AMD Launches New ATI Linux Driver

Michael Larabel writes "AMD has issued a press release announcing 'significant graphics performance and compatibility enhancements' on Linux. AMD will be delivering new ATI Linux drivers this year that offer ATI Radeon HD 2000 series support, AIGLX support (Beryl and Compiz), and major performance improvements. At Phoronix we have been testing these new drivers internally for the past few weeks and have a number of articles looking at this new driver. The ATI 8.41 Linux driver delivers Linux gaming improvements from the R300/400 series and the R500 series. The inaugural Radeon HD 2900XT series support also can be found in the new ATI Linux driver with 'the best price/performance ratio of any high-end graphics card under Linux.' While this new driver cannot be downloaded yet, in their press release AMD also alludes to accelerating efforts with the open-source community."

Comments

Is the driver open-source?

[Anonymous Coward]

If not then forget it. I will only go with those who release their source code. Hell, for all we know there could be a root-kit in the driver.

Re:Is the driver open-source?

[Anonymous Coward]

Here, you dropped your tinfoil hat.

Re:Is the driver open-source?

[spectrokid]

98% of all Linux machines are used for tasks where 3D graphic performance doesn't matter. Another 1% are people who jumped from Unix/SGI and know what they are doing/buying. There are not that many graphic-intensive games for Linux. This will only be important for people who want to run Aero-style 3D desktop environments. Your local banks webserver won't run it. Don't worry.

Re:Is the driver open-source?

[Anonymous Coward]

There's tons of games that run on Linux that tax 3D graphics to no end...

UT2004, doom3, quake4, .........

Not to mention Cedega offering options for 'windows-only' games

Re:Is the driver open-source?

[AJWM]

There are not that many graphic-intensive games for Linux.

Not many, true, but the FlightGear flight simulator is the main reason I upgraded from a cheap generic graphics card to an ATI 9250 based card (the highest level card with FLOSS drivers based on specs ATI released back when they were doing that). (And yeah, compared to current state of the art graphics cards, 9250s are still cheap.)

I wouldn't mind something a little faster, though.

Re:Is the driver open-source?

[CastrTroy]

Exactly. I got a laptop with an Intel GMA. Not a powerful video chip, but it has enough power to do all that 3D desktop stuff. And there was no fuss getting drivers. No extra stuff to download. No configuration to do. Everything just worked. For all my new computers (for the foreseeable future, until other graphics cards manufacturers release good open source drivers), they will all be using Intel GMA, because these video chips are good enough for my uses, and the drivers are extremely solid. If I want to play video games, I'll use my console (Wii).

Re:Is the driver open-source?

[KDR_11k]

There are not that many graphic-intensive games for Linux.

Nonsense, while the list isn't as big as Windows's there's still a fair number of graphics intensive games on Linux (though admittedly there may not be any ones that are so current that they absolutely need the latest hardware). Even just playing Doom 3 or UT2004 needs a 3d capable driver.

Re:Is the driver open-source?

[riffzifnab]

I don't think he dropped it. I don't think he trusts the tinfoil manufactures not to collude with the government and put small holes in their foil to permit the government's rays through. He's going to beat them through shear force of will by not thinking anymore.

Re:Is the driver open-source?

[Anonymous Coward]

Is there a particular reason you need the driver to be open-source? Are you planning on re-writing it? I've been hearing the open-source arguement for years but never understood it. Unless you are a driver developer yourself or plan on going through the code line by line, what does it matter? Why should a commericial company release their intellectual property for free when there are hundreds of people creating these products for a living. This is completely different than FOSS. What if you started giving away what you did for a living? /rant off

Re:Is the driver open-source?

[bvimo]

Don't forget GPL Nexuiz from http://www.alientrap.org/

Re:Is the driver open-source?

[Anonymous Coward]

Don't take it! It could be wired!!!

Re:Is the driver open-source?

[Anonymous Coward]

It's a trap!

Re:Is the driver open-source?

[BrainInAJar]

What about people like me that use Solaris? or any othe esoteric operating system other than the big-3 ?

or if there's strange bugs that you think are the drivers fault, and you happen to know enough C to fix them right now instead of whenever the snail-slow vendor gets around to it?

as for your comment about giving away what you do for a living... AMD doesn't write drivers for money, they make hardware. Intel manages to make hardware and open-source a good majority of their drivers, so that's just a stupid argument.

Re:Is the driver open-source?

[Anonymous Coward]

Spoken like a true button-pusher who has no clue how technology works, but who can push buttons.

Homer Simpson, is that you?

Re:Is the driver open-source?

[chris_eineke]

That wasn't a tinfoil hat, it was Sony/BMG music CD.

Re:Is the driver open-source?

[kc2keo]

Also... there could be DRM and a r00tkit. Who knows? Trust noone! :-P :-O

Re:Is the driver open-source?

[Spokehedz]

Yea, because you would be able to find the rootkit in a driver with thousands of lines of code...

Re:Is the driver open-source?

[Cajun Hell]

And that's a shame, because people probably really ought to keep their hats on.

Re:Is the driver open-source?

[Anonymous Coward]

Okay, I'll bite. Would you even know what to do with the source if you had it? Could you make sense of any of it beyond programmer comments? Or are you just talking out your ass because you expect everything to be free and open?

Re:Is the driver open-source?

[The Warlock]

Accidental security hole, deliberate rootkit, what's the difference, right?

Oh, wait, I forgot: fuck Heinlein's Razor; every mistake is deliberate! nVidia is consipiring with Microsoft and Halliburton to take control of your computer and obfuscate the true nature of The Milkman. It's all clear to me now!

Re:Is the driver open-source?

[nuzak]

> Is there a particular reason you need the driver to be open-source?

So it works when the kernel changes their *&^!%@! ABI yet again in the latest patchlevel. To port it to other OS's. So smarter people than me can look at it and find bugs or interoperability problems with it and send vendor updates to it.

I can understand their reasoning -- video cards are more or less big FPU arrays these days, and the actual 3d graphics is all software, so they might not want to expose their secrets. The other problem is that the competition would use it to find potential patent infringement. It's a Nash equilibrium: the first one to open-source loses. If I were to put the number generously at 50,000 extra customers due to OSS, that simply wouldn't cover the potential loss. But the fact is, there aren't any solid numbers as to what the market effect would be, and uncertainty is in a lot of ways worse than outright losing -- at least you can write off the latter on your balance sheet early.

Re:Is the driver open-source?

[GreatBunzinni]

Is there a particular reason you need the driver to be open-source? Are you planning on re-writing it? I've been hearing the open-source arguement for years but never understood it. Unless you are a driver developer yourself or plan on going through the code line by line, what does it matter?

You sure are distracted. By releasing open source drivers it is possible to support (i.e., new features, bugfixes, security fixes) the same hardware in a larger variety of platforms, some which were never supported officially and others that the hardware maker simply dropped the support. You may not notice it if you are that kind of user who adopted the dominating platform and needlessly upgrades hardware and software every year or so. If someone uses a niche platform along with outdated hardware, you bet that open source drivers are important. They are a life saver.

Oh and by the way, not everyone needs to be "a driver developer planning on going through the code line by line" to make the case for open source drivers. It takes a single user to write a patch that influences the entire community, which means that an isolated contribution by a random user can benefit the whole project and community. For example, do you ever heard about a project called linux?

Why should a commericial company release their intellectual property for free when there are hundreds of people creating these products for a living.

Really? Who, exactly is "earning a living" selling standalone ATI drivers? AMD sure isn't doing it. Oh you mean no one?

This is completely different than FOSS. What if you started giving away what you did for a living?

Just because someone may have found a way of getting paid for doing something it doesn't mean that anyone who intended to do the same thing for free should be barred from doing so. Do you believe that those bastards working on all FLOSS operating systems that release everything for free should just stop doing that just because some poor sap decided to try ask for money after providing that exact same service? Moreover, why exactly do you believe that locking down a driver is a decision that helps the hardware's paying customers? Do you believe that no one should offer something for free, something that undeniably improves the life of the community, just because a poor sap somewhere got rich by selling that exact same service?

Re:Is the driver open-source?

[Dan Ost]

Development in the open source world happens fast and it doesn't take long for a driver that isn't continuously updated to fall into obsolescence.

Therefore, I submit to you the reason that I prefer the following pragmatic argument that open source drivers are better than closed source drivers: closed source drivers lag behind kernel development.

Re:Is the driver open-source?

[Schraegstrichpunkt]

Is there a particular reason you need the driver to be open-source?

Uh, because the driver's authors can't anticipate all the ways in which people might want to use it?

Nvidia's binary blob driver doesn't work with Xen, so I can't use Xen on my desktop, and it'll only ever get fixed if Nvidia gets around to it.

Re:Is the driver open-source?

[the Hewster]

In that case, you can join the 170 people (so far) that have signed my pledge to "support graphics card manufacturers with open source drivers". At the end of the year I hope to have 1000 signatures and I will send a letter to ATI/AMD and nVidia representatives with the pledge and the signatures.

Re:Is the driver open-source?

[Anonymous Coward]

as for your comment about giving away what you do for a living... AMD doesn't write drivers for money, they make hardware. Intel manages to make hardware and open-source a good majority of their drivers, so that's just a stupid argument.

Im sure AMD is happy that you have cleared up what they can and cant do for a living. As one of the people that wrote that driver for ATI I can assure you they paid all their software engineers, and that the driver is less trivial than you think. And since everyone gripes about ATI's software compared to nVidea's software and vice versa, I think that suggests pretty strongly that software IS a major factor in a buyer's decision and it follows that AMD is certainly in the software business.

Personally I hope you stick to your principles and use nothing but IBM's shitty display adapters. How's the rest of your crusade going to get your cell phone software open sourced, or disk drive firmware, or your car's engine computer, or the other thousand devices you use everyday? Because, like, all those companies are in the hardware business and owe you their source code too.

Re:Is the driver open-source?

[dubbreak]

As a counterpoint, I recently bought a laptop with an ATI vid chip. No, I didn't have 3D accel working immediately, but all I had to do was click on "System"->"Admin"->"Manage Restricted drivers" (I'm using Ubuntu). I restarted my x server and I was good to go.

Compiz runs great (faster than my GMA850 based laptop I had previously) and no issues. This was waaay easier than my previous experience with installing drivers for both Nvidia and ATI, of course the majority of the thanks has to go to Ubuntu in this case. Stuff is looking up for drivers under linux. I'm crossing my fingers that the whole Dell/ubuntu thing will help accelerate the progress.

Re:Is the driver open-source?

[Anonymous Coward]

AMD has acceptable drivers for Windows, shitty drivers for linux and no drivers for Solaris.
It seems that install base is an important factor, don't you agree?

Re:Is the driver open-source?

[Schraegstrichpunkt]

Or you could put your money where your mouth is and support the Open Graphics Project. If this gets off the ground, the other manufacturers might wake up.

Re:Is the driver open-source?

[Vanders]

Is there a particular reason you need the driver to be open-source?

Yes

Unless you are a driver developer yourself..

I am.

Re:Is the driver open-source?

[BrainInAJar]

How does this detract from my statement that I need open-source drivers because I use an esoteric operating system ( I happen to think it's the best OS out there, so I use it. This means to me that I'm limited to intel & nV cards, and if I were a fooBSD person or something, I might be limited only to intel )

Re:Is the driver open-source?

[mashade]

Absolutely! Nex is great fun, and the community is growing.

Re:Is the driver open-source?

[just_forget_it]

I'm guessing you think the only application for 3D acceleration in Linux is games.

Re:Is the driver open-source?

[tyrione]

Perhaps the Linux ABI developers should coordinate with 3rd Party OEMs more closely to keep them abreast of any dev changes that would break their driver processes. It's a two-way street.

Re:Is the driver open-source?

[jack455]

ATI and AMD can do whatever they want. No one owes Linux users an opensource driver, but Intel increased our loyalty by giving them to us. Even though they're not the best hardware. nVidia has excellent hardware and while the drivers aren't opensource, at least their binaries work fairly well.

I used to really like ATI when I used Windows. I might even recommend them to a Windows user.

We may not be the largest demographic, but there is no doubt that they could sell Linux users more hardware if they got out of the "software business"

If they're even in it...don't they provide the drivers at no cost only to people who have their hardware? Sounds like they're really diversifying.

Re:Is the driver open-source?

[arodland]

Another interesting thing about the Intel driver is that due to being pretty much the most capable open-source driver around, it gets a lot of attention from XOrg way, including being compatible with the latest nifty standards. If I want TV-out on ATI, I have to use a driver that's been in continuous beta for the past four years and reboot the machine with the TV plugged in so the card notices it. If I want TV-out on NVidia I have to put weird crap in my X config file and then run nvidia's custom settings app to configure displays. Okay, better. If I want TV-out on Intel I use xrandr -- from the commandline or from any of the GUI utilities already out there... and it works. Bang. Just like that.

