace. But now they're back, and this is a nice first step. If they drive nVidia and ATI (and especially nVidia) out of business, I wouldn't shed a tear. Truth is even Microsoft by taking over Shaders with HLSL has done a better job that nVidia with their proprietary Cg language. Open sourcing their drivers shows good faith. Come on Intel!

I think we're missing the point here.

[Ruff_ilb]

Many, MANY home users out in the field use on-board video for everything. Now, I'm not saying this'll have them all converting to an Open Source OS, but this is yet another advance that would make sending the average noob user over to Linux without any sort of performance hit.

Taking a 180 degree turn and looking right back at your interpretation of the story, I find it very likely that Intel will be teaming up with nVidia sometime soon. Now that AMD owns ATI, Intel should be wide open to purchase nVidia if they want, and (although I'm not saying they'll need it), pairing Intel's massive resources with nVidia's enthusiast motherboard chipsets and universal video options, things could improve rapidly for the both of them. However, if Intel is going to enter the market as a third video force, that seems unlikely, although we could see Intel graphics cards interfacing well only with intel boards and intel CPUS, and the customer could likely lose if such a situation becomes possible.

Anyway, I think I've speculated enough. The bottom line is that open-sourcing these drivers is a very interesting and likely harmless move for intel to make, and it should make the jobs of many OS coders easier in the open source OS circles.

Re:I think we're missing the point here.

[Ruff_ilb]

INTEL
Type: Public (NASDAQ: INTC)
Founded: 1968
Location: Santa Clara, California, USA (incorporated in Delaware)
Key people Paul Otellini, CEO
Craig Barrett, Chairman
Industry Semiconductors
Products Microprocessors
Flash memory
Revenue $38.83 billion USD (2005)
Operating income $12.1 billion USD (2005)
Net income $8.7 billion USD (2005)

NVIDIA
Type: Public (NASDAQ: NVDA)
Founded: 1993
Location: Santa Clara, California, USA
Key people Jen-Hsun Huang, CEO
Industry Semiconductors- Specialized
Products Graphics processing units
Motherboard chipsets
Revenue $2.375 Billion USD (2005)
Net income $302.5 Million USD (2005)
Employees 2,737 (2005)
Website www.nvidia.com

Check out those rows, especially Revenue and Net income. Intel is a MUCH larger company.

Compare to

AMD
Type: Public (NYSE: AMD)
Founded: 1969
Location: Sunnyvale, California, USA
Industry Semiconductors
Products Microprocessors
Revenue $5.848 billion USD (2005)
Net income $165.483 million USD (2005)
Employees 18,100 (Nov 2005)
Website www.amd.com

ATI
Type: Public (NASDAQ: ATYT)
Founded: 1985
Location: Markham, Ontario, Canada
Key people David E. Orton, CEO
Industry Semiconductors
Products Graphics cards
Graphics processing units
Motherboard chipsets
Video capture cards
Revenue $2.222 Billion USD (2005)
Net income $41.676 Million USD (2005)
Employees 3,469 (2005)

Ati, suprisingly enough, has MORE employees than nVidia, an essentially equivalent revenue, and a higher next income.

If AMD can buy ATI, Intel should be able to buy nVidia with little problem.

Which brings up the question...

[japhering]

Are they making my plans to open source the rest of their graphics drivers ?

Talking out of both sides of my face,

[MrCopilot]

First, Brilliant move. They know, they just know AMD is going to blow open wide the company formerly known as Ati's drivers. They drop this announcement before the paperwork is even dry on the AMD/Ati deal. Bravo, kudos, well played... etc.

Second, Thank You Intel, so very much.... BECAUSE Even the laziest of our part-time hobbyist programmers will be able to improve your driver performance.
All these years I just refused to believe Intel could develop and ship newer and newer Card/integrated Video chips that were lightyears behind in performance and features. I instead chose to think of them as a Hardware Company full of Hardware Engineers who look down on the few "soft ones". I can understand how that might develop there.

I believed, some day, they would come around, and hire some PC Software/Driver Engineers. Someday the driver would rescue their possibly brilliant designs.

Well this is even better. We get our open graphics card with every e-machine.

Except, Of course Intel doesn't pay for it and yet reaps the rewards, and naturally perpetuates the undervalued view of us software guys.

Vicous cycle.

/rant heh, And then there were 2.

Re:Talking out of both sides of my face,

[ewhac]

Second, Thank You Intel, so very much.... BECAUSE Even the laziest of our part-time hobbyist programmers will be able to improve your driver performance.

Erm... I doubt it.

For the past few years, off and on, I've been porting the XFree/Xorg Intel 8xx graphics drivers to BeOS, so I have a fairly close relationship with that code, and unusually detailed knowledge of the chip series. Unless this represents a completely different codebase (which I doubt), it's really not that bad. Unless you're planning on turning it into a full kernel-mode driver, taking advantage of native interrupts and so forth, there's not a lot that could be improved.

The most annoying part with this driver release is that it still needs the BIOS to set display modes. BeOS can't access/execute the BIOS, so the driver has to be full native. I'll probably still have to do some fairly icky things to make it work...

Schwab

does the 9250 driver really work?

[j1m 5n0w]

Actually, the 9250 is the fasted fully supported ATI card under Linux.

This is anecdotal, but my experience (as of a couple months ago) is that the ATI 9250 SE doesn't work doesn't work properly with the open source driver. It renders, but appears not to be double-buffering. The screen flashes in a very ugly manner and I get to see frames of partially-rendered geometry. If I remember correctly, I got similar behavior with a radeon 7000. Currently, I'm using a cheap Nvidia card with the binary drivers (which seem to work ok this time around; the previous release would hang the machine if I tried to run any of several of the xscreensaver modules). This is with a 64-bit dual-core processor; 32-bit may work just fine.

Re:OpenGraphics is not dead!

[castle]

Your second paragraphs illustrates the thing stopping me from really considering it. However if Intel were to release a separate card they would most likely grab me as a customer. I'm fairly disgruntled about NVidia cards right now, but I was sick of starving myself for the few games on linux that required a spiffy graphics card.

Perhaps there's hope that AMD can open up ATi's drivers, unlikely, but perhaps.