Dell Asking ATI For Better Linux Drivers

Open Source IT writes "According to a presentation at Ubuntu Live 2007, Dell is working on getting better ATI drivers for Linux for use in its Linux offerings. While it is not known whether the end product will end up as open source, with big businesses like Google and Dell now behind the push for better Linux graphics drivers, hopefully ATI will make the smart business decision and give customers what they want."

Comments

Better drivers?

[JohnnyBGod]

Let's hope they come. I'm not holding my breath, though.

Re:Better drivers?

[mabhatter654]

Dell has something the community doesn't have... sales orders for chips!! If Dell wants drivers they stand a good chance in getting them. They just started building AMD systems and bundling ATI chipsets is a key part of the sales pitch. Hopefully several hundred thousand computers will be enough to get the ball rolling!!!

Re:Better drivers?

[mrjb]

I'm not holding my breath Me neither. I recently switched from ATI (on which I spent several days to get it to work but *still* suboptimal) to NVidia to get accelerated dualscreen and it Just Works. Never looked back. Sorry ATI, you're too late.

Re:Better drivers?

[Gazzonyx]

I'm not holding my breath Me neither. I recently switched from ATI (on which I spent several days to get it to work but *still* suboptimal) to NVidia to get accelerated dualscreen and it Just Works. Never looked back. Sorry ATI, you're too late. Exact same here, I fought with an ATI X1050 PCI-E for 2 days before tossing it on the "I need any part I can find, right now!" shelf, and got a GeForce 8500GT just last week. This is running under Solaris Express Developer Edition.

The NVidia driver update was a single .bin that removed the old drivers, installed the new ones, and setup xorg.conf. It also moved the old xorg.conf to xorg.conf.bak, I was surprised to see that they did the Right Thing throughout the entire install. Fire and forget, reboot and move on to more pressing issues. These drivers were only a few days old, but they don't feel 'beta' at all; they feel very well tested.

I've been a long time ATI user except for a single Geforce4 back in my gaming days. So long, and thanks for all the fish, ATI.

Re:Better drivers?

[mrchaotica]

The NVidia driver update was a single .bin that removed the old drivers, installed the new ones, and setup xorg.conf. It also moved the old xorg.conf to xorg.conf.bak, I was surprised to see that they did the Right Thing throughout the entire install.

The Nvidia driver might be admirably well-behaved, but don't forget that the real "Right Thing" would be for it to be released under a Free Software license so that it can simply be distributed with X.org to begin with, like the nv driver.

Re:Better drivers?

[Gordonjcp]

Yup. This is exactly the point - *you* aren't an ATI customer just because you bought an ATI card. Dell *is* an ATI customer because they buy ATI chips by the million.

Where this does help you is that if ATI is well-supported in Linux, because Dell have requested better drivers, and then they see their card sales to end users go up, then other companies might get the idea. Look at how well having open drivers has worked for wifi card manufacturers; if someone asks what kind of card to get for their Linux laptop, you just tell them to get something by Ralink.

Re:Better drivers?

[piojo]

if someone asks what kind of card to get for their Linux laptop, you just tell them to get something by Ralink. This is off topic, but I feel the need to mention it anyway (could potentially save someone some grief): you still need to check compatibility lists, or buy it and be prepared to return it. I have a ralink card that is supposedly supported by linux, but I could not make it work even with about 3 kernel recompilations and following instructions. I could probably make it work by simply installing a much older kernel/distro, but I'm not willing to do that.

Re:Better drivers?

[Gazzonyx]

So true, friend. I guess I was speaking relatively when I said 'Right Thing'. :)

Re:Better drivers?

[pakar]

That sounds strange... I got a ralink card bundled with my asus mainboard and it worked out of the box (2.6 kernel) with the ra2x00 drivers. But maybe you are running one of the more dists that are a bit more restrictive on updates, like debian?

Btw... Not really using the wifi-card, but it's always fun to see all 60-70 neighbors running their networks unencrypted.. :)

Re:Better drivers?

[Schraegstrichpunkt]

I'm not holding my breath Me neither. I recently switched from ATI (on which I spent several days to get it to work but *still* suboptimal) to NVidia to get accelerated dualscreen and it Just Works.

... until you try to get it running on a paravirtualized Xen kernel (dom0 even). I tried and gave up.

I need complete, publicly-available interface specifications. Anything less holds me back.

Re:Better drivers?

[Gordonjcp]

But maybe you are running one of the more dists that are a bit more restrictive on updates, like debian?

Ralink is actually recommended for use with Debian, because the factory drivers are GPLed. They've been very supportive of the developer community, with things like free hardware for developers to hack on and publishing the docs for the chipset. Part of the reason for this is that they are late to market and haven't any particular technical innovation, so the best way to get market share is to push their "open-ness" and sell to the growing Free software market.

Nothing for you to see here...

[andrewd18]

Nothing for you to see here. Please move along.

Has there ever been anything to see in ATI's Linux drivers?

Re:Nothing for you to see here...

[shaitand]

'Has there ever been anything to see in ATI's Linux drivers?'

Or more to the point, has there ever been anything you could see WITH ATI's Linux drivers?

Re:Nothing for you to see here...

[TruePoindexter]

Sure there is....oh wait we're talking about display drivers. Yeah, no there is nothing to see. Or more to the point you can't see anything. Screens were always overrated anyway - ATI wants us to move back to binary displays.

Re:Nothing for you to see here...

[AaronW]

The only thing I see is garbled text with xemacs, a garbled cursor when moving between screens with Xinerama, Google Earth failing to start without adding a hack, and general slowness with my screen saver running maybe one frame per second.

Oh, I also see response emails from ATI that they don't support the Linux driver when I submit bugs using their bug submission package (which has all the support for selecting Linux).

I curse the ATI graphics card where I work daily.

Re:Nothing for you to see here...

[Calyth]

Yeah, my hair, on the keyboard, freshly torn out of my scalp, whne I was trying to configure their bloody driver.

I use to have a bit of respect when there was open source 3D accelerated drivers for some of the older Radeons, while nVidia had none, but right now, screw that. I just want the thing to work.

Re:Nothing for you to see here...

[ncc74656]

The only thing I see is garbled text with xemacs, a garbled cursor when moving between screens with Xinerama, Google Earth failing to start without adding a hack, and general slowness with my screen saver running maybe one frame per second.

The only issues I've run across are sluggish compositing (ran into this when experimenting with a Mac OS X-lookalike theme) and...hmm, I think that was it. While most of my computers are equipped with nVidia video, I have a workstation at work with a Radeon 9200 and a notebook that uses the Radeon Xpress 200M chipset. Getting accelerated video working on both of them has been no more difficult than getting accelerated video working on nVidia cards...it's just emerge ati-drivers instead of emerge nvidia-drivers, followed by some slightly different changes to /etc/X11/xorg.conf.

(Turns out that the workstation uses an open-source driver, but I know the notebook uses the ATI driver because the open-source driver doesn't support it.)

Re:Nothing for you to see here...

[deander2]

what's really ironic is back in the day (say mid 90s) ati was the ONLY option you had if you wanted X to work...

Re:Nothing for you to see here...

[ConceptJunkie]

Everyone complains (rightly so) about ATI's Linux drivers, but my last laptop had ATI graphics and it wouldn't even run 3D games right in Windows. I was never able to resolve the problem and I made darn sure my newest laptop has nVidia. I hated ATI back in the Windows 3 days, and it seems their drivers are still as awful as ever. This used to be a typical hardware manufacturers' attitude of thinking drivers aren't important: Most companies have been smart enough to realize how stupid that is.

Re:Nothing for you to see here...

[Keith_Beef]

My first two graphics cards were a Cirrus (Alpine) and a SiS (ViRGE).

Both worked with X11R6.

Beef

ATI Linux

[Anonymous Coward]

I must be the only person on Earth who hasn't had much problems with ATI's Linux drivers. Them dropping support for anything older than the 9600 series sucks, but I have been Thrilled with with the Linux performance of my Mobility Radeon x1600. Easily bests my Geforce 6800.

Beryl, XGL, Compiz, UT2003, Enemy Terrority, America's Army, all glass smooth and stable. I can run Beryl while playing high-def (1280p) x264 videos at the same time, too.

Still, better is better, and ATI's drivers do have some problems entering/leaving the console.

Re:ATI Linux

[Nimey]

How is parent a troll?

I've gotten the ATI drivers to install on my old Athlon XP box (9600XT), and Beryl worked for a while, but then after an update it didn't anymore and it stopped accelerating 3D. Nvidia's drivers Just Work, and so did the Intel 3D accel on my old laptop with 830 chipset.

Re:ATI Linux

[brunascle]

i had a problem with an old dell laptop (radeon 9500 or something) when i tried Beryl. the right half of the screen got screwed up. i shouldve disabled it when i had the chance, because eventually X crashed and i had to manually edit xorg.conf.

Re:ATI Linux

[Anonymous Coward]

It is kind of sad and shows the smug attitudes of some people on slashdot who moderate someone as troll based on their experiences being better than their counterparts.

Re:ATI Linux

[MrCoke]

Try the same thing on a dual-screen setup with Xinerama enabled.

Re:ATI Linux

[RAMMS EIN]

``Them dropping support for anything older than the 9600 series sucks''

Aren't thos supported by the open-source driver?