Re:Is the driver open-source?

[eldepeche]

ATI and nVidia are in the software business because software is required to make their hardware work: the hardware only does what the software allows. Since they don't charge money for their Windows drivers, I find it hard to believe that money is any obstacle to making their video drivers free software.

Also, for many applications (3D desktop effects, most games) IBM's shitty display adapters work pretty well, and I'm perfectly happy to use them because they are fully supported on Linux and some other free platforms.

Re:Is the driver open-source?

[Provocateur]

Any luck with Beryl.Compiz, or AIGLX? Or you didn't bother? I was thinking of upgrading soon...

What distro and which driver are you using? Just curious...

Re:Is the driver open-source?

[msh104]

I am actually running nvidia under xen right NOW.

http://en.opensuse.org/Use_Nvidia_driver_with_Xen

a big shame its not the latest driver... but it works :D
no idea if it is easy to hack this fix into later versions of the nvidia driver though..

but i agree that official nvidia support would be a lot better...

Re:Is the driver open-source?

[Anonymous Coward]

Tremulous is where it's at!

Re:Is the driver open-source?

[DrSkwid]

I call BS, someone who writes video drivers professionally should really be able to spell nVidia.

Re:Is the driver open-source?

[wilder_card]

>> Yea, because you would be able to find the rootkit in a driver with thousands of lines of code...

Well, yeah, actually I could. You mean you can't?

Re:Is the driver open-source?

[AJWM]

Any luck with Beryl.Compiz, or AIGLX?

Only to the extent that I tried a live CD with Beryl.Compiz on it and played with the eye candy for a little while. It worked fine, but I'm not a big enough fan of eye candy to worry about it too much. I'm not even sure what driver it loaded now.

My default distro is SUSE 10.1 (from before the Novell-MSFT deal, I've been using SUSE for years), with the radeon driver. AMD64 in 64-bit mode.

Re:Is the driver open-source?

[maglor_83]

As a counterpoint, I recently bought a laptop with an ATI vid chip. No, I didn't have 3D accel working immediately, but all I had to do was click on "System"->"Admin"->"Manage Restricted drivers" (I'm using Ubuntu). I restarted my x server and I was good to go.

Yeah thats what I had to do too. Then I couldn't play videos, and I couldn't sleep my laptop. So now whenever I want to do 3D, I get to go through this process, and then undo it when I'm done.

Re:Is the driver open-source?

[TwilightXaos]

It would help if the driver were open-source, so that the ABI developers know what changes are likely to affect the driver without guessing.

Re:Is the driver open-source?

[Vexorian]

Is there a particular reason you need the driver to be open-source

For example:

• If AMD stops supporting it, someone else will be able to continue updating it
• If another platform wants it X person will be able to easily port the drivers
• X driver developer may find a way to improve it
• If X driver developer improves it, I will be able to use the improvements
• In total , I, the customer will be able to use my hardware for longer and in more platforms and the quality would even improve.
• And I paid for the hardware anyways, so AMD guys can still make a living from that money...

Re:Is the driver open-source?

[Rich0]

Well, it might except that the ABI changes are probably designed largely to make life as difficult as possible for anybody who distributes binary drivers. Helping them to cope with the changes would sort-of defeat the purpose in making the changes in the first place.

The kernel developers aren't exactly supportive of non-GPL binary modules...

Re:Is the driver open-source?

[Enderandrew]

Sabayon supports Compiz-Fusion out of the box on both ATI and NVidia cards. You get codecs, proprietary drivers, and everything. It just works.

The Live DVD is horribly slow, but the installed system is pretty nice. Just make sure to uninstall Beagle, which is a performance hog, and then you get a great desktop which includes Skype, Google Earth, Picasa, tons of 3D Games, Second Life, etc.

www.sabayonlinux.org

Re:Is the driver open-source?

[Anonymous Coward]

HA. I also bought a new laptop with an intel card(laptop's an FujitsuSiemens Amilo PI1505, graphics intel gma950) and it's been a complete nightmare. The "intel" driver is extremely unstable for 3D, can't run googleearth for more than a few minutes, compiz, or anything that uses 3d. The latest xorg/intel driver in debian crash with Fatal lockup or Ring errors and it's been like this for quite some time from what I gather from the xorg bug list.
I remember Linus (and people here) saying good things about the open sourceness of the intel graphics cards and damn was I wrong to trust this.
On the other side, nVidia's blob works great with no problems across all the machines I have, at least for me.

Re:Is the driver open-source?

[pakar]

Well.. It's the chicken and the egg story but for computers..... Without good drivers there cant be any good games..

And it's not just about 3d... good 2d performance is also a big issue, and without open specifications it's very hard to write good drivers..

Re:Is the driver open-source?

[vipulgoel]

http://slashdot.org/~vipulgoel/journal/182135

Put up or shut up...

[MMC Monster]

Really, it's not that I like nvidia. But I've been hearing reports on /. since the beginning of the year of ATI linux drivers coming soon. How about we wait until they're actually release before bothering to give them any support.

Re:Put up or shut up...

[thegnu]

But I've been hearing reports on /. since the beginning of the year of ATI linux drivers coming soon. How about we wait until they're actually release before bothering to give them any support.

1. You're new here, aren't you?
2. ATI is under new management, or haven't you heard?

Plus, isn't AMD generally considered pretty good for Linux support? Not that it's not mostly to the NFORCE chipsets' credit, but AMD purchased ATI because their Nforce "monopoly" was over, so Linux users were no longer locked into broken Intel chipsets if they purchased an Intel processor. Am I right?

Am I the only person who suspected that AMD would, in due time, put Linux driver support as one of its main priorities? I seemed to be one of the few (excuse me while I toot my own horn) that Snape killed Dumbledore because Dumbledore wanted him to, so maybe.

Oh, and [/spoiler], btw

Re:Put up or shut up...

[Prof.Phreak]

Amen. Same for AMD in general. Come on... actually release -SOMETHING-!

Re:Put up or shut up...

[Brandybuck]

Ditto. I am a FreeBSD users. There are no drivers for newer ATI cards on FreeBSD, because ATI refuses to open source their drivers. I have been hearing rumours of open source drivers for a year now, but nothing. At LWCE the AMD guys said we would see some imminently, but nothing.

I am building a new system in the next month or two, and if ATI still hasn't come out with open source drivers, I'm going with Intel instead.

Are they open?

[MarcQuadra]

They're useless to me unless the source is available, preferably under the GPL. I really wish they'd work -inside- the framework of the kernel, Mesa, and xorg projects instead of building one-off binary drivers. What if I want to use their card on PowerPC, want to link against the latest (or a non-mainline) kernel, or just want to run an all-open system?

Re:Are they open?

[gmack]

Right now I would settle for a driver that works on recent kernels since one of those improvements mean much to me if I can't actually install them.

I used to be a huge ATI fan but I've completely stopped buying their stuff. If they can't be bothered to make working drivers or have useful support answers. I can't be bothered to shell out money for something that's just going into the garbage bin anyways.

NVIDIA is marginally better.. at least these stuff works even if I have to reinstall the X.org drivers every time I update a kernel.

Re:Are they open?

[Bert64]

Supporting PPC makes even less sense now that AMD owns ATI...
Why would they want to support their cards on a processor type they don't produce?

Re:Are they open?

[corvair2k1]

NVIDIA has long since had a handle on your "latest/nonstandard kernel" problem. It builds its own interfaces to conform to the kernel's.

Re:Are they open?

[david.given]

What if I want to use their card on PowerPC, want to link against the latest (or a non-mainline) kernel, or just want to run an all-open system?

On a related note: does anyone know if it's possible to get standalone graphics cards with Intel 3D graphics hardware on them?

I know that on an absolute scale, the Intel chipsets aren't particularly fast... but they're certainly faster than the Radeon 9600 mobility I've got right now, and there are genuinely open source accelerated drivers for them. Which means they ought to be much less of a hassle to use. For 2D and lightweight 3D use, they should be ideal.

But I've only ever seen them in integrated chipsets, and I'd rather not buy a whole new motherboard just to get a new graphics card...

Re:Are they open?

[Constantine XVI]

AFAIK, Intel graphics chips only come on motherboards with Intel chipsets, which only handle Intel processors.

See a pattern here?

Re:Are they open?

[TheRaven64]

They're useless to me unless the source is available, preferably under the GPL. Out of interest, why GPL? The rest of DRI (and x.org) is MIT licensed, including the Intel drivers. The only parts that are GPL'd are the kernel modules (which do a small amount of validation and pass instructions to the hardware). Keeping the majority of the drivers MIT licensed makes it much easier for people to add support for other operating systems, such as FreeBSD and Solaris (both of which are supported by nVidias blobs, although only FreeBSD has good support for Intel chips since no one has ported DRI to Solaris yet).

Re:Are they open?

[KDR_11k]

To get at least some sales from people who don't buy one of their products?

Re:Are they open?

[Bert64]

But seriously, how many PPC workstations get sold nowadays?
Especially ones with slots able to take new videocards...
It's such a small niche that it's probably not worth it for AMD to pursue.

Re:Are they open?

[Anonymous Coward]

> What if I want to use their card on PowerPC
Then you will also need a PowerPC version of the BIOS to initialize the adapter. It worked with a Radeon7000, where you could download a PPC version of the BIOS for use in a Macintosh, but that's history now.

Re:Are they open?

[fimbulvetr]

Are you aware of the concepts of capitalism? I don't mean to be condescending, but serious. Do you understand there are significant costs on the programming side for an entirely different architecture? Do you understand AMD needs to make money to survive as a company? Do you understand that only a fraction of their customers are running Linux, and of that a trivial fraction are running PPC?

Re:Are they open?

[Mattintosh]

Perhaps instead of "PPC" the OP should've said ${nonX86Processor} so you would get the drift. The same applies to Sparc, ARM, Power-and-siblings, and a half-dozen others that I can't be bothered to name. Not all of these are out of mainstream production (even PPC is still in mainstream production, just not by Apple) and there is certainly a need for Linux video drivers for these platforms. Not to mention embedded stuff...

Re:Are they open?

[bfields]

Yah. Not to put words in the original poster's mouth, but I'm guessing what they meant was "under a license compatible with the upstream projects", and the upstream project they came immediately to mind was the GPL-licensed kernel.

The important thing is not to end up in a situation like openafs, say, where the code's under an open source license that has some technical conflict with the GPL and as a result has to be maintained outside the tree for ever....

Re:Are they open?

[mrchaotica]

The same applies to Sparc, ARM, Power-and-siblings, and a half-dozen others that I can't be bothered to name. Not all of these are out of mainstream production...

I hate to break it to you, but you're using a very non-standard definition of "mainstream."

Re:Are they open?

[Anonymous Coward]

I really wish they'd work -inside- the framework of the kernel

My sentiments exactly. I don't want open source drivers for idealistic reasons -- I want open source drivers because they follow the rules, unlike closed-source drivers which make up their own arbitrary rules.

I'm tired of hearing about how "closed and open can work together", or "closed is better than nothing". Sure, that's true, but it doesn't change the fact that I must attend to a special, arbitrary solution for a freaking video driver, unlike every other driver on my system. I consider it a wart, something that doesn't belong.

When it comes to computers, and operating systems in particular, conformity is a good thing.

Re:Are they open?

[Cajun Hell]

See a pattern here?

Yep, a strong pattern indeed. Intel is saying, "Buy an Intel processor," whereas AMD is saying, "Buy an Intel processor." Anyone who can't spot the pattern in that, has to be pretty dense. I wonder why AMD's stockholders haven't noticed it.

Re:Are they open?

[howlingmadhowie]

if they documented the interface, they wouldn't have to support it. someone else would do the work.

Re:Are they open?

[Anonymous Coward]

What if I ... just want to run an all-open system?

Yeah, what if you just want a pony too? They're not going to give you one are they, the bastards!

Re:Are they open?

[The Warlock]

The only people who see it like that are the same people who would rather have an Intel Integrated Whatever than a Radeon 9600.

Re:Are they open?

[nuzak]

Actually, ATI does work with the Linux infrastructure, using GLX and DRM (Direct Rendering Manager, not that other DRM). It's nVidia that goes off and does their own thing. Of course, nVidia just seems to do it better, quite possibly because they control more of the API, but more likely it's all in the driver.

I've always liked ATI's hardware (the X1950 AGP was a real gift to us folks with old boxes) but nVidia's software (objectively speaking, I hate nVidia's software less). At least ATI's build system isn't as bad now, though I still haven't seen a "one-click" script wrapper for it like there is with nVidia.

Re:Are they open?

[Anonymous Coward]

See a pattern here?

Yes, that Intel uses it's in-house components to support their own in-house components and build hardware.

Re:Are they open?

[edmicman]

GM should make me a 3-wheeled car, too, because well, I want one. Why won't they do this? I'll buy it, that'd be one extra sale for them. Why aren't they jumping all over it?

Re:Are they open?

[Anonymous Coward]

http://www.3wheelers.com/gmlean.html Go get one of those :)

Re:Are they open?

[MarcQuadra]

I used to buy ATI for all my Linux machines, but the past few years the R300 rift opened wider and wider, and the open-source support for the new cards got more and more flaky (not the drivers themselves, they just keep falling farther and farther behind the hardware offerings). I've switched to Intel integrated graphics for my machines, the GMA 3000 and X3000 are actually enough horsepower for all my needs, Intel is commited to providing open code, and the picture quality seems just as good as the two 'titans' offer.

Re:Are they open?

[msormune]

Then they are useless to you and they always will be. The reason for closed drivers is intellectual property, to put it very shortly. AMD does not own all of the hardware and logic in these graphics chips, instead they license it. They are not allowed to just hand it over to OS people, no matter now much they huff and puff. For example, S3 owns several patents for texture compression although I'm not sure they even make graphics chips anymore. Do you really think they will give that away for free? Why would they?

Re:Are they open?

[eviltypeguy]

Incorrect.

Solaris now has DRI support for chips such as the Intel 915 since October of last year at least:
http://mail.opensolaris.org/pipermail/xwin-discuss /2006-October/000356.html

Re:Are they open?

[Lost Race]

GPL would prevent competitors from using the AMD-developed driver technology without sharing back their own improvements. Seems like a win for both AMD and the Linux community (developers, distributors, and users) though a bit of a loss for everyone else. I'm sure Nvidia and Intel would rather see AMD's software released under a more permissive license.

Re:Are they open?

[init100]

Then maybe you are interested in this?

Yeah, I know, sounds too good to be true. But last time I checked, April 1st had already passed.

Re:Are they open?

[spitzak]

"GPL it" is really shorthand for "release the source in some way". I'm sure the people asking for it would be equally happy with MIT licensed or BSD or public domain. Or even some strange license from ATI as long as it allows the source to be modified and the result put onto a Linux distribution disk.

The advantage of the GPL for ATI is that others could not take the code and put it into closed source things (such as the NVidia driver...)

Re:Are they open?

[777a]

For the last several years I've alternated between nVidia and ATI (nVidia 4200Ti, ATI 9800XT, nVidia 6500, ATI X1900XTX).

ATI's windows drivers are terribly bloated (NET framework catastrophes), the only reason I still bought ATI was because of the excellent, efficient 3rd party Omega drivers (Win32 only).

My last card (the ATI X1900XTX) cost ~$600, since then I've started paying a lot more attention to Linux, and unless ATI do some decent drivers it'll be the last ATI card I buy.

Re:Are they open?

[Jaysyn]

Crap, I forgot about those. I wonder if they have a compatible driver for the Radeon Express 1100 in my laptop. Acer can't be bothered to update the drivers & ATI's drivers won't work with them.

Re:Are they open?

[Anonymous Coward]

They're useless to me unless the source is available, preferably under the GPL.

what I would like to see, is not ATI or AMD or ANOther company releasing a driver under the GPL, but actually releasing the specification for the chips. If you've ever looked at a badly written, uncommented driver, the source isn't that much help.

Re:Are they open?

[j0ebaker]

Absolutely Right on -- If they are not Open Source I'll just suffer a little less and use Intel's Open Sourced chipset with 3D. It seems to work good enough on my Ubuntu Gutsy Gibbon Tribe 5 test system. Now this is going to push 3D card vendors to open up their code or open up their hardware documentation for open code to be written.

Re:Are they open?

[ConceptJunkie]

Why would they want to support their cards on a processor type they don't produce?

Because they aren't evil, vindictive bastards like, say, Microsoft?

Re:Are they open?

[idontgno]

If you're gonna use a car analogy, at least use one that fits. (Although I acknowledge that this isn't traditionally required on /.)

GM (after it buys out Goodyear) should make tires that can be used on 3-wheel cars, as well as 4-wheel ones. That way they can be competitive in the market for trike tires, especially since many trikes use perfectly conventional car wheels.

Re:Are they open?

[Ash Vince]

GPL would prevent competitors from using the AMD-developed driver technology without sharing back their own improvements. Not necessarily. It is perfectly possible to read someone else's code then use it to inspire your own development without actually copying it line for line. You can use their code as a roadmap to a solution without openly copying it line for line.

Also note that reading the code may give a valuable insight into the hardware it is meant to support and how to either base your future products on that hardware, or possibly design your future products such that code optimised for you competitors products will run well on your hardware but code optimised for your products will only run well on your products. Since both Nvidia and ATI have "friendly" games developers who already favour one company or the other this could give a substantial advantage.

Remember a few years ago when many games had the runs best on nvidia logo as soon as you launched them. I also seem to remember some strange shenanigans with regard to 3Dmark optimising their benchmarks to favour one card at around the same time although I cant remember which. I am also fairly sure that ATI are not above befriending a few games companies with free samples in order to make sure their games run slightly better on ATI hardware.

The fact is that the 3D card market is driven by people buying the latest tech to get the highest framerate in the latest hot title. Making sure that killer app ran slightly better on your product could translate into alot of extra sales for that company.

There is also the possibility that both teams of hardware designer have been reverse engineering competitors products and doing all these things for years, but open sourcing the drivers (and hence exact hardware specs) would reveal this to their customers. I know in the past John Carmack took issue with Nvidia for releasing a less advanced card with a higher model number but then tweaking the driver so that the new card appeared to be more advanced.

The fact is that we live in a capitalist system which encourages these sorts of tricks if you can get away with them. Microsoft has been walking a fine line of anti-competitive practices for years but have always managed to get away with it. I know they have been fined and had various punishments leveled at them but they have never made a loss to my knowledge and they still maintain a position of dominance in the application market by leveraging the fact that they produce the defacto OS that most people use. I just think that other companies also engage in similar practices but people make less of a fuss since they are not the pariah the MS have become.

Re:Are they open?

[Chandon Seldon]

I hate to break it to you, but you're using a very non-standard definition of "mainstream."

You'd think that, but it gets a bit less clear when you consider things like the 3D workstation market and the semi-embedded market. How "mainstream" are Sparc workstations among the potential customers of FireGL cards? How about PPC workstations? How common are ARM processors? What are the chances that someone will want to combine a commodity ARM processor with an AMD graphics card?

That last bit is sort of important. AMD makes a business of selling commodity hardware components. Their customers are companies building their own products, who expect to be able to use AMD hardware with other commodity hardware.

Re:Are they open?

[maglor_83]

Thats why its a benefit to open source their drivers - then they can sell to these people without having to do this work.

Re:Are they open?

[fimbulvetr]

It doesn't say it in the summary, but AMD says they also intend to release (all/some?) specs for ATI cards. This would be a tremendous boon for the OSS community - and wholly hope for support on every arch if this is the case, but I think it'd be unreasonable to ask AMD to support this natively given the very specialized niche from which they likely never gain any profit.

Re:Are they open?

[einhverfr]

How many processor architectures does it support?

Re:Are they open?

[rtb61]

I understand only this, I am a customers, I want to use Linux, well, specifically Kubuntu, I will only select a graphics card that supports this. I do not buy products to make profits for other people, I buy products that fulfil my requirements, that is capitalism.

Of course there is the whole it is a lot more fun to play with the penguin that to get burnt by M$.

It will only be a matter of time until M$ starts contracting out their own range of 100% Vista compatible video cards produced under contract by an ODM and of course every other video card will only achieve partial compatibility unless they pay the Vista (or what ever the hell the next M$ (P)OS will be called) licence fee per unit installed.

Re:Are they open?

[MarcQuadra]

Intel has similar problems, but they manage to work around it by releasing as much 'open' code as possible, then leaving the proprietary bits in an optional binary shared object. Intel hasn't released the *.so file yet, but that seems like the most reasonable solution, not to mention that an endeavoring outsider might be able to reverse-engineer or even completely re-implement those 'secret' bits.

The beauty is that there's a working 2D and 3D graphics driver that everyone can port and improve upon, there's 'extra' functionality to those who need it and can be encumbered by the restrictive license, and the interfaces and methods of access for the proprietary parts are clearly labeled in the open code, leaving room for members of the community to implement them on their own.

Re:Are they open?

[msormune]

Actually, S3 texture compression is also available as an third party add-on to the Mesa libraries, or was the last time I checked. It worked also with ATI cards. The problem with the otherwise good approach you described is the lack of hardware documentation. And reverse-engineering is very difficult with the increasing amount of functionality gfx chips hold. And the implementation of that functionality might still be actually illegal without a proper license from the IP holders... But maybe Mesa and 3d driver libraries could be more modular in a linux kernel sort-of style so that the approach you described would be easier to take.

Re:Are they open?

[Bert64]

PPC and Sparc workstations are not commodity devices... If you buy such a device from IBM/Sun, then the respective vendor will have made sure it comes with a compatible video card.
ARM is really only used for low-power situations, where you'd also want to couple it with a low power video chipset (if any at all). Although highend videocards offload a lot of the processing, they don't offload it all so you still need a reasonably performant CPU to go along with it.

Re:Are they open?

[mSparks43]

I use Fedora 7 and an ATI Radeon X1600
Last months drivers finally worked (the previous ones would crash X when you tried to restart Xorg)
If I get AIGLX support.

YEY! thank you,thank you,thank you,thank you.

As for the 'it should be open source' calls, I'll agree with the earlier comment about patents, but also, I really don't think it should be an issue, and I'm really quite unnerved by all the 'IT HAS TO BE GPL' people, using GPL is an individual choice for a software developer, just like using BSD is a choice, or keeping everything out of easy view is a choice. Why the H377 should you have any call on how someone else chooses to release software to run their hardware. That's like me banging on your door and demanding you let me drive your car, just because your neighbour gave me his keys.

Re:Are they open?

[gmack]

As for the 'it should be open source' calls, I'll agree with the earlier comment about patents, but also, I really don't think it should be an issue, and I'm really quite unnerved by all the 'IT HAS TO BE GPL' people, using GPL is an individual choice for a software developer, just like using BSD is a choice, or keeping everything out of easy view is a choice. Why the H377 should you have any call on how someone else chooses to release software to run their hardware. That's like me banging on your door and demanding you let me drive your car, just because your neighbour gave me his keys.

Well the main logic behind demanding GPL is that we want something that can be integrated with the kernel. Having to reinstall the drivers separately each time I upgrade the kernel is a huge pain and a major reason I'm considering going Intel on my next purchase.. it's just too annoying to bother.

Re:Are they open?

[mrchaotica]

You'd think that, but it gets a bit less clear when you consider things like the 3D workstation market and the semi-embedded market.

Like I said, you're using a very non-standard definition of "mainstream." Mainstream, for computers, means basically "stuff you can buy at places like Best Buy or Fry's." Now do you get it?

Re:Are they open?

[Chandon Seldon]

Like I said, you're using a very non-standard definition of "mainstream." Mainstream, for computers, means basically "stuff you can buy at places like Best Buy or Fry's." Now do you get it?

That's more "mainstream from your perspective as user of home desktop computers". I'm talking about "mainstream in the computer industry", where home desktops are just one niche (a small one compared to embedded ARM systems, and a low-margin one compared to graphics workstations).

Re:Are they open?

[Chandon Seldon]

PPC and Sparc workstations are not commodity devices... If you buy such a device from IBM/Sun, then the respective vendor will have made sure it comes with a compatible video card.

False. PPC and Sparc are both reasonably open architectures sold by numerous companies. Any company who wants to can buy Sparc or PPC in a compeditive market (much more compeditive than the market for Intel chips) and build whatever they want; some even make workstations.

I guess you're right. The workstations themselves aren't common enough to be considered commodity devices - but the chips are. And AMD doesn't sell graphics chips to end users, they sell them to device makers. If they want to sell them as commodity graphics chips then they will be expected to work with commodity processors - like PPC, Sparc, and (to some extent) ARM.

Hey AMD, A tip for you.

[Lumpy]

I purchase Nvidia only because the cards actually work under linux, or they used to. Lately there are issues...

If AMD steps up to the plate and gives us good drivers and actually listens and reacts fast to reported problems, they can come out way ahead.

Nvidia driver install used to be painless, now it can be incredibly painful depending on the Distro and Card you have. I still cant get a old Geforce4 card working on my wifes ubuntu PC. I gave up and switched to the intel onboard chipset. Far better support for that video chipset than nvidia is giving us even for the older cards that USED to work great.

Re:Hey AMD, A tip for you.

[ajs318]

What's wrong with the i-tal "nv" driver? Never installed nVidious's closed-source crap, never likely to.

Re:Hey AMD, A tip for you.

[colinleroy]

I purchase Nvidia only because the cards actually work under linux, or they used to. Lately there are issues...
That's why I purchase Intel only -- free drivers work and are actually supported by the distros, the kernel people, and Intel.

Re:Hey AMD, A tip for you.