Re:ATI Linux

[stinerman]

I've got an X1600 that performs worse than the 9200 I used to have in here. Whenever I use ZSNES, MPlayer or any other programs that have a lot of motion for X to keep track of, the CPU usage for X goes off the charts. X barely keeps up when I'm watching a DVD! My system is getting old (Athlon XP 2400+, 1GB RAM), but this is ridiculous. I'm looking to buy a 9600XT and sell the piece of junk I have now. At least then I can use the free Xorg drivers rather than the crap ATI puts out.

I've checked with several people who have no clue what the problem is. I'm running Debian testing with fglrx 8.38.6. Yes, DRI is enabled and running. glxgears gives me ~900FPS.

Re:ATI Linux

[Anonymous Coward]

Parent here.

Can't say I ever tried Xinerama+beryl, but I have used multiple screen support. I don't use Dual Head often (use Dual Computer with Synergy instead), but when I do, I can drag playing video between the monitors in Linux without a problem.

The real problem with ATI's Linux drivers is that they are rather picky about your xorg.conf. I use Kubuntu, and I had to manually rip my xorg.conf to pieces and back to get it to the state it is in now. I can pastebin it later if anyone wants it. IIRC, I had to manually specify how much VRAM I have before 3d support worked at all. This could be a result of hypermemory, as I have 128meg of dedicated VRAM + 128meg of Hypermemory.

I'm not being a troll. When I bought my Geforce 6800 a couple of months ago, I was impressed at how easy it was to get working. nVidia's drivers are better, that's for sure, but ATI's aren't useless.

I think I'm going to create an account as "kevman" later, if its open.

Re:ATI Linux

[TheThiefMaster]

I had an X1900XTX in my pc, and just installing the drivers didn't even enable hardware opengl acceleration. Instead I had to manually edit xorg.conf to disable some other feature for it to enable. Movies decoded in the wrong colours. I had to manually switch gnome from aiglx to xgl to get beryl to run on it, and it then after a couple of minutes it frequently blacked out new windows (inc. menus) and frequently crashed.

Dunno which ati drivers you were using. Fortunately I was only borrowing the card, and switching to an nV 8800GTS was like a breath of fresh air.

Re:ATI Linux

[Benanov]

Not really. The 9200's are supported, but that's it.

Plus the 'open-source' 9200 driver is a rat's nest of firmware.

Re:ATI Linux

[Mad Merlin]

Your experiences with ATI are rather atypical. I have a Radeon X1400 that's half the speed of a Geforce 6600 in 3D apps, 114x times slower than a Geforce 6600 in 2D apps and 15x times slower than a Radeon 7500 with the open source drivers in 2D apps.

Re:ATI Linux

[BobPaul]

He's not so much a troll as flamebait. The AC's comment really isn't very helpful.
A) It's only marginally on topic.
B) It was an AC comment. These are invariably treated more harshly.
C) Most of his comment waxes ATI fanboy. This is given away by comments like "It easily beats my nVidia card."
D) He mentions a whole bunch of stuff that runs once GLX is working regardless of which vendor is providing GLX. Games and Beryl don't depend on nVidia, Intel, or ATI drivers present, they depend on the accelerated OpenGL interface provided by these drivers.
E) The comment starts out "Am I the only one having problems..." People running Beryl don't live in a box. They're running Beryl, more likely than not, with the help of information found on their chosen distribution's discussion forums. He's neither blind to the problems encountered nor the general sentiment. Knowing that lots of people are having difficulty and then simply stating "Works for me!" (which is all he did) is not interesting, insightful, or informative. It is, however, quite baiting.

Re:ATI Linux

[Anonymous Coward]

I do too. But you can't run composite (Beryl) and 3D hardware rendering (UT2k4) at the same time on 95%+ of ATI's cards on Linux.

Re:ATI Linux

[aperion]

900FPS? That sounds pretty low, perhaps the 3d acceleration is not working. Just because DRI is enabled doesn't mean the 3d drivers are working. with an nvidia 6600 GT I get more than 3000 fps (I don't remember the exact number, but it's very high) ~900FPS is what I'd get with out the 2d nvidia drivers(aka nv).

Re:ATI Linux

[j79zlr]

I had an X700 and it would only allow me to log into X once. If I logged out and tried to log back in, the machine would hard freeze. Bug 239. Acted the same way when switching VT's. Bug 37.

I bought a GeForce 7950GT when I built my latest PC and was so happy with its linux performance I bought a GeForce 6800 to go into my other box. I fought with that ATi card for 2 years and it never really worked right. It could just be certain chipsets/versions but I know I wasn't alone. The Radeon 9000 I got works just peachy with 3D acceleration from the open source drivers OTOH.

Re:ATI Linux

[Anonymous Coward]

Do you have subpixel rendering for font anti-aliasing enabled?

Re:ATI Linux

[BobPaul]

The real problem with ATI's Linux drivers is that they are rather picky about your xorg.conf. I use Kubuntu, and I had to manually rip my xorg.conf to pieces and back to get it to the state it is in now. But earlier you said

I must be the only person on Earth who hasn't had much problems with ATI's Linux drivers. It's certain you aren't trolling anymore and perhaps you didn't intend to troll originally, but these two comments are worlds apart... This is the difference between "STFU, Works for me!" and "It *works*, but you have to fuss with it a lot to get it going."

Re:ATI Linux

[a.d.trick]

My experience with a Mobility Radeon 9600/9700 has mostly been always "just worked" (except for some funny stuff under Gentoo, but I think that was a problem with the ebuilds and all sorts of Gentoo magic). That said, AIGLX doesn't work, which is a real bummer. XGL does not "just work" so that has been something of a pain point. That alone is enough to make me run for Nvidia cards. The performance on it is pretty stellar though.

I think it depends a lot on the card though. There are a large number of ATI cards that don't work for various reasons. And if it doesn't work, your pretty much out in the cold. Nvidia drivers aren't perfect either though, the 8 series have a bunch of annoying bugs with animations and such.

If ATIs drivers ever go open source though, I would dump Nvidia in a moments notice.

Re:ATI Linux

[cachimaster]

HP nx9420 here, ATI mobility 1600.
GLXGears: 2392 FPS, always.

Funny, but 3D seems to work fine with the new Ati drivers and Ubuntu x86.

But video (Mplayer, totem) works very bad. I have the Driver v8.38

Re:ATI Linux

[exi1ed0ne]

Them dropping support for anything older than the 9600 series sucks

I'll say. My Laptop has one of those chips, and my life has sucked since. It's not like I can pop in the latest and greatest card and be done. Trouble is that any vendor that ships binary drivers can drop you any time they want to and there isn't squat you can do about it. NVidia is just as guilty.

Re:ATI Linux

[mrchaotica]

I'm looking to buy a 9600XT and sell the piece of junk I have now.

Wait a second, let me get this straight: you're complaining about how your ATI card sucks (which is very likely because it is ATI, and has crappy drivers), yet you're planning to replace it with another ATi card? What are you, stupid? A glutton for punishment? Did Nvidia and Matrox gang up and steal your lunch money as a kid? Stop me if I guess it!

Re:ATI Linux

[Anonymous Coward]

im using beryl with a geforce6100 and its quite fast.

Re:ATI Linux

[Anonymous Coward]

How the fuck is it a troll to report that one hasn't had any problems?

Re:ATI Linux

[stinerman]

In order:

Yes.
No.
No.
No.

I bought the X1600 before I made the move to Ubuntu (and then Debian). The proprietary drivers (fglrx) are pure shit. The FOSS ones made by our friendly hackers that are released as part of Xorg are quite good. The 9600XT is an R3xx card and is about as cheap a card as you can get at that performance level.

The nVidia cards don't have any FOSS 3D support. I took a look at Matrox cards. They have FOSS drivers but they are more expensive and I can't afford them at the moment. Their 3D performance is nearly non-existent.

It comes down to a decent FOSS driver with the 9600XT @ $40 or a lesser Matrox card for $70-$80. nVidia isn't on the radar.

Re:ATI Linux

[stinerman]

I've posted on a few forums and no one has any clue what the deal is. I'm willing to believe some configuration is off, but I can't find the problem. I get no errors from /var/log/Xorg.0.log and only trivial warnings. If you have any idea what the problem could be, please contact me.

Re:ATI Linux

[Anonymous Coward]

In the art of trolling, it's not about the subject or content but how you write it. The key is to contribute nothing useful or something that will result in lots of angry or otherwise useless replies.

Re:ATI Linux

[Xtravar]

Well, except for the 'Beryl + NVidia = Black Window' bug. Have they fixed that yet? Last I heard, the ball was in NVidia's court

Re:ATI Linux

[BobPaul]

Fine, whatever... I was just answering his question. This time I won't suppress me karma bonus.

Re:ATI Linux

[sigaar]

Works fine at home (HP nx6125 - Radeon m200 with 1400x1050 LCD and external 1440x900 LCD, ATI proprietary driver) and at work (two 19" 1280x1024 LCDs on a Radeon 7000, old OSS driver). So there...

Re:ATI Linux

[Nimey]

Haven't had that problem. I do occasionally have the window decorations go on vacation, though, and take the effects with them. I expect that's a Beryl problem.

Re:ATI Linux

[jmorris42]

> Not really. The 9200's are supported, but that's it.