[Lumpy]

Um no 3d, no Xv acceleration, makes it useless to even have a video card outside the built in cheapie no 3d no acceleration anything card.

nv driver is good for install or limp mode only.

Re:Hey AMD, A tip for you.

[corvair2k1]

This is not likely NVIDIA's fault. Ubuntu has various issues dealing with 3rd party drivers that conflict functionally with those in the repository. In this case, the various "nv" drivers conflict with the blobs released by NVIDIA. One way to fix this is to use the nvidia-glx packages from the repositories. The other way, if you have to use NVIDIA's blob, is to blacklist the included nv drivers.

Re:Hey AMD, A tip for you.

[Zonk (troll)]

What's wrong with the i-tal "nv" driver? Never installed nVidious's closed-source crap, never likely to. It's 2d only. Ie, no Beryl/Compiz. It also doesn't seem to be very stable. I use that driver on my PowerBook (it runs Ubuntu), but X freezes half of the time I try to play a video*. If I'm at home I have to ssh in from another machine and kill X so I can use the thing again, or if I'm anywhere else I have to reboot (keyboard and mouse are frozen as is everything on the desktop, ctrl-alt-backspace doesn't do anything).

* It doesn't matter whether I'm using VLC, Xine, Mplayer, or Totem. I happens very often, which is why I'll usually just boot it into OS X if I want to play a dvd or avi.

Also, in reply to Lumpy (gp), why is it so hard to go to "System->Administration->Restricted Drivers Manager"? I've done that with a few GeForce4 (integrated) cards and it's as easy as typing your password and clicking a button.

If you're not running 7.04, then just do "System->Administration->Synaptic Package Manager" and do a search for "nvidia-glx". Install that and it should work (you might have to change /etc/X11/xorg.conf to "nvidia" rather than "nv", I don't remember if that's automatic or not).

Ubuntu is by far the easiest distro to install 3d graphics drivers on since they provide the packages. No compiling and it will always work across reboots since the driver gets updated when the kernel does.

Re:Hey AMD, A tip for you.

[non]

at the risk of being offtopic:

with Ubuntu you have a choice of three different NVidia drivers; new, normal, and legacy. you should probably use normal 'nvidia-kernel-' & 'nvidia-glx-'. if, on the other hand, you have a brand spanking new card, you will need the beta drivers direct from nvidia and you will have to install them yourself. in the event you choose to go that road do *not* install the linux-restricted-stuff - it will interfere with the drivers, and remember to re-install the drivers from recovery mode every time you upgrade the kernel more than a minor version (Ubuntu point releases, ie. 15-27 ->16-31 won't cause a problem).

Re:Hey AMD, A tip for you.

[everphilski]

Must be a Ubuntu thing. I've never had a problem with Red Hat, CentOS (I know, virtually the same thing), Fedora or Slackware. I've had 4 different cards in 4 different computers over the years (as a gamer and then as a graphics/visualization person at work, programming under Linux). NVidia's drivers always seemed to work on the first try.

Re:Hey AMD, A tip for you.

[MrNemesis]

As I think someone else has said, nVidia maintain a "legacy" driver for older GPU's. Unfortunately this keep changing - damnit, I wish someone would come up with an app that can install the correct driver based on an lspci output... (unless the nVidia installer already does this, but IIRC it doesn't).

I also had problems with my venerable Ti4200 until I installed the legacy driver, after which it worked fine.

You are better off using the Intel onboard though - solid for 2D stuff and good for light OGL stuff too. The GMA 3000 series are especially capable little GPU's.

Re:Hey AMD, A tip for you.

[(H)elix1]

Have you tried a current Intel graphics card lately? I've been struggling with a Intel 945GM graphics card in a laptop. Yes, drivers in source code form exist at http://www.intellinuxgraphics.com/ but I'll be damned if I can get everything to work with the directions provided using Centos/RHEL. Using the 915Resolution hack got it running at 1280x800, but for all the googling and failed attempts, 'supported' is a very strong word. Yes, I'm not a guru when it comes to the Linux configuration stuff. An RPM, yum, apt-get, emerge or deb for the 'major' distros is what I'd call supported. Source code is good, but I had to figure out what GIT was and how to install it before I could even start with trying to get native resolution. Way more work than it has to be compared to some of the other graphics cards/chipsets.

Re:Hey AMD, A tip for you.

[chill]

How does it work for gaming? I'd really like to support Intel on all my machines, just because of their support for the GPL drivers they have done. However, I've yet to see adequate 3D gaming performance on their 945 and 950 chipsets.

If you have one, can you give me an idea how Linux native 3D games play? Hardware specs and framerates would be good. America's Army v2.5 is popular, freely available as a download, has a native Linux version and is a total dog on the old ATI drivers. With nVidia I get faster performance on Linux than the same PC on Windows.

I'm going to be purchasing two new laptops for my kids and am torn between nVidia and Intel graphics chipsets. They're doing 3D graphics and need good 3D support -- FOSS or not.

Thanks in advance.

Re:Hey AMD, A tip for you.

[hax0r_this]

For gaming I think you will find the Intel chips to be lacking (although they might play Americas Army at low resolution ok, I'm not really sure). Intels chips are called "Graphics Media Accelerator" for a reason - they're designed for people who want to work on spreadsheets and maybe watch the occasional movie. If you want something good for gaming try to get an nvidia 7600 or higher. Last I looked some of the 17 inch Dells can be configured with a 7900GS, which would make for a pretty sweet gaming laptop, (but would be heavy, hot and low battery life probably). My new Macbook Pro has an 8600M GS which plays Americas Army 2.8 pretty well in Windows at 1440x950 (but doesn't run Linux worth crap yet - stay away if you want Linux). But yeah, if you want to play anything more graphics intensive than maybe BZ Flag I would recommend nvidia.

Re:Hey AMD, A tip for you.

[richlv]

yeah. and nvidia is fixing bugs only in the new drivers. woohoo.
i am buying and recommending nvidia only still, but if ati would provide opensource drivers in kernel/xorg... that could change. not likely, given that we've heard all kinds of such vague promises before and nothing changed.

Re:Hey AMD, A tip for you.

[Anonymous Coward]

I don't use ubuntu (I use Debian), but I know Nvidia changed their legacy driver package so there's two of them now. The old "nvidia-kernel-legacy" package in Debian was only packaging the 71XX series driver, which doesn't support the Geforce4. I submitted a bug report and the Debian Nvidia maintainer made a package out of the 96XX driver that contains the correct driver.

Re:Hey AMD, A tip for you.

[MrNemesis]

To be fair, they do update their legacy releases to some degree. I think the major stumbling block now is that, with different legacy "versions" knocking about, many users don't want to have to grep through a "here's the supported cards for this driver" list to find out which driver they need to install. Be nice if $distro_package_manager could be made to figure this out for themselves...

ATI have really, really dragged their heels on this. There was a time when I'd have jumped to ATI in a flash, but support for their Linux drivers has been so terrible I have serious doubts of them being able to maintain a decent release. ATM I'm recommending Intel to everyone that doesn't need gaming performance.

Still no word on when accelerated H.264 playback will be supported on either nVidia or Intel hardware, both of which are capable of it...

Re:Hey AMD, A tip for you.

[Zebedeu]

I still cant get a old Geforce4 card working on my wifes ubuntu PC That's not only a problem with the Linux drivers, as I had exactly the same happening to me in my mothers Windows XP computer. It appears that nVidia stops testing their drivers with old iterations of their video cards, though it would be at least helpfull if they acknowlodged the problem and made available on their website old versions that are know to be working.

In the end I made it work by searching for old versions of the nVidia drivers on the internet. Perhaps if you try an old version of the linux drivers you will be luckier.

Re:Hey AMD, A tip for you.

[pato101]

If I'm at home I have to ssh in from another machine and kill X so I can use the thing again, or if I'm anywhere else I have to reboot (keyboard and mouse are frozen as is everything on the desktop, ctrl-alt-backspace doesn't do anything). Next time try Alt+Print_Screen+K (search for Raising Skinny Elephants is Utterly Boring at the wikipedia)

Re:Hey AMD, A tip for you.

[AaronW]

I concur. We have some newer PCs at work with integrated Intel graphics and the drivers don't work at all. The best I can get is frame buffer mode, which is useless with the LCD monitors, since the FB drivers cannot match the monitor's native resolution and are incredibly slow. I just gave up and installed an nVidia card with nVidia's driver, and everything just plain worked. I've rarely had problems with their proprietary driver. The only thing I need to remember is to reinstall after I boot into a new kernel (but even this is fairly painless).

3-D works great on the nVidia driver as well.

Re:Hey AMD, A tip for you.

[CCW]

That's my position exactly - but I'm looking at the chipsets, not the video cards. I don't 3d game, so embedded video is perfectly adequate for me as long as I can get solid drivers for it. I'm in the market for a new quad core machine in the next couple months - I run AMD machines nearly exclusively now, but Intel is my top choice for the next system primarily based on the solid open source support for their on board video chipsets.

Re:Hey AMD, A tip for you.

[digitalaudiorock]

What's wrong with the i-tal "nv" driver? Never installed nVidious's closed-source crap, never likely to. It's 2d only. Ie, no Beryl/Compiz. It also doesn't seem to be very stable. ...and even for a lot of 2d use, like MythTV, it's useless due to lack of TV out support.

Having said that, as an owner of an RPCRT TV and a MythTV system, it kills me that the proper output of 1080i content at 1080i in the nVidia drivers has been hopelessly broken apparently since the 8xxx linux drivers...so I'm not about to get too evangelical about them. If the nv drivers, or for that matter the new ATI drivers, could handle that correctly, I'd use them in an instant.

Re:Hey AMD, A tip for you.

[celtic_hackr]

What's wrong with the i-tal "nv" driver? Never installed nVidious's closed-source crap, never likely to.

... I use that driver on my PowerBook (it runs Ubuntu), but X freezes half of the time I try to play a video*.

Try apt-get update && apt-get dist-upgrade. The latest nv driver seems to be a bit more stable, or maybe it's the latest X.Org.

Re:Hey AMD, A tip for you.

[Provocateur]

in reply to Lumpy (gp)

Glad you clarified what Lumpy was, or who, otherwise, I was thinking, was there a mutant Ubuntu released just recently that I wasn't aware of?

Seriously, though, go AMD! Don't forget us ATI Radeon 7000-series users! I'm sure your development team of 1 won't be too distracted by updating those drivers as well. (But if you make them open, that team of 1 can easily become 100 or 1 dozen...)

Re:Hey AMD, A tip for you.

[RobertLTux]

yes I have in my box a radeon 7500 aiw and the open driver works well but

1 no Close captioning
2 no stream capture (this didn't work well at all but..)
3 the 3d performance sucks

and what IP is actually left in the driver (heck mail the source to the Windows driver and a Spec to the X.org team)

Re:Hey AMD, A tip for you.

[Anonymous Coward]

<OffTopicHateRant>

I agree. Too often the words "supported," "working," and "easy" are abused by the Linux community.

"Supported" is often used to mean "working" rather than "you can get authoritative technical assistance."

"Working" often means that the fundamental minimum functionality of the device or software works, but forget about any thing beyond that. Example: A Sound card can be made to produce sound, but forget about surround sound. Check in the box... it "works" and gets placed on the HCL.

"Easy" often means that it can be accomplished with only one or two commands at the sh prompt which seem to work for most people (some are SOL). Yes, grandma can easily copy and paste those commands... ergo it's easy to use?

</OffTopicHateRant>

Now just so I don't get marked up as off topic.... ATI drivers suck and I have no reason to believe that the new drivers will be any better because I'm bitter over past experience. Nvidia drivers work well without a lot of effort. Intel has uber cool open source drivers which are good enough to turn your desktop into a spinning cube, but don't expect too much more--at least your not using those "evil" binaries from nVidia and ATI/AMD.

Re:Hey AMD, A tip for you.

[dbIII]

Underneath it all it's still linux, so ignore the distro. If the nvidia driver doesn't work with the kernel ubuntu comes with by default then upgrade the kernel to a newer ubuntu package of it, install whatever they call the kernel development stuff on ubuntu if it comes seperately (kernel-devel, linux-src or whatever) and then run the nvidia installer from the nvidia website.

I've put a variety of nvidia cards in quite a few desktop machines and the only hassles I've had are for things like the TNT2 stuff being in a seperate driver now. I even managed to get ATI and Nvidia cards to play together in a dual head setup. It should be obvious that you may get hassles using a GUI tool to set up display hardware - so avoid the ubuntu updater and do it with text and the instructions from nvidia. Debian politics possibly also may be creating problems - they are opposed to this kernel module so it's possible the default installer for Ubuntu may not be set up to handle it.

Re:Hey AMD, A tip for you.

[Anonymous Coward]

Wow I'm sorry to flame but this just has to be said.

Really? You have to ask? Was that a serious question?

Because it has NO FUCKING 3D SUPPORT, DUMBASS.

Some of us actually USE our computers for many things. No 3D is a HUGE DEAL.

2D performance is pretty much a non-issue nowadays. Nobody does 2D benchmarks anymore because all cards are good enough at every resolution they support. If I didn't care about 3D, I would use the cheapest card available that supported the resolution I wanted.

So maybe people buy expensive video cards because they want to use them. Otherwise the $10 bargain bin special would have been good enough for them.

In other words, YOU ASKED A VERY STUPID QUESTION.

Re:Hey AMD, A tip for you.

[Anonymous Coward]

uhm...because powerbooks do not have a printscreen key? And upstream cannot get their act together to be able to map the key anywhere via a sysfs interface or some such?

Re:Hey AMD, A tip for you.

[gsking1]

I completely agree with you on the nvidia cards. I have used them quite a bit in linux and been happy in Ubunutu as long as I don't make too many changes to the system. I even played all of Doom 3 in linux (using nvidia). However, the problems with Nvidia cards (and non openness) arises when I start to do things such as compiling my own kernel, or switching to newer drivers from the nvidia website. I always have problems with getting the Nvidia graphics to work and spend way too much time messing with it. For now I'm only using the onboard Intel video in my new computer and it has been great so far. The open GL from the intel chip is acceptable for anything (such as google earth) but the games. I've never used the AMD stuff, but this is great news - at least it can be considered now.

Re:Hey AMD, A tip for you.

[Zonk (troll)]

Try apt-get update && apt-get dist-upgrade. The latest nv driver seems to be a bit more stable, or maybe it's the latest X.Org. The system is completely up to date. I really only have stability issues with it when I do video, especially if I try full screen. It's strange.

It's nice to see DAAMIT finally getting there

[Trelane]

It's only been 3-4 years since I bought an ATI card in the (vain) hopes that they would continue supporting X devs. Sadly, I found poor support and lots of bugs. Unless they pull an Intel and release/fund Free drivers for their graphics chips, for me it's Intel for ease-of-use and NVidia for performance. I've lost faith in them.

Underwhelmed

[Anonymous Coward]

Have ATI even stopped violating the GPL by shipping old code from AGPGart in their binary? This is too little too late, I've already given up on high performance 3D and decided to stick with intel graphics because of the open drivers. What's the betting this 'driver' requires mono? Seriously, last I looked the windows drivers required the .NOT framework for the craplet and settings manager.

Re:Underwhelmed

[Svartalf]

Well the settings manager happens to be a Qt app if that matters...

I'm glad they woke up- but it may be a bit late unless the rumblings of turning up the volume on the open source side
of things on their end are true.

Re:Underwhelmed

[Anonymous Coward]

Intel drivers are OK, but crash EVE under Wine. Actually, X crashes. You can trigger it by downloading the eve client and scrolling up/down in the license screen. That is on Macbook.

nVidia, no problems. *ever*

ATI? hehe, haven't bothered with ATI since Radeon 64MB AGP card was @ $300 a pop. There were messages all over the forums about great ATI drivers. ATI sponsoring OSS developer(s) to make real, OSS drivers, unlike nVidia's closed source driver. Well, ATI *never* delivered. And there is NO excuse. I will NOT buy ATI mobos chipsets (AMD now) for desktops. On-board nVidia chip? Perfect! Until ATI has good drivers and Intel AND nVidia both screw up Linux support, I will not be an ATI customer.

AMD should have merged with nVidia instead of ATI. nVidia at least has Linux support and AMD would get more business from me instead of less. Oh well....

Re:Underwhelmed

[Akaihiryuu]

Well, if it makes you feel better, I haven't succeeded in getting the ATI settings craplet to work even in Windows. I have an X850XT and an Athlon XP 3200+ (no PCIe, DDR2, or any other fancy new stuff), I made sure to install .NET 2.0 (which the craplet requires), and it just gives some strange .NET error when it tries to start. The driver itself seems to work fine, I haven't had any problems in any games I've tried to play, but I've never been able to access the settings. The same thing has happened on 2 different Windows installs, I don't know what the deal is. I haven't had any problems at all on Gentoo with that same computer, although I can't install the ati-driver-extras stuff, because it depends on an older version of the driver (?) which won't compile against my current kernel (2.6.22.1 vanilla). I haven't had any problems with the latest ati-drivers though. All my other machines except my server (which doesn't have x.org installed) use the open source radeon driver which I've never had any issues with, I just hope it eventually supports the R480 (which the X850XT is). Not that the closed source driver has given me any technical problems, but for moral reasons the open source drivers are far better. I haven't heard whether or not the GPL issues have been resolved.

still no support

[jameseyjamesey]

There is still no support for all-in-wonder cards. Nice try though.

Re:still no support

[etnoy]

Not even support for the r100-based cards that can be found in, among others, the lovely Thinkpad T30:s. Why? Because ATI decided it was too old for it. The open source radeon driver does a great job, though, and I can use Compiz Fusion with no problems on this four-year-old laptop of mine. Four years! Too old for drivers, you say?

Re:still no support

[brunascle]

ATI is apparently discontinuing the all-in-wonder cards. there's some major incompatibility with the drivers and Vista that they cant find a workaround for, so they're stopping them all together.

Re:still no support

[Anonymous Coward]

1982 called. They want there video card back.

I definitely made a brilliant move

[erroneus]

In previous discussions about ATI and their Linux driver support, I had mentioned that I made the bold move to move away from ATI on my laptop to nVidia. (Dell makes these kinds of changes fairly easy) My laptop is an Inspiron 8600 which I had originally ordered to use the ATI Mobility 9600 card. Through eBay, I ordered and later installed the 128MB version of the nVidia card to replace it. (Not terribly expensive either.) I just checked AMD/ATI's web site to see what the current hardware supported under the current driver is. Sure enough, my mobility 9600 is now at the very bottom of the supported hardware list and with the new release, it is certain to fall off entirely.

If it hasn't been stated clearly enough in the past, I'll state it again. Even if you don't care about whether a driver is OSS or proprietary from a technical standpoint, users are advised to understand that proprietary drivers places control over your hardware's obsolescence firmly in the hands of the manufacturer. And these days, with limited hardware selection for things like laptops or very tiny PCs, your options are pretty limited. These proprietary drivers are damaging the viability of Linux on older hardware which has been one of Linux's strongest motivators for adoption.

Moving to nVidia helps because at least with nVidia, they have a legacy hardware program to support and update drivers for older hardware. AMD/ATI does not. Ultimately, though, I should probably settle in and get comfortable with the OSS drivers for my hardware even if the performance is lower... it's a damned shame though.

Re:I definitely made a brilliant move

[nametaken]

I think the big problem is not that people don't understand the pitfalls of proprietary drivers. I think it's more that people buy hardware first, and opt to install an alt OS down the line. Aside from myself, I don't know anyone who was careful to purchase a computer that would be well supported by anything other than Windows.

The net result is that a LOT of people end up with ATI video cards, not wanting to buy replacements, and aggravated that driver support sucks. It's a crappy situation all the way around. :(

Re:I definitely made a brilliant move

[TheRaven64]

Looks like the ATi card in question has (unstable, but more or less working) support in DRI at the moment. ATi tends to drop support for old hardware with their blobs once there is good community support for it, so it makes sense that it would be near the bottom of the list.

I try to avoid ATi hardware in general, but I've had good support for it under FreeBSD with the DRI drivers. Their own code tends to cause crashing on any platform I've tried (OS X, Linux and Windows).

Re:I definitely made a brilliant move

[erroneus]

This is very true... the FIRST go-around. Many of us tend to learn from that though... my first lesson was long ago with Sony and the NeoMagic video chipset... that video REALLY sucked. I learned to avoid it and I'm guessing everyone else has as well since I haven't heard of it in years. And now people are learning the same with ATI... the lesson is coming slower for many people, though, as ATI is rather entrenched. Fortunately, however, Vista and Direct X are pushing Windows users in the same direction for similar reasons, so it's not just a Linux thing.

Re:I definitely made a brilliant move

[zlogic]

Moving to nVidia helps because at least with nVidia, they have a legacy hardware program to support and update drivers for older hardware. AMD/ATI does not. Nvidia didn't make a Vista driver for Geforce 4 (and older versions), meaning it works in some kind of safe-mode with no 2D acceleration. Widgets are redrawn slowly, and scrolling large text or HTML pages is also slow. And WinXP drivers cause random BSODs.

Re:I definitely made a brilliant move

[Svartalf]

The 9600 is an R300 chip. They don't plan on cutting R300 support for a while yet.

Otherwise, your observations are largely dead-on. Unless the rumblings of ATI stepping
up the pace with helping the Open Source community do their own drivers or a jointly
developed driver (we can only hope...) eventually, they'll drop support for your chip.
Having said this, the NVidia chip is only a better supported version of the same problem
really.

Re:I definitely made a brilliant move

[Anonymous Coward]

The 9600 is supported by the reverse-engineered open source ATI driver. It is included in x.org yet for some inexplicable reason it is nearly completely ignored in all slashdot graphics driver discussions. As a happy user, I couldn't be more baffled.

Re:I definitely made a brilliant move

[celle]

nvidia is no angel either, they've been dropping old chipsets from their drivers as well.

Re:I definitely made a brilliant move

[Annymouse Cowherd]

no, they've been moving cards to nvidia-legacy.

Re:I definitely made a brilliant move

[AbRASiON]

Which model nvidia card did you purchase to replace the ATI? I have exactly the same problem with my Dell 8600.

What's worse is Ubuntu 6.10 was fine but 7 - not so much.

AIGLX....

[werdz]

All I want is AIGLX. If I get AIGLX I'll be happy.

Re:AIGLX....

[Anonymous Coward]

read the article, page 2, wait a month ... masturbate of happyness.

And?

[n0dna]

Even a blind pig will find an acorn occasionally.

Lets suppose that this driver does all it says, and more. That'd be one in a row for ATI. They have even had drivers that will sometimes work under Windows. Not very often, and not by any stretch routinely.

Why would I put my money behind a product that I can be fairly certain will never have another driver that will ever work?

Re:And?

[eqisow]

Because ATI is now owned by AMD? I'd say that gives them a clean slate. AMD is clearly trying to turn things around.

Re:And?

[n0dna]

Unless they fired everyone responsible for writing the drivers and the entire QA department, then that still makes this one in a row. Even if AMD holds them to a higher standard, it'll take at least one more good driver to convince anyone of it. Like I said, ATI has had working drivers once or twice before.

Besides, people have a long memory when it comes to garbage hardware. A $40 game that blows can be a fluke. A $200 (or more) video card that only does 640x480 in 16 colors is harder to forget.

Once a company burns you on hardware, there's no reason to ever have to go back to them if there is any competition at all. Look at the other options you have for graphics. Hell, people are even using built-in video instead of ATI. How bad do you have to be for people to prefer onboard video?

Re:And?

[DirkGently]

Mod this guy up. On the merits of that last sentance alone.

Re:And?

[WhiteWolf666]

I know what the "proof" will be; or at least, the start of a trend.

Full AIGLX support in 8.42 (the article is discussing 8.41). The claim at Phoronix is that AMD has claimed AIGLX is going in at 8.42.

Continuing the trend would be MPEG-4/H.264 Xvideo support in 8.4x or 8.5x, preferably within the next 6 months or so (keep in mind that the Radeon 2X00 series have excellent video capabilities).

If they hit those two goals, I'll most likely purchase 2-3 ATI cards for my Linux boxes; the AIGLX and Xvideo things are a big deal to me, and Nvidia cards don't currently accelerate MPEG-4/H.264.

Re:And?

[Anonymous Coward]

Hell, people are even using built-in video instead of ATI. How bad do you have to be for people to prefer onboard video?

My onboard video is ATI, you insensitive clod!

Re:And?

[baadger]

I think the problem is there isn't an open source framework out there with an API ready to accelerate H.264/MPEG4 AVC. XVideo only accelerates colour space conversion and scaling. XvMC only does motion compensation and is only compatible, as far as I know, with MPEG 1 and 2 (It works with one of the ffmpeg projects decoders).

Windows actually lacks such an API too, but thats not surprising since various companies out there (Cyberlink for example) want to *sell* you an H.264 decoder that supports NVidia's proprietary acceleration API (PureVideo HD).

Actually you have been previously been better off on an FOSS system with XvMC than with MPEG 2 acceleration provided on Windows by Nvidia's own PureVideo (pre-HD branding) MPEG2 decoder, which wasn't free (~$30 USD)

Re:And?

[Samizdata]

When I recently did a large upgrade on my main desktop (which is Windows, yes), I finally gave up on ATI and went Nvidia. I got tired of getting "Zero Display Service" errors, and always waiting for the latest Omega drivers...