Eh? I'm running an ATI X850XT here on my testbench just fine with the stock X.org driver and I have the 3D bling turned on. Stellarium does have a habit of going kaboom unless I turn off the bling first.

Re:ATI Linux

[kevmatic]

kevman was taken :(

If you're right, then nearly every comment here is "fanboy" in the opposite direction. Everyone is like "lol ati drivers are completely useless" which simply isn't true. And THAT is sure helpful. Yeah, a lot of people have problems. But a lot of people don't. And arguing over the definition of troll and flamebait isn't helpful either. It is, in fact, very baiting. Which it seems everyone agrees with. There are more than enough trolls here that we don't have to read between each others lines.

Since I posted that, however, I discovered and fixed a slowdown my machine with the nVidia card (which I mentioned I recently bought. Fanboy, right) and gained 20-30%. Still slower than my ATI machine, but its the CPU at this point. I was pointing out things that work for me to establish a benchmark of my (impressive) performance. A machine more than a few years old won't play high def video at full speed at all, let alone within beryl's spinning cube.

ATI certainly has room for improvement. My ATI machine just gives me a scrambled image if I switch to a console, for instance. And the Xorg.conf takes a TON of work. But I'm confident that the correct xorg.conf can get just about everyone (Radeon 9550+) working. And the Opensource drivers are slower than dirt. My Radeon 9200 is completely useless in Linux.

Re:ATI Linux

[dabrepus]

I had this problem with my laptop a while ago. Seemed to be because X couldn't use xv to output video.

Nothing to do with 3D.

Is xv working on your computer? I think mplayer will tell you what output driver it's using.

Re:ATI Linux

[Magada]

You don't have acceleration at all. 900 fps is what the system manages to squeeze out of the proc. Play an mp3 and watch that figure drop. You have driver trouble.

Re:ATI Linux

[Caetel]

That's probably because the GeForce 6800 is a year older than the Mobility Radeon x1600.

GPL or nothing

[Werrismys]

Even if ATI released 100% working, fast drivers, they would be useless if they weren't OPEN and FREE.

Re:GPL or nothing

[apathy maybe]

What if they were 2 clause BSD licensed? Would that be good enough for you?

Though, I'm sure they would prefer to release it under something more like the GPL so that they can poach any changes back again (just like they are allowed too of course...).

Re:GPL or nothing

[Anonymous Coward]

While I don't agree with the "GPL or nothing" position on the level of principles, I often wonder why hardware vendors keep their source closed. It's not like a driver is of any value without the hardware and quality open source drivers would boost hardware sales. Maybe just a little, but still.
Just publish the source to whatever you have and see what the community makes of it. How could it possibly hurt?

Re:GPL or nothing

[RGRistroph]

Often, not as much is done in the hardware as is advertised. I have been told that examination of some modern graphics drivers reveals them to be very good implementions IN SOFTWARE of graphics libraries. If those companies were to release the source, their competitors drivers would gain in ability, and projects such as OpenGL might suddenly become a lot better.

Essentially, it is partly the case that graphics cards are hardware dongles for graphics libraries (drivers).

I would be nice if ATI released open source drivers, both for Linux and for Windows. However, none of the big graphics card manufacturers are likely to do that unless they believe that their own card can compete on a pure hardware basis alone. The fact that they don't do it, is evidence that these overpriced 3D watt-burning powerhogs aren't really all they are hyped up to be.

Re:GPL or nothing

[iminplaya]

They are probably trying to keep certain hardware specs out of sight to prevent others from building a competing product. However, I thought that what patents on the hardware are supposed to do. Eh...I use whatever I find in the dumpster. I feel so poverty stricken using an nVidia with only 32mb of RAM :-), but it seems to be sufficient for playing Solitaire.

Re:GPL or nothing

[Maxo-Texas]

Another faulty mod at slashdot...

Your opinion is that only open free code is of value.

However, in the real world, a lot of things won't get done unless someone is compensated in some way.

Over 45 years, I've seen that most people who declare things should be free do not contribute a bit of their time to help things be free. They want to be compensated for their time but they want to get everything free.

Given a choice between no driver and a closed driver that works and is installed as a binary object, I have to disagree with you.

Since ATI is a hardware company, I think they'd be foolish to ignore a rapidly growing portion of the market to sell their hardware too. However, if that market doesn't buy enough cards to pay the salary ($150k a year with benefits) of the driver developer then ATI is being extremely rational to blow them off. $150k would probably require a minimum of 20k card sales per year on linux boxes.

Re:GPL or nothing

[PalmKiller]

I would use them, as would most any other linux desktop user. Heck I have used the closed source nvidia drivers and it didn't make me ill or nothing (but lsmod did say my kernel was tainted as I recall...oooo I said, tha 'taint good at all...but at least I got good resolution on the desktop).

Re:GPL or nothing

[e2d2]

Speak for yourself mate, I run Ubuntu on a dell laptop with an ATI video card, I'd like some native driver support regardless of license. As my mother always said - beggars can't be choosers.

Re:GPL or nothing

[dctoastman]

If this is true, that there is some secret sauce in the drivers that make the card run sweet, then they should remain closed.

At least, according to ESR's point of view (as espoused in 'The Cathedral and the Bazaar'.

He outlined several reason why a company _shouldn't_ open source its code. I can't recall them all in detail. But, if the value of those secret bits are sufficiently high, then ATI benefits more from closed source than open source.

Re:GPL or nothing

[twilight30]

Like Nvidia?

Re:GPL or nothing

[jZnat]

I think they'd be useless until they were released under the X11 license like the rest of X.org is. It'd be nice if other platforms other than Linux can benefit from proper drivers for ATI cards as well.

Re:GPL or nothing

[MMC Monster]

If the software violates a software patent, it is unknown until the source is seen. Given the amount of time it takes to search for software patents, it's extremely reasonable for commercial entities to hide their source code.

Re:GPL or nothing

[DrXym]

Even if ATI released 100% working, fast drivers, they would be useless if they weren't OPEN and FREE.

And how do you come to that conclusion? Personally I really could not care less if my drivers were open or closed just as long as they exist. Open might be preferable, but closed is just fine by me.

And I expect the same holds for a great many people who expect decent graphics performance out of Linux but don't want to wait the month of sunday's for either NVidia and/or ATI to consider their IP no longer worth protecting.

Re:GPL or nothing

[RGRistroph]

I don't believe it is ESR's position that you should write Free code except when you can make a lot more money by writing un-Free code. He's a bit more of a capitalist and right-winger, but he's the kind of right-winger to whom Freedom is important.

ATI used to produce a full spec for their cards. We should encourage them to return to that practice, because then we will have better computers. It's that simple.

If they instead want to produce hardware dongles for proprietary code, given what we know about how a proprietary code computer industry evolves into crap, we should not buy their stuff and should even support projects such as the open graphics card ( http://wiki.opengraphics.org/tiki-index.php ) until ATI goes out of business and is run by someone who wants to sell us something we want to buy.

If, as a marketplace, we pay ATI to make more money for keeping secrets, than for producing good hardware, then we will get what we pay for -- bad hardware and closed drivers. It's a free market, and you should buy neither ATI now NVidia nor Intel if they aren't selling what you want; fortunately, since 3D acceleration is a trivial unnecessary decoration on a computer, and there are plenty of old ATI Radeon 9500s for sale on Ebay, you don't have to send your money to these guys if you don't want to.

Re:GPL or nothing

[DrXym]

That depends if the source code contains trade secrets, licenced 3rd party software, code shared with other platforms, patent pending code, or simply because NVidia / ATI don't feel like showing their source code off when they're in such cutthroat competition with each other.

I think it's totally unrealistic for an open source proponents to expect / demand source code from a vendor who has their own reasons for keeping their code hidden.

I think a far more realistic and pragmatic approach would be for dists and vendors to work on an dist neutral ABI to maintain drivers, but also to highlight the advantages of open source such as more optimal performance.

Re:GPL or nothing

[Rakishi]

The fact that they don't do it, is evidence that these overpriced 3D watt-burning powerhogs aren't really all they are hyped up to be. No it shows that some things can be done more cheaply in software than in hardware especially on something as complex as a modern video card. It's as a simple as that. It's like saying that the OS should be hard coded into the CPU as otherwise the CPU is worthless.

Re:GPL or nothing

[RGRistroph]

I don't know if it is better done in software or not, but it is plausible to me that it might be.

If which case, why are we paying so much for a video card ? Why don't we write a very good version of Mesa3D or something, so we can run $30 SiS cards and get the same performance as the $300 cards ?

Either way, whether the value is in the card or in the software, hiding things is not in the benefit of the consumer or the computer industry in general. If the hardware cards really are that good, let's see a full spec so we can write the drivers we want, or an open source driver from which was can derive a spec. If they aren't not that good, let's figure that out write drivers just as good that work on cards that will save use $50 a year in smaller power consumption costs.

Re:GPL or nothing

[darth_linux]

oops. You've made the classic mistake of confusing free with free. Remember, the term 'open and free' means open source and free as in freedom. closed-source binaries do not offer the community the freedom to adjust the driver (read hack - in a good way) to make it fit any related hardware not directly supported by the company. One of the first fires that lit up Richard Stallman was the closed-source printer driver necessary (but not provided) to make his older server play with the new printer.