If you want Linux and open

[AHuxley]

Please try and support The Open Graphics Project.
http://wiki.opengraphics.org./tiki-index.php?page= AboutOpenGraphics

Re:If you want Linux and open

[tomstdenis]

While I wouldn't be openly against the idea of an open graphics processor, I don't see the need for it. So long as the GPU enables me to use an open API like X or OpenGL, who cares really how it's done underneath.

Because honestly, who's gonna pay for support? Or afford to recall defective chips, etc? At least if my nvidia card doesn't work the store I bought it from knows they can return it back through the chain to eventually get their money back. If I buy some no-name card from a small time manufacturer, returns might not be an option, etc, etc.

As for the "binary blob vs src" debate, again I don't see the problem. Once nvidia stops releasing drivers I'll stop buying their cards. Until then though, I'll use their cards because they "just work."

Re:If you want Linux and open

[Kjella]

Please, it's an interestic academic idea but GPUs are now more complex than CPUs and demand huge R&D efforts. They have a snowflake's chance in hell of producing anything competitive, and that's if the snowflake is taking a lava bath during a heat wave while soaked in gasoline and lit on fire. If you don't want top performance there's Intel, and then you at least get an actual, well-working product. Hopefully they can bring open drivers to more performance oriented markets too, but even for Intel that's a huge investment. This project just doesn't have anywhere near what it takes.

Re:If you want Linux and open

[TeknoHog]

If you want Linux and open, you can use Intel's graphics chips right now. They have opensource drivers in the stable kernel and X.org trees. If you need badass performance for the latest games, I don't think OGP will be much better than Intel. But for example, my oldish Centrino laptop runs things like Tuxracer and Quake 3 smoothly, so the basic 3D stuff definitely works.

On the other hand, I do appreciate a good hardware hack, but that's a completely different realm from most Linux geeks' needs. A completely opensource (software) system is easy to have right now, but with an open graphics chip you'd still have most of your hardware closed.

Re:If you want Linux and open

[init100]

Or you might want to try this: AMD to open up graphics specs

the forgotten ones...

[Anonymous Coward]

I have a x700 and I _know_ it is going to be my last ATI card in much time (i.e. forever or when the issues get fixed, whatever happens first). I know, I know, FreeBSD may not be a mainstream OS, nor a first choice for a desktop computer but at least Nvidia shows some interest for their customer. (psssst, ATI execs, a clue: your customer are the ones who make you earn money to buy those fancy cars you own...)

Re:the forgotten ones...

[Anonymous Coward]

Please point me to a place where I can download a proprietary nvidia driver for FreeBSD/amd64. IA-32 has been obsolete for 3+ years. I didn't buy a 64bit CPU to run 32bit junk.

Glass

3D is important; Do what Linus does: buy Intel

[Kludge]

98% of all Linux machines are used for tasks where 3D graphic performance doesn't matter.

Wrong. Many Linux machines are now desktops. 2/3 of the Linux machines in my home are desktops. I don't use fancy 3D desktops, but I do use everyday apps like Google Earth and the occasional kids' games that are much faster and smoother with hardware rather than software OpenGL.

However I have solved this problem by only buying Intel graphics hardware. They work from the moment Fedora first boots up.

Re:3D is important; Do what Linus does: buy Intel

[plague3106]

Intel graphics are also shit compared to Nvidia or ATI.

Also, I don't think your number "prove" most linux installs are desktops. Many probably still are just servers.

Re:3D is important; Do what Linus does: buy Intel

[CastrTroy]

Yes they are. But once you bring driver quality and stability into the equation, then Intel wins hands down. I'd rather have a slower video card that actually works, than a fast one that doesn't. Also, unless you are playing games, you won't notice the speed difference. Even if you're running a 3D desktop.

Re:3D is important; Do what Linus does: buy Intel

[Anonymous Coward]

Many probably still are just servers.

Or other non-desktop devices. All my routers and external modems (heck, even my external dial-up modem) are running Linux.

Re:3D is important; Do what Linus does: buy Intel

[Afrosheen]

"I'd rather have a slower video card that actually works, than a fast one that doesn't"

That's why nearly all Linux gamers and more than 60% of Windows gamers buy Nvidia cards. They've had better drivers for ages. Not open source, which disturbs some of the hardcore, but great drivers nonetheless.

Re:3D is important; Do what Linus does: buy Intel

[plague3106]

I agree, but for some reason you've ruled out Nvidia. Great drivers and a fast card. ATI is the only company that thinks its ok to put out shitty graphics drivers.

Re:3D is important; Do what Linus does: buy Intel

[AaronW]

I'd say Intel is worse shit than even ATI (which I detest). Out of the two Intel systems at work, neither will work outside of frame buffer mode, and worse than that, they cannot run at the native resolution of the LCD monitor.

The nVidia drivers just plain work. They detected the monitor correctly and have worked flawlessly ever since. Open or close source, they are by far the best drivers I've used. The ATI drivers I use on one of the machines support the monitors, but introduce periodic 2-D corruption when running Xemacs and corrupt the cursor when moving between Xinerama panes, but at least they can do 1680x1050 and Xinerama. I don't even mind that the 3-D is slow since this is a work machine and I don't really need 3-D.

The Intel drivers are far too slow at 2-D, and given I can only do 1280x1024 or 1600x1200 and both look like crap on the LCD monitors. One machine is a P-4 and the other a core 2 duo machine, and both are unusable with Intel.

I only want 2-D and the Intel drivers can't even do that right.

Also with nVidia, I don't really even need to care which chipset is used as long as it isn't too old, since the drivers just work. Even the open source nVidia drivers work well for 2-D.

Re:3D is important; Do what Linus does: buy Intel

[Anonymous Coward]

Not to mention how bloody hard it is to get the Intel 915GM line to output to a TV or external monitor (at least on my monitor). I must have blown my xorg up at least a dozen times trying to get S-Video out working and the i810switch program Does not work right, I get a very distorted image. My Nvidia card in my desktop, however, works seamlessly.

Re:3D is important; Do what Linus does: buy Intel

[Schraegstrichpunkt]

The nVidia drivers just plain work.

Right...

LD [M] /usr/src/modules/nvidia-kernel/nv/nvidia.o
Buildi ng modules, stage 2.
MODPOST 1 modules
FATAL: modpost: GPL-incompatible module nvidia.ko uses GPL-only symbol 'paravirt_ops'
make[4]: *** [__modpost] Error 1
make[3]: *** [modules] Error 2
make[3]: Leaving directory `/usr/src/linux-headers-2.6.21-2-686'
NVIDIA: left KBUILD.
nvidia.ko failed to build!

Re:3D is important; Do what Linus does: buy Intel

[AaronW]

And from my quick reading on this it looks like it may have been addressed in the kernel and in fact may have been a bug introduced into the kernel.

http://lists.openwall.net/linux-kernel/2007/04/30/ 530

-Aaron

Re:3D is important; Do what Linus does: buy Intel

[Schraegstrichpunkt]

Some more quick reading (or searching with Google) will make you realize that this is just the latest way that the nvidia drivers come far from the characterization of "just works". There are other examples, like trying to get them to work at all under Xen.

Re:3D is important; Do what Linus does: buy Intel

[miro f]

try

sudo apt-get install nvidia-glx

or if you have a new card

sudo apt-get install nvidia-glx-new

always worked for me

Re:3D is important; Do what Linus does: buy Intel

[dbIII]

98% of all Linux machines are used for tasks where 3D graphic performance doesn't matter.

Probably true. Linux clusters are useful so there are a very large number of machines out there that fit exactly into this catagory. The birrare thing is many salesfolk assume that if you have a processing node you are doing movie special effects on a monitor hooked up to every node so they often try to sell expensive 3D graphics gear with it no matter what industry you are in.

Personally I think the problem that people have with nvidia will go away when the people that don't want the driver opened leave the place. Until then the open driver is good enough for a lot of purposes - the same goes with ATI.

Re:3D is important; Do what Linus does: buy Intel

[cibyr]

nVidia drivers have NOT been great lately. What with their efforts being spread by the 8800, SLI and vista things have really deteriorated. There are a few (official, WHQL) drivers that simply bluescreen on boot - had to boot safe mode to uninstall them to get things going again. BSODs are a regular occurrence these days, especially when playing new or beta games - DRIVER STUCK IN INFINITE LOOP is a message I've seen a lot lately.

And of course there's no way to actually report any of these problems - if you're not running vista, they apparently just don't want to know.

Re:3D is important; Do what Linus does: buy Intel

[wilymage]

emerge nvidia-drivers

Under Gentoo, this command will install (proprietary) X11 driver and (proprietary) kernel module for Nvidia cards.

You can have the "nv" and "nvidia" drivers installed simultaneously, and they play nicely if you wish to swap between them in your xorg.conf.

Don't quote me, but I think the VIDEO_CARD="nvidia" line in your make.conf will pull in nvidia-drivers. VIDEO_CARD="nvidia nv" is allowed.

Re:3D is important; Do what Linus does: buy Intel

[Anonymous Coward]

HOLY FKING CRAP! It's 2007 and we're still hoping drivers will support shit, or not crash. Like seriously WTF!

s/Launches/Announces/

[theMAGE]

The title is misleading - AMD did not launch anything, they announced it. Just the fact that some random hardware site got a sneak peek at the driver does not change anything...

Any video accel lovin'?

[DirkGently]

I didn't see any word about MPEG2/MPEG4 offloading, or even word of proper Xv support/controls. I've got my fingers crossed, but for those of us who live & breathe MythTV, I fear it's still a one-horse town.

Re:Any video accel lovin'?

[tji]

Seconded. Linux+MythTV is already a top notch DVR option, which would really benefit from video improvements.

But, there are a few horses to choose from (all with their share of warts):

NVidia - Closed binary blob, supports XvMC for MPEG2 accel. Works some of the time, for some that try it.

Intel - Only very basic XvMC support today, but they have a very nice effort towards open source drivers and a new video acceleration API for MPEG2, MPEG4, H.264 support, and VLD support. Looks like a great MythTV option, when it's available/mature.

VIA Unichrome - some features similar to intel above. In theory it's a nice option, and has been available for a long time. In practice, not so much. Spotty support for hardware versions, most GPUs don't support HD playback, plenty of pitfalls. But, for some people, with the right chipset, and OpenChrome drivers, this works very well.

Re:Any video accel lovin'?

[DirkGently]

I'm very curious to see what comes about from the myth-vid branch of development. The devs are making a solid attempt at moving away from XvMC and using the 3D engine, as that's what nVidia does in the Windows drivers for their "PureVideo" stuff. It's got a lot of promise and opens up a lot of better deinterlacers beyond bob & weave. However, because ATI drivers have stunk so badly (and because they already own nVidia cards), that's what the devs are working with. For the forseeable future, if you want it to work, that's also what you buy.

As an aside, I've had excellent luck with XvMC for both SD and HD, though I ended up sacking the HD because the Bob deinterlace was kinda ugly and I had the CPU to spare.

Re:Any video accel lovin'?

[Mad Merlin]

fglrx currently has Xv acceleration using the GPU on R500 series cards, and it works well enough that I can watch 1080p H.264 content with no dropped frames. It's the one (and only) thing it currently beats the Nvidia drivers on.

Re:YAY

[Anonymous Coward]

Cool! Now we can play games which were on Windows only about five years ago!

Who's chasing tail lights now? In your face, Apple!

Whoa! Spiteful much?

[TheGreatOrangePeel]

What the hell is up with all the scathing remarks?! Let's remember that the ATI acquisition by AMD is new and let's be impressed, considering past support, that progress is being made in the Linux ATI drivers arena AT ALL! I really do believe that AMD is going to do the right thing by Linux. They're two underdogs that stand a lot to gain from each other and it would only stand to hurt any gains to be had by such a relationship by continuing what ATI was doing before the buyout. The fact of the matter is, ATI has undoubtedly undergone a mass re-organization and is, doubtlessly, also operating under a new philosophy. Anyone who knows someone who had their division bought out knows this to be true. Let's just sit back and see what happens before we start (effectively) blaming AMD for ATI's past mistakes and poorly written code.

Re:Whoa! Spiteful much?

[m50d]

What the hell is up with all the scathing remarks?! Let's remember that the ATI acquisition by AMD is new and let's be impressed, considering past support, that progress is being made in the Linux ATI drivers arena AT ALL!

But there's no *real* progress. All there is is more *talk*, and we've been hearing for *years* that the next lot of ATI drivers will make everything right. They never do, and so we've lost patience with all these *empty* announcements. When they actually release a working driver I'll start cheering. Not before.

Re:Whoa! Spiteful much?

[TheGreatOrangePeel]

I've ATI hardware. I know how terrible the ATI drivers are. That said, AMD has, historically, delivered what they say they will. As a result, I'm tentatively optimistic. To put it simply, the acquisition of ATI by AMD injects new hope.

I defiantly agree that they need to have something to (metaphorically speaking) hold up and point at before they go spreading the word, though.

Re:Whoa! Spiteful much?