You can and will get compensated for your work under a 'free as in freedom' open-source model. That's one of the reasons the GPL exists - to give authors credit for 'free as in freedom' work.

Re:GPL or nothing

[Anonymous Coward]

AyN RaND!!! AyN RaND!!! AyN RaND!!!

Re:GPL or nothing

[Anonymous Coward]

Yeah because we purchase the drivers and not the cards. I am pretty sure ATI is compensated when users/computer manufacturers order the hardware. Maybe it seems silly that people cry about it in the Linux world from a Windows user perspective because could you ever see ATI or Nvidia hijacking support from Windows drivers? I mean it's not like some shitty printer we are talking about where we can just go buy a better one from a list of hundreds. You have a few video card chipsets, and fewer that work well with 3D on Linux. Having said that, it isn't fear of not getting compensated that they are afraid of. It's encumbered IP they probably have in their driver. I wouldn't be surprised if they have some proprietary libraries or code that aren't explictly owned by them and it would be pulling teeth to convince those other guys to open their code as well.

Re:GPL or nothing

[mark-t]

The problem with giving away driver source code is that it inhibit a company from being able to recover its research and development costs effectively because of the likelihood of somebody else figuring out the hardware interface specs from the driver and reverse engineering a compatible product for a fraction of the cost (because figuring out a way to do something that somebody else has already done is a lot easier than inventing the idea in the first place), and it would price the first company's product out of the market before they've recovered their R&D costs.

Re:GPL or nothing

[dook43]

Not in most areas of the country. Especially in my area -- there are tons of MS .Net and ASP jobs, and the few LAMP jobs that are available are at small startups ill equipped to pay too much more than 30k/year. Joe Engineer doing kernel hacking at RedHat is the exception rather than the rule. For the everyday software developer (read: RAD Business/Web apps), OSS is not the way to earn a living.

Re:GPL or nothing

[Kjella]

Look, there are two types of hardware which I think could easily be released as open-source, for different reasons.
1) Everything in hardware, "stupid" interface with a trivial driver. Basicly any card that does all the interesting bits in hardware. There's essentially nothing to do and could easily be maintained by the OSS community.
2) Everything in software, "stupid" interface which relies almost entirely on the driver, release specs so others can try to emulate what the driver does, e.g. software RAID. Basicly, you have to it all yourself anyway.

Modern graphics card drivers are neither of the two. They're basicly a highly optimized hardware-assisted graphics library where it's of great importance where and how in the different algorithms it's assisted by hardware. Being able to study the interaction between hardware and driver in detail could easily give the competition vital information on how to design their hardware better or where to shave cycles off their software implementations. In a market where performance is vital that could easily lose them many millions in sales.

That is ignoring all the other possible pracical and legal problems which I'm sure someone will repeat to you for the n'th time. In a market with many competitors and stable technology like say RAID cards or gigabit NICs, I could easily see one try it anyway for the chance of a market grab. But ATI and nVidia are tightly locked in a duel and think the drivers are part of their crown jewels. Maybe you're right - or maybe they're right and would shoot themselves in the foot, but I don't see any of them risking it over the small potential upside.

Re:GPL or nothing

[jimicus]

Well, I can only speculate, but hardware companies frequently are lousy at software. There's a strong chance that some of the driver was outsourced, and they don't have the rights to open source it.

But that doesn't explain keeping the specifications closed. IME, that's a sign that the hardware is actually pretty sucky and it's only through minor miracles in the software that it works at all.

Re:GPL or nothing

[Anomalyst]

it would be pulling teeth to convince those other guys to open their code as well.

We don't WANT their crufty code, just accurate and detailed hardware specs and maybe a little email back and forth with a knowledgeable liason to elaborate the places where things are uncertain or missing. For the most part we aren't even seeking the latest and greatest features, a solid, performing and MAINTAINABLE driver is the goal.

Re:GPL or nothing

[schwaang]

You're saying that the Chinese (e.g.) couldn't disassemble a binary driver, and that's the only thing holding them back from developing a cheaper graphics card clone? I don't buy it at all.

Re:GPL or nothing

[Anonymous Coward]

Another faulty mod at slashdot...

Yep, whomever modded you up.

You are (deliberately?) mixing up free as in beer and free as in speech. ATI drivers are always free as in beer. You buy the graphics card or chip and get the drivers for free. And ATI will always be the ones to pay for the development of those drivers, because no one would purchase their hardware otherwise. None of that has anything to do with whether or not the drivers are free as in speech (open source.)

Over 45 years, I've seen that most people who declare things should be free do not contribute a bit of their time to help things be free. They want to be compensated for their time but they want to get everything free.

There are many people willing to contribute their time to the development of free (as in speech) drivers for ATI hardware. The problem is that ATI will not disclose the necessary details of the hardware.

Re:GPL or nothing

[Maxo-Texas]

Um. If the source code is available, I just gave up any ability to be further compensated.

Except for "thank you". I can't put food on the table with "Thank you".

Re:GPL or nothing

[Maxo-Texas]

The faulty mod was referring to making the original author I was responding to "flamebait."

He was expressing an opinion- it wasn't flamebait. We have a lot of crappy down-modding to censor people at Slashdot over the last 12 months. I disagree with down-modding opinions you dislike. "Troll" should be reserved for posts like that "gay nigger" guy who tried to first post everything. The grandparents post was on topic and a legitimate opinion.

I can see your point that if they would release the hardware specs then you could write your own drivers. As long as it is about you writing your own drivers from scratch, I'm for that. Asking them to pay someone to write a driver and then release the driver and the code is what I disagreed with.

Re:GPL or nothing

[immel]

My primary linux box has a FireGL T2. I've been through this level driver hell, and yet I am inclined to agree. The weekend after ATI releases the source code (which, when reverse engineered, may reveal in great detail the workings of the proprietary stuff in their chips that makes them go real fast), every chip manufacturer in China will be making unlicensed knockoffs of their chips (Every chip manufacturer in China is making mostly licensed knockoffs for now). A lot of R&D went into making the stuff that would be revealed in the source code.

You can't just expect a company like ATI/AMD to give up the secrets, garnered by expensive R&D, which comprise a great deal of their income just to please a (let's face it) very small fraction of their installed customer base. Sure, if it was released, the linux community would jump on the code and make a driver roughly a million times better than anything ATI could produce, but realistically I just don't see it happening any time soon for a very good reason.

Re:GPL or nothing

[Anonymous Coward]

Yes that would be a wonderful thing to happen as it would follow the OpenBSD goal of wireless drivers. Unfortunately no one really knows what they want. 90% of the posts on this board ask for their code and I am explaining why they won't open their shitty hack-assed code. Specs are always better but getting universal healthcare in the US is probably an easier goal to go after. ATI and others would love to have their products as big black tamper proof boxes.

Re:GPL or nothing

[davidsyes]

"Just publish the source to whatever you have and see what the community makes of it. How could it possibly hurt?"

Well, I suppose the shareholders and companies and wall street are afraid of loss of IP ownership... While I agree with your stance, I also presume those other parties will feel:

-- loss of control

-- loss of potential income streams

-- strengthening competition...

Re:GPL or nothing

[compro01]

newsflash, you're already giving away the driver, you're not making money off it. you're making money off the hardware.

Re:GPL or nothing

[mark-t]

Don't put words in my mouth. Did I say that others couldn't reverse engineer from a binary? No. I said that providing source code or specs only makes it unacceptably likely that cheaper compatible products would surface before they recovered their R&D costs. Whether you buy this or not, there is plenty of historical precedent that shows that companies that produce useful hardware and publish too many details about how they do things have their ideas copied. Anyways... reverse engineering from binaries is still multiple orders of magnitude more difficult than working from documented source or specifications, the latter being therefore proportionally more likely to result in the surfacing of cheap clones than the former... at least before the company is ready to lower its prices or discontinue the product (and bring out another new innovative product).

Re:GPL or nothing

[schwaang]

Whether you buy this or not, there is plenty of historical precedent that shows that companies that produce useful hardware and publish too many details about how they do things have their ideas copied.

It would be more persuasive if you could provide some relevant examples. Examples where an open-source driver have lead to cloning would be relevant.

But seriously, the specs needed to implement a driver are *far* from the detailed specs needed to clone hardware.
If you doubt that, just go read some open source drivers and see if you could design the hardware from them.

(PS - Your argument is remarkably combative and irrational for a Canadian.)

Re:GPL or nothing

[Rakishi]

*sigh* Think of a video card as an airplane and the software as the pilot. If your pilot is a moron even a jet plane won't do much. Likewise even if your pilot is god like he can't make a low end prop plane reach 600mph. The software is there to take the best advantage of the hardware which DOESN'T mean that the hardware is pointless. The software also covers up limitations of the hardware and bugs that would cost too much to fix in hardware. A car analogy is that cars don't self-diagnose most problems but rather require a human to do so even though the former is doable to a large degree. There are also patents and license agreements involved in there.

Re:GPL or nothing

[Anonymous Coward]

In an environment with very specialized, high-margin hardware with convoluted drivers and algorithms, (medical devices, MRI machines, etc) you might be right, but these are 3D cards.

I'm not belittling them. I'm saying the interface is set (DirectX and OpenGL), the specs are set (geometry/vertex/pixel shaders) and most of the algorithms are well understood. The only thing to do is implement them well, get to market first and then lower prices on a reasonable schedule while keeping a good margin.

- Any company that waits until ATI publishes an open source driver is not getting to market first with the new features.
- Any company that takes a slow start will not be able to undercut ATI who has proven hardware and, by then, savings from high-volume production and experience fixing everything wrong with the hardware.
- By the time the waiter gets hardware into production, a new generation of ATI hardware will be out.
- Any company that wants to improve the driver also releases improvements to ATI, who can then use their improvements as well.

In fact this last one is what makes the GPL interesting for these cases. By releasing their code under the GPL, they create a level playing field. Sure, you could read the code and reimplement their ideas in your own closed source but that's marginally less expensive than reverse engineering a binary.

In fact, the GPL creates just about the perfect playing field where ATI and NVidia could collaborate without fear of backstabbing. Release it under v3 and your competitor has to automatically cross-license their patents if they want to use your code. Maybe add a Tivoization exception, although it is an interesting question if that would be desired or not by either party (e.g. NVidia adding code that only runs under NVidia cards).

A GPL driver shared among companies decreases the development and support costs of the software side by spreading them among companies (not to mention users). The only problem is that it lowers the barrier to entry, but in this market that doesn't seem like a worry right now.

Re:GPL or nothing

[Anonymous Coward]

Your opinion is that only open free code is of value... in the real world, a lot of things won't get done unless someone is compensated in some way.

Come on, you should know that there is big money being made from open source as we speak. Compensation for software development is no longer dependent on whether or not the code is proprietary -- the more popular open source gets, the more that argument is reduced to a strawman.

I have no problem with proprietary code myself -- that's a decision that every software developer has the right to make -- I'm just pointing out the reality of the situation.

Re:GPL or nothing

[Deanalator]

Open and free like the GPL? The GPL is so full of red tape an legal hassles that it is almost worse than the nvidia or ati licenses. I publish under the "fair license". It's simple and has been tested by the US legal system. The funny thing is that since the fair license is more "free" than the GPL, people writing GPL apps can use my code, but GPL code is totally worthless to me because my apps aren't "GPL compatible". We will never progress in a world where people have a hard time advancing knowledge just because they are bound by legal hassles.

As for mixing computers and religion, I think you would be better off finding Jesus.

Re:GPL or nothing

[heinousjay]

It's an interesting sort of dichotomy - people who are gung-ho on the open source typically have closed minds. Strange, that.

Re:GPL or nothing

[Jherek Carnelian]

However, in the real world, a lot of things won't get done unless someone is compensated in some way.
...
Over 45 years, I've seen that most people who declare things should be free do not contribute a bit of their time to help things be free. They want to be compensated for their time but they want to get everything free. Gee, just what about the CURRENT drivers is not little-f free already?

You've misconstrued the OP's point in an entirely non-sensical fashion.

The FREE the OP was talking about wasn't the free-for-no-money kind of free, because we already have that with the current drivers. It was the free-as-in-freedom kind of free. Nothing about freedom requires that people go without compensation.

Re:GPL or nothing

[Virgil Tibbs]

does ati make any money from drivers?
  - as you have said they have to PAY developers money to develop software they then let anybody download
  - search ati drivers and you will find millions of hits..
To produce a binary driver for linux, with its relativly small market share, would probably be considered not cost efficient
hence it's a much cleverer idea to give the people who use it the resources and say heres the source, make it work, take the credit and BUY or products...

Re:GPL or nothing

[the_greywolf]

The problem with giving away driver source code is that it inhibit a company from being able to recover its research and development costs effectively because of the likelihood of somebody else figuring out the hardware interface specs from the driver and reverse engineering a compatible product for a fraction of the cost (because figuring out a way to do something that somebody else has already done is a lot easier than inventing the idea in the first place), and it would price the first company's product out of the market before they've recovered their R&D costs.

I have trouble parsing long run-ons, but I think I figured out what you're saying. And I really don't see your point.

Knowing how a driver interfaces with a chip does not make it easier to understand how a chip works. The only way to understand that is to have a deep knowledge of the silicon.

Example: While Intel used the 80x86 mnemonic for their commodity CPUs, there was no legal means by which Intel could prevent the duplication of those chips in their primary market. (In fact, when AMD began producing AM386 and AM486 chips that were pin-compatible with Intel's 80386 and 80486 chips, Intel sued AMD and lost, since this is common practice throughout the semiconductor industry. For example, the 74xx and its low-power variants - 74Cxx, 74HCxx, etc. - has been manufactured by everyone from National Semiconductor to Motorola, where the competition for price has been tradtitionally very good.) Thus, after Intel failed to prevent AMD and Cyrix from producing clones of their chips, they tademarked the "name" of the chip, calling it the Pentium. IIRC, AMD licensed the x86 ISA to continue producing their AM586 chips. When their license terminated, they began developing a "80x86-compatible ISA" that made it possible to produce the K6, and later the Athlon and friends. They have invested billions in R&D to produce a chip that is ISA-compatible, but not pin-compatible with Intel's offerings. Cyrix (now ViA) and Transmeta did the same.

Simply because one is "intimately familiar" with the x86 instruction set because it is well-documented does not mean it is therefore "easy" to produce such chips. They contain an enormous amount of complexity at the logic level, just to interpret a 0x90 byte as "no-op." Previously, any company wishing to produce an electrically-compatible variant of another company's offering would crack open the plastic case and analyze the silicon (if it was complex, as in the case of the 80x86) or simply work their design around the electrical properties of the package (as in the case of the 74xx), especially since the components are well-understood: all engineers are familiar with the operation of a transistor.

Thus, I don't see how your argument applies. If a Chinese company wishes to duplucate NVIDIA's GPUs, they would either crack open and analyze the silicon or conduct industrial espionage (i.e., raid TSMC). However, knowing what commands mean when sent over a PCI bus to the chip in question is a completely different aspect of the problem. It's truly like saying you know how to build a car because you assembled a steering wheel.

Re:GPL or nothing

[darth_linux]

that assumes the purchaser can do what you did or pay someone else to work with your code. in the case they pay someone else, under gpl you still get credit (ok - credit != pay)

Re:GPL or nothing

[Maxo-Texas]

No. I think immel got to the root of the problem best.

This is a combined hardware/software package.

The hardware without the driver is worthless therefore the driver has value.

Given complete specs of the hardware, it becomes possible to make a knockoff.

Re:GPL or nothing

[WNight]

People already can disassemble video drivers. In fact, other video card companies are quite likely to employ someone who can already do this.

If nVidia was, let's say, 50% slower on dual-textured polys and couldn't figure out why, they'd have no trouble looking at the relevant sections of ATI's driver.

It's not even that they don't have the rights to all the code. They'd just open the parts they had rights and provide specs for the rest, if that was it...

The most likely reason is that their code is horrible. ATIs drivers have always been subpar and now we want to *see* their code!? Also, probably realizing that they must infringe on a ton on submarine software patents and thinking that letting patent-trolls read your source code is a good way for them to decide to sue you.

Re:GPL or nothing

[WNight]

Why is that too much to ask? I've written a few hardware companies over the years expecting them to improve their drivers because of my corner case configurations.

IMHO, based on years of QA, it's *much* harder to get a product working properly under Windows than Linux. "They" make so much of all the versions of Linux, as if the differences between Suse and Debian are half as large as that between versions of Windows. (98|NT) -> 2k was a difficult time for hardware companies. The company I was at had to support a million oddball things, many different products or windows actions would hose product and we had to recover from registry munging, files being overwritten, poor documentation on parallel port driver requirements, and a million things. If it were open source we'd have just made it work in a few default configurations and let the users support the weird crap themselves.

If ATI/etc want to keep selling products they should make ones that work. If one of the requirements is that it be open source so it can be properly maintained, then so be it.

I'm not expecting something for nothing. I'll buy the hardware, if they manage to make it work with my OS. If not I'll send them nasty letters for failing to make it work even where they attempted to.

Re:GPL or nothing

[WNight]

If there was a question nVidia had about ATI's drivers they'd simply disassemble them. They'd have a really good idea about what all was going on so it'd probably be pretty easy even.

Closed source prevents no bugs or leaks, just fixes.

Re:GPL or nothing

[WNight]

The hardware people have to pay for. The driver they give away but nobody wants.

Really, there is no value in an ATI driver. Even to them. For years I've recommended against buying ATI at my work because of the support nightmare.

As far as knockoffs go, do you think someone's going to try for an ATI-like card through reverse engineering crap when they could just implement the DirectX spec directly? They'll go snooping if ATI does something better, but they do that now without source.

Re:GPL or nothing

[TheRaven64]

2-clause BSD? Do you mean MIT, also known as X11 licensed?

For reference, the Intel drivers are MIT licensed for the DRI component (most of it), GPL'd for the Linux kernel module and 3-clause BSD licensed for the FreeBSD kernel module.

Releasing a driver under the GPL pretty much limits it to Linux-only. No one is going to be able to port it to BSD, Solaris, or any other platform. Some of the more obscure (non-UNIX) free operating system projects have started using DRI-derived drivers, and it would be a bit silly to release a driver that would explicitly prevent other operating systems supporting your hardware.