[Samizdata]

Well, if they can't make Windows drivers that work reliably, or even install for that matter, then I don't expect much from their Linux codebase. That certainly doesn't seem spiteful to me...

Re:Whoa! Spiteful much?

[TheGreatOrangePeel]

I agree. Its just that it reads like a lot of people take it personally.

AMD to open up graphics specs

[xer.xes]

It was announced today at the Linux summit they will open up specifications for all graphics cards, and release a 'reference'/minimal open-source driver for all cards.

More here: http://lwn.net/Articles/248227

Re:AMD to open up graphics specs

[chill]

All graphics cards from the R500 going forward, specifically.

Still, THIS should be an article on Slashdot with the new drivers being a footnote -- not the other way around.

Re:AMD to open up graphics specs

[lotho brandybuck]

That's huge. I hope this is true. I hope it's done fast.
Available, truly open sourced drivers are going to be a big factor in any hardware purchase I make.
I'm just one, but I think I'm one of many. Even if you're not "paranoid" (concerned) it's obsolecense protection.

Re:AMD to open up graphics specs

[MrHanky]

I was planning to purchase a Radeon X1950 Pro (since I still have AGP and can't afford/don't need an upgrade). Now it doesn't look like such a stupid idea after all. Excellent news!

Re:AMD to open up graphics specs

[Kjella]

That's excellent news, but I'm always weary of paper releases. When the specs are available for download, and someone with more driver writing skills than me has said "yep, this is a good and complete documentation that we can actually use" then it's time for celebration. Then maybe next time I'll cosider an ATI card, it's been a long time since last time.

Re:AMD to open up graphics specs

[Svartalf]

Interesting. R500 forward would be a major improvement, really. While I'd like to have R300/R400 and better R200 info, I'll take what I can actually GET from someone giving it like that.

Re:AMD to open up graphics specs

[MORB]

Even if they don't open up the older cards, it won't matter anymore in a few years - as long as they stay commited to open up the specs for all their new cards.

As far as I'm concerned, unless nvidia follows my next graphic card will be an AMD.

Re:AMD to open up graphics specs

[Anonymous Coward]

Hmm, that R500 forward imples that R300+ chips have some IP issues, or at least they are so buggy that AMD/ATI don't want to be embarassed :) - but the thing is already reverse engineered, at least big part of it and those who did it really did discover that chips are buggy

IMO, AMD should hire some people to work on merging their proprietary driver back to upstream DRI/DRM and enhancing the framework in the process, though the closed driver they just released probably is some kind of unified Vista+Linux architecture and hence it's probably very complicated to maintain it as a single source tree while at the same time have it working on DRI and be a public OSS project (if they don't want to open up Windows driver code).

I don't expect OSS driver to be as fast as closed one. It should be a clean and as stable as possible design, not compromising stability (and security!) for a few FPS. After all, it is dirty hacks and various "secret" math approximations that squeeze last few bits of performance out of cards and such nasty things are well hidden from public. ATI and Nvidia probably have dozens of experts with a full-decade experience of driver-level GPU/VRAM optimisation tricks for games (NOT for composited desktops though!).

Re:AMD to open up graphics specs

[Svartalf]

That'd be my take as well. I've got 1 R300 and 1 R500 based card as it stands that I obtained to do verification of game code ports against the "other" GPU in the space. The R500 support & specs got my attention- if they keep doing things this way, they'll keep it.

Money where your mouth is

[MrNemesis]

Since the beginning of the year? Hell, I've been hearing murmuring for years on "support for XYZ will be coming soon!" - and yet today the disparity between the ATI/nVidia feature set and stability under Linux are still huge. How long since nVidia got support for AIGLX? ATI only just adds it now?

You'll also note that, GeForce 8x00 series notwithstanding (which are marginally slower under Linux), nVidia maintain a very small performance delta between the Linux and windows version of their drivers. ATI's performance delta can sometimes be as much as 50% (top-of-my-head BTW, Phoronix had another full-of-crappy-graphs article about it a while back).

I'm hoping AMD can pull some weight and at least get better support for laptop chipsets and IGP's in their otherwise pretty nice chipsets. Until then, I have to stick to Intel or nVidia for graphics, and since I only need the one gaming box, I'm getting through alot of Intel motherboards. Guess what CPU goes in an Intel motherboard, AMD? Despite me wanting to use X2's for their lower idle power envelope, I find it hard to justify.

Sigh.

Ungrateful...

[spirit of reason]

Holy cow, folks. Can't you be just a little happy that ATI has finally gotten their crap together performance-wise before criticizing their potential lack of openness? I would like OSS drivers as much as the next guy, but at least we ought to appreciate AMD/ATI is finally putting some effort into Linux. Besides, one of those Phoronix articles is insinuating that there is more for the OSS community coming than just higher performance.

Re:Ungrateful...

[Anonymous Coward]

No. How many instances of vendors leaving nasty buffer overflows and other badness in their closed source blobs (leading to exploits down the road) will it take to make you see that this is not good? Blobs which have to interact with the kernal are BAD. I'll stick with Intel chipsets/video until ATI and Nvidia start releasing drives I know won't bite me in the ass.

Re:Ungrateful...

[LWATCDR]

I am looking forward to reviews of the new drivers. I need a new Video board and an 1950 would be a nice card for the money.
But they are not supporting the all in wonder boards so that is a big negative for me.
And they are not supporting their older GPUs which is also a negative for me.

So I will have to be in the wait and see mode.
As far as openness? Well I use nVidia now. If I have a choice between two good boards with good drivers I will pick the one that are GPL. Intel isn't an option since I currently have AMD systems and I like them.

Re:Ungrateful...

[KillerBob]

I'm actually going to wait to see how it handles my Radeon XPress 200M on my laptop before I say they've gotten their act together. That card has 128MB soldered on. Dedicated 128MB. That the Linux driver simply can't see. I have to use 128MB shared memory, taking away from the rather anemic 1GB of system memory taht I have in the lappy. And even under those conditions, performance is barely better than using the Mesa driver. Completely unusable for games, even though I never had any problem running games such as GuildWars or SW:KOTOR under Windows.

When ATI comes out with a driver that can actually see/use the 128MB dedicated graphics I have on my laptop, *then* I'll say they've gotten their act together. Until then, I'll keep using the open source driver.

Re:Ungrateful...

[spirit of reason]

Yeah... that's not cool. It seems like cards with hypermemory or turbocache are hit and miss on both sides.

But geez, Phoronix keeps hinting at open specifications or an open source driver, it's driving me mad (for example, see here). I have only used nvidia cards thus far (because performance was my top priority--games, CAD, etc. really need it), but that could quickly change if AMD does either of the above. If they do, surely someone from the community will fix those problems for you (if they still exist in 8.41).

I would love to get into coding drivers for graphics cards and help fix things myself, but... I've only done very simple firmware for CAN and serial chips in an embedded AVR system (no OS). I need a good book (preferably free ;-) ) on Linux device drivers, I think.

Re:Ungrateful...

[spirit of reason]

Ahh, now I see why there's a redundant mod. I didn't pay close enough attention to the person a few threads up (who may have been below my threshold at the time). Sheesh, give me a break, lol.

It's about time!

[WhiteWolf666]

Awesome!

Even though its not "out" yet, there are plenty of benchmarks available. It'll be out soon.

What does this "prove" for me? That AMD's commitment to make ATI a first-class contender on the Linux front was for real. I'm guessing that Windows users will also see improvements in OpenGL performance, and we'll see better adoption of OpenGL on all three major platforms (Windows, OS X, Linux).

I'm happy as hell about this. About time us Linux users got to take advantage of GPU price wars!

I'm still an NVIDIA fan, because they've been good to me for all these years (on Linux), but I'm at least willing to look at ATI these days; particularly because the ATI peripheral GPU software is much better (better control panel, better install program). I wonder if the driver quality is good (not just performance, but does it always compile correctly, does it always fix broken installs (the way NVIDIA's does?)).

This is a good day for Linux.

Bah. Fuck ATI

[Greyfox]

Ironically you have a lot more choice on your Linux PC than I do on my Apple desktop. Thing comes with a defective ATI video card that overheats the moment you actually try to make it DO anything. And my choices are another (probably defective) ATI video card or a less capable Nvidia one. Well there's always a massive amount of suckitude associated with my experiences with ATI, from months spent with no PCIe support on Linux to drivers that would randomly break X to outright defective hardware. And now promises of a new driver that's going to crap daisies and rainbows? I'll believe it when I see it. Until that time, I'm going to be going out of my way to avoid buying any more ATI hardware.

Re:Bah. Fuck ATI

[leoc]

Ironically you have a lot more choice on your Linux PC than I do on my Apple desktop.

How is that ironic?

A little late.

[seebs]

Thanks, AMD!

Unfortunately, under the existing driver, any time my rogue applied poison in WoW, I had a full half-second freeze. Which means that, last december, I got an nVidia card.

Since I only upgrade graphics cards every couple of years, it might be a while before this matters -- especially because, until the "complete freeze on certain texturing operations" bug is documented and acknowledged and someone who had it before tells me it's fixed, I'm not about to buy an ATI card on the off chance that it might work.

Re:A little late.

[Anonymous Coward]

Same here.. I gave up on ATI under Linux a long time ago. I once used them on my Windows machines, but I realized that as I upgraded my Windows machines, I'd put Linux on the old ones. But ATI drivers sucked under Linux, so now even on my brand spanking new Windows boxes I'm choosing NVidia to be ready for when Linux gets installed.

X.Org Mafia

[Anonymous Coward]

So if you are a graphics card vendor and you don't give money to the X.org mafia, they will call you a liar and a thief and nobody will buy your hardware (not only the graphics card, but even CPUs and other perpihals, even if faster/cheaper).

On the other hand, if you pay a wealthy amount to the x.org mafia, this will not happen.

Screw X. Framebuffer is just fine.

Opening

[alexgieg]

I don't think ATI and nVidia, the two big graphic chipset manufacturers, will keep their drivers closed for much more time. GPUs are more and more being seen as advanced mathematical co-processors rather than "mere" gamers' hardware. Keeping them closed is akin to keeping most if not all of a CPU's opcodes closed under NDA's. What good would that do to a CPU manufacturer? There'll come a point where software companies will simply start demanding open low level access to GPUs for performance improvement purposes (think advanced video editing, strong cryptography, grid computing etc.), and it'll be hard for GPU manufacturer to offer any reasonable explanation for not providing it.

Re:Opening

[andrewguy9]

I'm not sure that is true. nvidia's cuda c compiler does not open anything, and yet it allows me to write / run arbitrary c programs on my 8 series gpu. The culture in the gpu market is the problem. Those guys are paranoid about loosing ip to eachother. I mean seariously, they got into a fist fight last year! Tensions run deep. Hopefully Intel's offerings will continue to improve and put pressure on the big boys to calm down and open up.

Re:Opening

[init100]

I don't think ATI and nVidia, the two big graphic chipset manufacturers, will keep their drivers closed for much more time.

If the quote from the Linux kernel summit is true, you may be right: AMD to open up graphics specs

Now I wonder how nVidia will handle this. :)

Re:Opening

[Anonymous Coward]

I think that AMD wanting to add GPUs as one of its cores on die would cause this sooner or later anyway. Once the GPU is on the same piece of silicon as the CPU, they'll pretty much have to open hardware specifications, as its no longer just an addon card isolated by the OS, but a core piece of tech to the computer. Nvidia may be king of the hill now, but I really don't see it staying that way once things like that start being deployed. Intel has its own GPU branch, AMD now has its own. Nvidia is going to have to make some tough choices really soon, if it doesn't start doing something, anything, to improve its position in that market, it could quickly find itself being obsolete. So ya, its a good bet we'll start seeing open hardware specs on all manufacturers.

Pathetic

[BlueParrot]

So... I don't even bother trying the fglrx drivers since the reverse engineered free driver is more stable, and actually works. I mean seriously ATI, a non-profit project which bases its code on guessing how your hardware works has not only better, but in some cases superior, stability than your shitty driver, that really says something. I think it is time for a bad car analogy. Imagine a driver who memorises the layout of the town by carefully noting down where his car crashes as he drives. This guy's taxi company is currently beating your top of the line staff, even thou you have a full map of the town, a military grade GPS receiver, and real-time information about traffic congestion. Oh, and btw, your competitor's car has opaque windows, can only use the reverse gear and he is only able to turn left. Even so, the customers prefer him in front of you. In short: You suck! Big time...

Well...

[Xenographic]

There WAS that request from some PR type in the Firehose to give Slashdot an interview over the subject. So I suspect that they plan to do something. Because if they're plotting a PR blitz over a lot of nothing, that'd only backfire.

Honestly, I suspect this has more to do with Dell selling Ubuntu than anything. Hopefully that'll at least get them to improve the drivers, although I wish we could get some open ones. I mean, the kernel team is willing to work under an NDA... what more do they need?