Re:GPL or nothing

[apathy maybe]

Nope, I mean the two clause BSD licence. Of course it is basically the same as the MIT licence, but is different from three clause BSD and of course the four clause BSD. The Free BSD licence is two clause as well.

Of course, if an OS has a decent driver system, then having a GPL driver shouldn't limit anyone from using it. As well (an I am not an expert in the field at all...), if people can write proprietary drivers using DRI, why can't people use GPLed DRI drivers on software such as FreeBSD? Or do you just mean that people won't distribute it?

And of course, it wasn't me saying "GPL or nothing".

Re:GPL or nothing

[TheRaven64]

I don't know why the grandparent suggested GPL for video drivers. Most open source video drivers are not Linux drivers, they are DRI drivers. DRI inherits the same license as the X server, the MIT license. Possibly the confusion comes from the fact that every DRI driver requires a kernel-specific DRM driver. These tend to be written with the same license as the rest of the kernel; GPL for Linux, BSD for FreeBSD, etc.

Re:GPL or nothing

[Maxo-Texas]

So what I am hearing is folks would be happy if ATI would make a linux only card that they published full accurate interface specs for and provided a skeletal driver (40 to 120 hours of coding max but functional) with source code which the community would finish and implement all the features that minimal driver did not implement.

To me that seems reasonable. The only thing I disagreed with was expecting the company to pay someone six months salary to develop code and then give it away. The company would not waste any effort on developing the software perhaps 40 hours of one coder.

I'm not sure the community would deliver but it seems like the cost to ATI would be minor and would likely be covered by the additional hardware sales.

I think as the hardware gets more stable this kind of thing may happen anyway. Computers are not changing as fast as they used to.

Re:GPL or nothing

[Taxman415a]

That could well be part of it--I don't know about graphics hardware specifically. But If you read what driver writers say, a big reason many companies don't want to release specs or opensource drivers is because their hardware is full of bugs and they don't want everyone to know how crappy it is. Add that to some hardware vendor's belief that the open sourcing of the driver will give away even small bits of competitive edge and you have the situation that we do where a lot of hardware doesn't have gpl or bsd drivers. What I can't understand is why they don't release specs for older hardware that's not on the competitive edge. I suspect it's because of the buggy hardware reason I gave above since that's the only reason I can see left.

Re:GPL or nothing

[Anonymous Coward]

You're underestimating the cost of designing and building new hardware.

This may help a lot

[Bullfish]

If you consider that AMD owns ATI and that AMD needs (considering how they have been hemorrhaging money) Dell to buy their CPUs, Dell just may be able to get what the Linux community has been asking wanted for quite a while.

While Dell doesn't have a lot of fans on Slashdot, they may also be able to get a lot more hardware supported as well.

Strange bedfellows, but...

Re:This may help a lot

[jd]

AMD also needs the One Laptop project to buy chips - they're potentially a gigantic customer, far bigger than any domestic slice of the pie Dell might have - and OLPC is also Open Source. I don't know exactly what graphics OLPC uses in the current version for graphics, but you can bet your left sock that AMD is going to want to supply as large a fraction of the components as possible.

Dell doesn't have fans on Slashdot, but I doubt they're doing that badly in Slashdot League Division 1, and they seem to be improving on the Good Idea/Bad Idea ratio. Of course, that could just be a phase they're going through too.

Re:This may help a lot

[nine-times]

My immediate thought when I read the headline was, "And this is the answer to all those people who asked, 'Why should I care if Dell sells computers with Ubuntu?"

I remember when that story broke, and loads of people were saying, "I use Linux, but I'm not going to buy a Dell," or "Well I don't use Ubuntu, I use [insert distro here], so this doesn't help me!"

Re:This may help a lot

[nahdude812]

Absolutely, this is a major player in this market, with a lot of buying power and therefore a lot of weight to throw around. It would just be incredibly costly now for AMD to not improve the quality of their Linux drivers!

Re:This may help a lot

[zCyl]

I remember when that story broke, and loads of people were saying, "I use Linux, but I'm not going to buy a Dell,"

Personally, I swore off Dell laptops a long time ago. But their support of Linux is what has changed my mind. For my next laptop purchase I intend to seriously consider Dell. Having the hardware pre-tested and known to function is a nice advantage which can save a lot of time in the buying process (and potentially even more time during the installation process if I don't have to fuss with tricky hardware).

Re:This may help a lot

[rbanffy]

From http://wiki.laptop.org/go/Hardware_specification:

    Beta Test 3 Systems (BTest-3, or B3)

This build, scheduled for May 2007, is the first to use an updated design for the laptop. Noticeable improvements over BTest-2 include:

        * A faster, lower power processor: the Geode LX-700
                    o 64 KB I/64 KB D of L1 Cache, 128 KB of L2 Cache (vs. 32 KB of L1 cache)
                    o Faster processor and memory clock (433/333 vs. 366/266)
                    o 1.5 W typ. vs. 3 W typ.
                    o Much better graphics processor, including support for rotated blits and depth conversion

Re:This may help a lot

[Anonymous Coward]

I purchased my first Dell, hell, my first prebuilt, after their Linux announcement.

Acer Laptops and Linux

[Calmiche]

Man, I hope this shows up. ATI Drivers for my x700 mobility in my laptop are one of the final problems I have with Linux. I use it at home a lot, but there just doesn't seem to be a reliable way of getting 3D graphics working with this chipset. So, I'm still dual-booting on my laptop. It's a pain, as honestly I'd rather go full linux.

For those wondering, I'm using Ubuntu on a Acer Travelmate 4400, and yes, I've one through EVERY walkthrough for 3d Graphics. Everything else works. Graphics, wireless internet, even my card reader. It's just 3D accelerated graphics.

Re:Acer Laptops and Linux

[Lord of Hyphens]

Try an old Thinkpad a22m with the P/M Mobility chip in it. Then you'll start screaming.

I got it working once. Once.

Re:Acer Laptops and Linux

[gnuman99]

Man, I hope this shows up. ATI Drivers for my x700 mobility in my laptop are one of the final problems I have with Linux.

Err... shouldn't that be,

"ATI Drivers for my x700 mobility in my laptop are one of the final problems I have with ATI graphics hardware". It is not Linux that is preventing you from using your display to maximum potential.

Re:Acer Laptops and Linux

[wild_berry]

I have a Mobility X700 in my notebook and use the X.Org 'radeon' driver without hiccups; I have used ATI's FGLRX driver also without problem, having followed the ATI driver walkthough at wiki.ubuntu.com. I believe it's become even easier with the new Resstricted Drivers Manager in Ubuntu Feisty and Gutsy.

Re:Acer Laptops and Linux

[Calmiche]

Oh, I can get the display drivers working. Both the initial drivers (Which I understand to be the same drivers as the x.org ones) and the Restricted Drivers work fine for desktop use. (After some xorg.conf tweaking) It's when I try to play games that I have problems. There is NO hardware acceleration. It's just like running software acceleration. In Unreal Tournament, for example, I get about 4 frames per second. This is true for all benchmarking software and includes Wine and Cedega.

Ok

[Billosaur]

Not much to the article. I mean it's a given that Dell would want better drivers -- no one's going to buy a PC that they can't hook up to their favorite monitor and use right out of the box.

ATI just released new drivers

[jshriverWVU]

here just a couple days ago. Not sure how much better they are, but they are making some efforts.

Re:ATI just released new drivers

[digitalderbs]

Most of their recent releases will either fix/patch bugs, support newer cards, break previous features or increase/decrease performance -- including this one. For example, version 8.31.5 broke suspend and cpu usage for video has increased in recent versions. Support for AIGLX still hasn't been included.

I really hope this pushes ATI to push feature development and improve their QA on the drivers so that performance and features are maintained between releases.

Re:ATI just released new drivers

[stinerman]

Yeah, I still haven't figured that one out yet. I'm ready to just sell the damn thing and get something with an R3xx core.

Re:ATI just released new drivers

[timeOday]

ATI has completely dropped support for the Radeon Mobility in my T40 Thinkpad. It sucks, I don't need to play the latest 3d games but I would like to use google earth and a few other OpenGL apps. And since the drivers only work with certain versions of X, I can't use an old driver unless I want to set my whole system back to RHEL3 or something like that.

Re:ATI just released new drivers

[DF5JT]

> ATI has completely dropped support for the Radeon Mobility in my T40 Thinkpad.

So what? The Radeon Mobility with an R250 core is perfectly supported by the original xorg-driver "Radeon", including 3D accelaration. My old R51 with an Radeon 9200 could do GoogleEarth and Quake3 without problems.

However, my new laptop took me weeks to find, because I have sworn to NEVER again use any notebook with an ATI-card - and Thinkpads use these in most of their models. Finally found an SXGA+ T60 with Intel Chip.

No more hassle with either NVIDIA or ATI.

Thank god.

Re:ATI just released new drivers

[timeOday]

Thanks for the tip, I'll try the OSS Radeon driver. I remember having troubles with it years ago, but it's time for another look.

Re:ATI just released new drivers

[HobophobE]

Just for the record they released that driver last week too... but they had a bug that caused the beta test watermark to be stuck on the screen if you used several of the common installation methods. To those of us who follow their drivers patiently hoping for ones that will work, it was a candid moment where you say to yourself, "this is a sick joke. AMD must be outsourcing their driver development to Microsoft."