AIGLX

[Anonymous Coward]

Where's AIGLX? Well, unfortunately due to delays later in their development cycle for the 8.41 driver, AIGLX was pushed out of this month's development and release cycle. The AIGLX support will, however, arrive next month with the fglrx 8.42 driver. Oh, for fucking shit.

Great News

[Anonymous Coward]

For you fucktards who are so dumb that you *have* an ATI card. Their drivers once told me my card, with an ATI logo on the box and RADEON in big fucking letters, was not an ATI card and therefore unsupported. No idea why the screen was solid black.

I switched to a company that can make drivers that actually work with third-party boards, without whining and crying. Long live NVIDIA!

Multi Head Support with multiple graphics cards?

[e-Trolley]

Does anybody know if this driver supports multi head (Big Desktop) across several physical cards?

The reason why I ask is because using built-in Xinerama of xorg/xfree86 doesn't support DRI, and it looks like it never will...

I need at least 3 monitors and at the moment no affordable single cards has that many outlets, and living without DRI is a PITA.

As the ppl from xorg don't seem to implement DRI for Xinerama my only hope is a (proprietary) driver.

ATI Mobility FireGL 9000

[exi1ed0ne]

Is there a particular reason you need the driver to be open-source?

I have a perfectly good 3 year old laptop with a video card that ATI decided to drop support for. The last proprietary driver that it IS supported on (8.28.8) will not install on Feisty. My options are exactly:

1. Deal with it
2. Go back to windows (shudder) or Dapper
3. Drop a bunch of $$ on a new laptop (can't replace just the card now can I?)

Exactly none of those options is appealing, so I won't be buying ATI again until they open up.

Re:ATI Mobility FireGL 9000

[Constantine XVI]

In Feisty, my Mobility Radeon 9100 works perfectly (ie: compiz fusion) with the default OSS drivers.

Re:ATI Mobility FireGL 9000

[exi1ed0ne]

Interesting. Mine doesn't with the stock drivers for compiz.

He's come to the right place...

[Anonymous Coward]

> He's going to beat them through shear force of will by not thinking anymore.

Well, he sure came to the right place!

Forget new features, just fix the current ones.

[Frenchman113]

Before announcing/implementing these wonderful new useless features (XComposite/AIGLX), I'd like to see existing functionality fixed.

fglrx currently can handle dual-head setups in one of two ways. The first: you run two X servers, which wastes memory and makes you unable to move windows between them. To make it worse, Xv doesn't work on the second head. The second: use a combined framebuffer/xinerama setup which works if your monitors are exactly the same resolution but otherwise forces one to change res.

It is impossible to use both OpenGL and Xv at the same time with fglrx. That's just plain stupid.

VSync doesn't stop OpenGL video playback from tearing, it just makes the video tear diagonally/in the corner.

fglrx is just plain unstable. I've had it regularly lock up my machine with no apparent reason.

fglrx isn't even fast. Running 3dMarkSE2001 in WINE (not a great benchmark by any means), I get a good 60ish% performance loss compared to Windows. That's just plain stupid.

the rest

[celle]

How bout the rest of us in operating system land? *bsd, solaris...

Dual/Multiple Head?

[man_ls]

I've got an Ubuntu FF installation languishing on my hard drive in a secondary partition to XP, because I've never managed to get my 3 monitors spread across 2 ATI videocards (X800 and 9250) to work correctly.

Does this release make it any easier, at all, to make it work?

Something tells me from reading the rest of the comments that the answer is "no" but I don't know for sure.

Yes, but

[Anonymous Coward]

does it run on... um, sorry.

X1950 AGP

[Anonymous Coward]

I hope this driver atleast works with the X1950 AGP series cards so i can get better performance than with the Vesa driver.

I'm an old linux user that have been using linux on and off for 6-7 years but i have more or less stopped using it since my budget doesn't allow me to keep several graphicscards depending on what OS i feel like running at the moment.

Hooray!!!

[logicassasin]

... does this mean they've got time to fix the broken Rage128 driver for XP???

Whoa - what about winblows?

[Heembo]

Does this mean ATI will actually deliver stable drivers for windows? That would be a first!

Whats the point?

[steveaustin1971]

Last time I checked noone was making any games for Linux that needed a high end graphics card anyway... why would you put one in your linux box?

Re:Whats the point?

[houstonbofh]

There is this new Company called Id Software. They plan an entire line of games with native Linux support. I think the first game will be called Quake Wars : Enemy Territory. Or they may have had 1 or 2 games before it.

Re:Whats the point?

[steveaustin1971]

oh sorry I meant GOOD games, I have the Beta for ETQW and its pretty lame... how about something ppl actually play? Battlefield series? Source? HL2? Bioshock?

I pick shutup

[DarkTempes]

I'm pretty sure the whole reason ATI is in the mess of being regarded as having crappy linux drivers is because they've released crappy linux drivers. I'd be fine with them releasing some sort of 'beta' driver to help appease people with currently unsupported hardware but they should not put out a real release until it's tested and working properly.

And in my opinion the only other thing I would request out of them until that time is silence on the matter.

Hopefuly good enough

[asm2750]

If its the same or better compared to their chief competitor's linux drivers, then I buy a card for my next linux machine.

May I have those drivers in a Black Box please?

[WheatGrass]

;-) It's good news.

Re:And?- They need to work though

[Anonymous Coward]

The can claim x,y,z and if the driver just doesn't work on your system it is completely worthless! I have spent way to much time and many late nights trying to get direct rendering to work on my laptop(works with older driver), only to get
glxinfo | grep rend
direct rendering: No
arghhhhhhhhhhhhh!!!!!!!!!!!!
From what I can tell with my 9600 mobility series and it doesn't work right with the current releases(anything with pcie support). The kernel incorrectly returns the memory size of my AGP card, so I have to force the driver in PCI mode. Forcing the driver into pci mode causes it to fail when trying to allocate memory. I would put money on it that forcing an AGP card to PCI mode will cause issues on many machines with the current drivers. If you ask about distros I have tried many, gentoo, fedora,suse and finally ubuntu. Only reason I am working with ubuntu(edgy)is the kernel is older so I can use the older drivers. I am using ati driver that is about year old but it works and I can get xgl working.

They will be open!

[Anonymous Coward]

Actually, they will. And I'm not sure why this story also isn't on /. since this is much bigger news!

AMD To Open up Graphics Specifications
"A quick report from the kernel summit: AMD's representative at the summit has announced that the company has made a decision to enable the development of open source drivers for all of its (ATI) graphics processors from the R500 going forward. There will be specifications available and a skeleton driver as well; a free 2D driver is anticipated by the end of the year. The rest will have to be written; freeing of the existing binary-only driver is not in the cards, and 'that is better for everybody'. Things are looking good on this front. More in the kernel summit report to come."

http://www.osnews.com/story.php/18573/AMD-To-Open- up-Graphics-Specifications/

I'll wait for it and test it myself

[fudgefactor7]

Sure, they can claim all the want that they have their act together, that's fine; but I will be the final arbitrator of that morsel myself. If their 8.42 driver doesn't fix the issues I've experienced for a while now (see my journal for the info), then they can be assured that they are not going to get more of my hard-earned cash. I've been able to prove that FGLRX + Linux + 3D = broken. If they can fix that, I'll be happy. Mesa's driver doesn't fail like the ATI driver...pity you get far less performance from it...but at least it's stable.

ATI: don't prove yourself a liar with this. I will wait. I will see. Until then, give me no hype: give me results instead.

Re:I'll wait for it and test it myself

[paynesmanor]

Dude your Problem is not with AMD.. You need to have your facts, Before you go Bashing a company.. Its so easy to lay blaim on suspicion alone. This Linux driver is not going to solve your problem, Because its NOT a AMD issue... DUH...

Re:I'll wait for it and test it myself

[Anonymous Coward]

Uh....yeah, it is. The OP is right. AMD is the owner of ATI. ATI's driver is notoriously messed up. The blame goes up the chain.

Why it is important that the driver be open source

[einhverfr]

If you read my journal, you will see that I am hardly the Stalin^H^Hlman type. I tend to think that we should be looking at pragmatism as a foundation of idealism not the other way around.

THe basic issue, however, is that nVidia drivers generally (at least in my experience) tend to require more effort to get installed than they should if you start to move outside of the most common process ro configurations. For example, on one of my Fedora boxes, I can get a kernel that works well with my processor or a precompiled 3d video card driver. I cannot get both. This gets worse as you move away from IA32/64 architectures.

So if you want every Linux user to be able to *easily* use the driver, it *needs* to be open source.

I'm still waiting...

[Trevin]

for basic features like Resize & Rotate (especially rotate), and dual-headed display on monitors with different resolutions.

What's so hard about providing the same features in their Linux driver that they have in their Windows driver?

Let your faith be restored

[deek]

Unless they pull an Intel and release/fund Free drivers for their graphics chips, for me it's Intel for ease-of-use and NVidia for performance. I've lost faith in them.

I'm repeating what others have posted already, but what you wish for looks to be happening soon.

http://lwn.net/Articles/248227/

Re:Why it is important that the driver be open sou

[Afrosheen]

There's always the ancient nv driver if you want something open and free. It gets the job done, at least in 2d land. If you want a free driver, I guess we can all petition Nvidia to strip SGI's secret OpenGL magic out of the drivers, leaving us with a substandard card and a wonderfully open driver. It's been discussed to death, and the reason Nvidia won't give us an open driver is because they can't. At least, they can't give us one that performs as we expect Nvidia cards to perform.

what about Xv on AVIVO cards?

[chmod a x mojo]

What I want to know is.... is Xv working again on the X1xxx series cards again? I know it worked on the older drivers ( 8.24-ish?) but the drivers in debian stable ( 8.36? ) Xv was never there. Not to mention that watching video with GL or GL2 rendering would totally crash X whenever you hit the stop button. If they have Xv working ( and the AIGLX support I would happily throw my X1600 back in.... as it is for dvd / video watching my ti4200 outperforms it.

Re:Why it is important that the driver be open sou

[einhverfr]

I don't doubt that there are legal obstacles to fully open source drivers. However, I was mostly trying to use nVidia as an example of concrete, practical problems with closed source drivers.

With nVidia, I think that people like to blame them too much, but they do make a good example of the problems of closed source drivers on something like Linux.

On the other hand, maybe it is time we start petitioning SGI to allow an open soure driver for the nVidia cards :-)

Hey Guy, Tip for you

[Cocoshimmy]

AMD, Nvidia and Intel don't care if your old GeForce 4, RageXL or intel crap integrated 4 year old graphics cards don't work anymore and really, they shouldn't. That hardware was made several years ago, is no longer in production and I don't recall any of them offering a lifetime warranty on their products. By the same logic you should be upset that Sega isn't releasing any firmware updates or new games for Sega Dreamcast or that Nintendo has dropped support for the Gamecube. It simply does not make sense. The companies have moved on and have shareholders to answer to. Supporting 5 year old out of production products for users who are unlikely to upgrade soon anyways, does not make the company any money.

Further to this, they REALLY don't care if it works under linux. Complain and threaten to boycott them all you want, linux users make up what, 5 maybe 10 percent of all desktop users? Out of that lets say that each graphics vendor has an equal share of linux users so approximately 3.33% Out of that 3.33% only a small fraction, say 50% have bought a new graphics card in the past year (gross overestimate) that they aren't using for gaming (since only a few commercial 3d games are ported to linux every year as opposed to the hundreds made for windows) so it's almost certainly a low end and low margin card. This means that even if any one of these companies dropped linux completely, their sales would go down at MOST by 1.5% That is still an exaggeration since VGA mode works fine for the majority of linux users who use their systems primarily as servers and these are low margin parts.

Supporting linux is almost akin to charity for these companies, except without the tax benefits. Despite this, they continue to support something that makes them no money and is unlikely to make them money in the near future. Perhaps one day it may, but certainly not now.

Good (free) Linux device drivers book

[Anonymous Coward]

Please take a look at Linux Device Drivers, Third Edition. It's available for free but you of course will be purchasing it (you only mentioned it had to be free - not that you would be using the free version if there were multiple versions available).

I'll be expecting to see your name popping up on the Linux Kernel Mailing List in a few months announcing fixes and new features for my Intel, AMD/ATI, Via and NVIDIA graphics cards (some of which you will be reverse engineering). No need to thank me (you may also need to find a good book on X).

so that they don't abandon hardware

[Anonymous Coward]

when the closed fglx driver stops supporting 9600 series cards, you need to hold your upgrades to that machine until you buy a new card.

If it's an open source driver then it will be maintained by the X.Org team and when they drop it, you can take up support.

Out of your basement, enter the datacentre.

[jotaeleemeese]

In the datacentre you don't have screens connected to the machines. Since this is where most Linux machines are (fact) we will treat your anecdotal evidence as cute, misguided and uninformed.

stfu

[Anonymous Coward]

i hate amd, my gateway crashed because of the failure of the processor and ati graphics. gayyyy.