Not to deride their efforts, but it gets frustrating when month after month crawls by and they offer very little in the way of better drivers. And then they can't even be bothered to have VM setups so they can run through the various installs before uploading to make sure there's not something crazy like a watermark about to be imposed on their customers.

DAMMIT err I mean ATI

[Coniagas]

Since ATI has been swallowed by AMD, ATI has gone downhil. Paper launches of products and mssed shp dates Wha makes you thin ATI will make an effort or do the even have the means to do do it?

Joy in getting what you want.

[Anonymous Coward]

"While it is not known whether the end product will end up as open source, with big businesses like Google and Dell now behind the push for better Linux graphics drivers, hopefully ATI will make the smart business decision and give customers what they want.""

Free blow jobs?

Great news.

[lattyware]

The only reason I have ever bought nVidia exclusively is the drivers. If ATI perked up with something better than TwinView (give me multicard support so I can have 3 monitors without xinerama!) and I'll switch. Besides, I want to get in on that AMD/ATI action if AMD ever get back on top.

(Not my current machine is a C2D E6600 with a GeForce 8600GTS and 7600GS - So no bias here.

Besides, any extra competition is always good. If it forces nVidia to push more improvement, bring it on.

Re:Great news.

[spiedrazer]

Agree... I just built a new PC with an ATI card and the DVI port basically didn't work with XP and the management software they provided wouldn't even open. After reading other posts on the Internet with the same results I put in an NVidia the next day and was up and running with NO issues. ATI needs to get their software act together!

Its a trap!

[Anonymous Coward]

I hear the plans to port the Catalyst Control Center is actually a devious plot by MS to destroy Linux on the desktop.

Id be more hopeful if the ATI tray tools guy was doing Linux.

Oh yeah. Completely.

[Petersko]

"Even if ATI released 100% working, fast drivers, they would be useless if they weren't OPEN and FREE"

Totally. Unless, well... unless you want to some stuff that requires working, fast drivers. In that unlikely circumstance the drivers would be very useful.

When it comes to closed systems like video cards and their drivers, I think only a fool would turn up his nose at a binary simply because it doesn't come with source code. They should, of course, provide it for any GPL'd libraries they use.

Re:Oh yeah. Completely.

[serviscope_minor]

I think only a fool would turn up his nose at a binary simply because it doesn't come with source code.

Nice attempt to dismiss arguments using invective. Perhaps you should read what Theo de Raadt has to say about the security implications of binary blobs.

When it comes to Theo, you might think he's an asshole, but if that's the case, then he's an assole who knows his shit.

Re:Oh yeah. Completely.

[Curien]

I'll remember not to install close-source 3D-acceleration drivers on my servers. Thanks for the hint.

Re:Oh yeah. Completely.

[Petersko]

"Nice attempt to dismiss arguments using invective."

Actually I didn't notice I had done so. My apologies.

In my defense it slipped out because "fool" seemed like a description rather than an insult. People who need security above performance can use existing open drivers. Slow, but secure. People who require performance are more likely to be gamers or artists - but probably gamers. For them using a binary from a manufacturer is probably not verboten, or even a bad idea.

And somewhere in the middle is the guy who wants performance, hates binaries, and has to choose between his technology-based morality and his desire to make use of his fancy new hardware.

And dismissing a binary simply because it's a binary, without even considering where the best option lies, seems like a fool to me.

Re:Oh yeah. Completely.

[somersault]

Yeah, nobody should EVER use a binary that they haven't written from scratch themself - they should start off writing up an assembler in binary, then write a decent c compiler in assembly. Then, they should write all their own libraries, and they can start on writing their own OS (or at least thoroughly inspect every single line of source in the OS that they are choosing to compile). In fact, that's not acceptable.. they should really design their own computer first to avoid the security implications of using buggy x86 processors..

While it's obviously not as transparent as open source, you could always disassemble it if you're really bothered? Or you could just say thanks for actually writing this in the first place, and then use it to drive your display. They have no obligation to provide source, especially when they are using 3rd party licensed technologies...

Re:Oh yeah. Completely.

[twilight30]

Agreed.

Sooner or later I just want the damn thing to work. The point the reply poster makes concerning security as per De Raadt is a valid one, but the other reply about being able to recompile and tune doesn't apply as easily with Nvidia drivers (hell, I recompiled them just last night with a 2.6.22.1 kernel).

Now if I can get Nvidia and Hauppauge working together properly, I'll be a happy man...

Re:Oh yeah. Completely.

[richlv]

And somewhere in the middle is the guy who wants performance, hates binaries, and has to choose between his technology-based morality and his desire to make use of his fancy new hardware.

hey. sounds like me ;)
but to be more precise, i don't hate binaries, i hate problems that come with closed source software - which is mostly drivers these days.

i am using nvidia driver on my box, but a complete opensource driver would be very nice. actually, i wouldn't have any problems moving to ati (that i dislike and avoid a lot with now) if they released an opensource driver.

problems with proprietary drivers include separate upgrades of kernel/driver (can lead to unusable xorg), no support for older cards (nvidia recently dropped support for one of my oldest cards, but the last working version has buggy headers...), worse quality in some cases (check out supending problems)...

Re:Oh yeah. Completely.

[akumria]

When it comes to Theo, you might think he's an asshole, but if that's the case, then he's an assole who knows his shit.

One would hope every arsehole was aware of their shit.

Re:Oh yeah. Completely.

[exi1ed0ne]

I think only a fool would turn up his nose at a binary simply because it doesn't come with source code.

Yeah, right up to the day that ATI decides to remove your laptop's chipset from their binary driver. I really appreciate them forcing a total system upgrade on me for an otherwise perfectly serviceable laptop.

Re:Oh yeah. Completely.

[Anonymous Coward]

Slow, but secure.

Are you certain? Have you personally done a code audit? Do you know the person who did the code audit? Or was it just something you believe because you heard it on the Net or heard somebody you don't know bragging about it on the Net.

If you don't do the code audit yourself you're just counting on somebody else's word. Which is pretty much the same as having closed source drivers anyway.

Re:Oh yeah. Completely.

[cortana]

Do you administrate these servers from the clients that you install these drivers on? :)

Re:Oh yeah. Completely.

[un1xl0ser]

How is this troll insightful? :-P

The idea is that you have access to the source of the application if you want to review it. You don't have to look at it, ever ... but it is there if you want to. Without source code, you have to trust the company that wrote the software. I don't know that many companies that I would trust the software from, personally.

They have no obligation to provide the source, just like many people don't have an obligation to buy their hardware. In fact, because the consensus seems to be that nVidia's drives are better than ATI's, a lot of people are doing just that. Having their drivers be open source gives them a very big competitive edge, they just don't know it yet.

Re:Oh yeah. Completely.

[Schraegstrichpunkt]

The idea is that you have access to the source of the application if you want to review it. You don't have to look at it, ever ... but it is there if you want to.

More importantly, it creates a competitive market if you want to pay someone to review/improve/port it. Dell wouldn't have to plead with ATI for better drivers if the cards' interface was publicly documented---Dell could just hire whomever the company wanted to make the drivers better.

Re:Oh yeah. Completely.

[somersault]

People will still notice if the drivers contain spyware that is trying to send information back to ATI, and even other more nasty problems. It is easier to check for this stuff if the source is open, but when you are a company like ATI, that pretty much relies on geeks as early adopters to drive new hardware sales, you can't really afford to piss them off by having malicious or invasive code in your drivers, can you? I mean it's not the same as with Sony selling CDs to the great non-unwashed masses.

Re:Oh yeah. Completely.

[TheRaven64]

Excellent point. There are companies out there who have the expertise to write display drivers. Until they hired Keith Packard and Eric Anholt, Intel used to outsource their driver development, and I think Matrox did the same thing.

It used to be that graphics chip makers would release reference drivers, and provide the source to board makers who would be able to improve them if they wished. For the VooDoo 2, there was a constant debate over which board manufacturer's driver had the best performance or the most bug fixes. An open source driver is just an extension of this; instead of limiting the people who can make improvements to board makers, anyone with a vested interest in better drivers can. Want to sell systems with ATi cards? Pay someone to tweak the drivers. Want to deploy 10,000 machines with ATi cards across your organisation? Pay someone to test all of the applications you use in-house with them, and fix any bugs or performance issues they come across. ATi drivers don't work as well with your product (e.g. CAD tool, game engine, etc) as they do with your competitor's? Pay someone to profile them and optimise some of the code paths you use. None of these are possible without an open driver.

I guess I'm a pragmatist.

[i_love_unix]

If the drivers do what I want them to do (i.e. not suck), I will use them, GPL or no GPL.

Re:I guess I'm a pragmatist.

[Jherek Carnelian]

If the drivers do what I want them to do (i.e. not suck), I will use them, GPL or no GPL. And if they do suck, now you've got a $100+ piece of junk on your hands that you can't do anything with except run MS Windows.

A good friend of mine encountered a bug in the nvidia drivers that prevented his dual-$300 card system from even displaying. Nvidia's *informal* support (they do not have a formal support system for linux - they want you to post on some web-forum) just blew off the problem report once they established that it was beyond the "did you plug it in level."

My friend figured out what was wrong, and with source code he could have fixed it, probably with less than 5 lines of code. But he could not do a damn thing because the the drivers where not open source.

Ironically, this is exactly the same situation that RMS found himself in all those years ago - he had a printer with a buggy driver and no source code. He knew what was wrong, could have probably fixed it with minimal effort, but without the source the printer was just an expensive hunk of junk.

Re:I guess I'm a pragmatist.

[Anonymous Coward]

Aside from the fact binary drivers can't be distributed with the distribution, so you have to download them separately and may (read: often) break your system on update, yeah, binary drivers are totally practical.

They are completely more convenient than drivers that are included in X.Org and automatically loaded on install and updated with the rest of the distribution updates.

Re:I guess I'm a pragmatist.

[TheRaven64]

You are not a pragmatist. People describe themselves as pragmatists in this debate to try to phrase it in the form 'pragmatist versus idealist,' because we all think we'd rather be pragmatists and sneer at idealists. This misses the point. Most people described as idealists in the Free Software community hold their beliefs for entirely pragmatic reasons. They want stuff to work, and long experience has shown that Free Software is the only way of guaranteeing this.

The debate is not between pragmatists and idealists, it's between long term and short term thinkers. You are on the side of the short term. You want something that more-or-less works now, and it is not important to you that:

• The code you are running in ring 0 is not properly security audited, so you may be installing remotely exploitable code (see the nVidia bug).
• The code can only be updated by the original manufacturer, who have a vested interest in stopping supporting old hardware to encourage you to upgrade (see the support being dropped for old ATi cards in the new drivers), and reduce the resale value of your old hardware.
• Bugs, if found, can only be fixed by a single source, so no matter how important a particular bug is to you, nothing you do can get it fixed if it's not a priority for the card maker (see the nVidia / Beryl black window issue).
• Your choice of operating system is dictated by the card manufacturer. If you want to switch to something else, you'll need to reverse engineer the driver to get it supported (see ATi support on FreeBSD, and compare with Intel).
• Your driver is tied to a particular version of the kernel. If the kernel interfaces change, and the manufacturer doesn't think your card is important, then you will have to stay with an old kernel, or lose support for the card.

Please stop smugly proclaiming yourself a pragmatist. Pragmatic sounds better than calling yourself short-sighted, but we have enough newspeak from politicians, without adding to it.

Endless hand wringing

[tji]

Why does the never ending cycle of

    ATI Needs to produce better Linux drivers --> ATI announces they really like Linux --> ATI never produces drivers

keep exciting everyone enough to cause this constant hand wringing?

They are not going to ever really support Linux well. If that's not clear after 12 years of the above cycle, then you haven't been paying attention. Move on.. Get a board with an Intel integrated GPU if you want totally open. Get an Nvidia card if you don't care about open, but want working accelerated drivers.

If ATI does somehow produce open specs or drivers, great.. think about buying one then. In the mean time, vote with your dollars, buy something else.

Re:Endless hand wringing

[Anonymous Coward]

But that conflicts with my boycott of Intel processors. Intel is not a friend of Freedom either, just a tiny bit more competent and a lot bigger.

Re:Endless hand wringing

[RAMMS EIN]

Correct me if I'm wrong, but wasn't ATI the graphics company that used to provide specs for their cards? Now that is the real way to get your cards supported (much better than providing drivers for a few platforms). While I can't comment on the current ATI driver situation (I've never had problems with ATI cards, but I currently have only Intel and nVidia), the picture of a company that has forever paid lip service to Linux without ever delivering doesn't seem fair to me.

Re:Endless hand wringing

[GeckoX]

Because this time it's a big gorilla throwing their weight around, and thus it just might be enough to persuade ATI to actually produce for once. ATI has always seen Linux support as something that wouldn't make them any money...but DELL certainly DOES make them money.

If anyone can get ATI to pony up working drivers for linux, it's DELL.

However, I'm still waiting for the fat lady to sing on this one ;)

Re:Endless hand wringing

[TopSpin]

They are not going to ever really support Linux well. If you are AMD/ATI and the 800 lb gorilla of PeeCees sends you a memo, you read it and take steps or you answer to The Board because one wrong word from those people and your Ass is Grass. ATI is under new management and now Dell has a Linux agenda. Have a little faith. Things can change. Companies like HP (particularly on the server side,) Oracle and now Dell have been and will continue to end Linux indifference among hardware manufacturers.

It's working. It's not fast and good karma isn't the motivation, but it IS working.

Re:Endless hand wringing

[GauteL]

"They are not going to ever really support Linux well. If that's not clear after 12 years of the above cycle, then you haven't been paying attention."

Normally I would agree, but you might be interested to hear that ATI has new owners and it is possible that these new owners will enforce a change in policy.

Now, even if such a change in policy is shouted from every rooftop I wouldn't recommend that any Linux user buys ATI/AMD graphics products until there is actual working drivers for them, but it isn't fair to assume that ATI/AMD will always be as bad as ATI was on their own.

Re:Endless hand wringing

[JKDguy82]

Would you be satisfied with a new release every month? Maybe you should check out their 64-bit FireGL driver release page:

http://ati.amd.com/support/drivers/linux64/radeonp revious-linux64.html

Re:Endless hand wringing

[Anonymous Coward]

Get an Nvidia card if you don't care about open, but want working accelerated drivers. If nVidia has W+X, does that mean ATi has W^X in their driver code? :)

Re:Endless hand wringing

[kiddygrinder]

I don't know where you get that one from, intel have open sourced the drivers for their graphics chips. I suppose they could have done it in an unfriendly way.

Re:Endless hand wringing

[swillden]

I believe you're thinking of Matrox, who used to provide specs. ATI has never been very forthcoming.

Re:Endless hand wringing

[Chandon Seldon]

Some company had a bunch of money and used it to convince ATI to give out specs under NDA to a couple DRI developers. This resulted in the ATI 7000/8000 series cards having reasonably good open source drivers. None of those developers work on X anymore, so no one even has access to those specs any more - much less any recent specs.

Re:Endless hand wringing

[Spliffster]

hehe, good question! Back in the windows NT4 and NT5 days we had servers crashing because of bad ati drivers and had constantly install problems with ati drivers on windows, I never even bothered to buy hardware with ATI chips for linux.

If the microsoft stock drivers run more stable than the ATI drivers, this says something.

Who is to expect good Linux drivers, if they can't even produce decent drivers for their main platform ?

Re:Endless hand wringing

[Anonymous Coward]

The open source ATI drivers have been updated by reverse engineering efforts, and the work suprisingly well for cards up to X850. These drivers make ATI cards the best choice there is for open source operating systems.

M$ DRM and DX is what is keeping open drivers.....

[Joe The Dragon]

M$ DRM and DX is what is keeping good open drivers away from linux. Intel did open there but there hardware at the time was real bad next to ATI and NVIDIA and they had very little to lose as at the time 3D game play was very slow on Intel GMA video.

follow the leader

[ianare]

I think we will eventually see open drivers from amd/ati, after all Intel has open-sourced theirs, so AMD will want to as well. Anything to get their hardware selling I would think. Especially for the server market, where AMD is doing (relatively) well, and the video hardware isn't bleeding edge (I know there are some open reversed-engineered drivers, but it would be nice to have some official ones).

When it comes down to it, as the underdog, AMD has the most to gain, and the least to lose, by open-sourcing their video drivers. And if/when they do, Nvidia will be under pressure to follow suit.

In any case, I would be happy with better support, open or not. With a growing install base and good, stable, 3D graphics it would also make the game companies not completely ignore Linux. Here's to hoping civilization V is cross-platform! (hey, they already started down that path by using much python for IV)

Re:follow the leader

[RAMMS EIN]

``Intel has open-sourced theirs, so AMD will want to as well.''

If wishes were horses...

Holy gramatical error batman!

[CaptainPatent]

Since ATI has been swallowed by AMD, ATI has gone downhil[l]. [With p]aper launches of products and m[i]ssed sh[i]p dates[,] [w]ha[t] makes you thin[k] ATI will make an effort or do the[y] even have the means to do {do - delete} it? I was originally going to make a statement saying that ATI hasn't been too terrible lately, but after all of the corrections I had to make just to comprehend your post, you instead left me wondering if you're missing a finger.

Re:Holy gramatical error batman!

[8-bitDesigner]

Bah, he's probably just on an iPhone!

Re:Holy gramatical error batman!

[Anonymous Coward]

Since ATI has been swallowed by AMD, ATI has gone downhil[l]. [With p]aper launches of products and m[i]ssed sh[i]p dates[,] [w]ha[t] makes you thin[k] ATI will make an effort or do the[y] even have the means to do {do - delete} it? I was originally going to make a statement saying that ATI hasn't been too terrible lately, but after all of the corrections I had to make just to comprehend your post, you instead left me wondering if you're missing a finger. I' osting on a iPhne, y isenstve cld!

Re:Holy gramatical error batman!

[Anonymous Coward]

...and typing one-handed because he's frantically masturbating at the mere sight of it?

yes yes please please

[Anonymous Coward]

Open source the damn things, and then make sure everyone knows about it! (market it)

I *promise* you, AMD, there is big potential here.

ATI can't write ANY drivers

[Tihstae]

ATI is going to write better Linux drivers. How many times have we heard this? What you have to realize is ATI can't write drivers period. Their Windows drivers are the biggest piece of crap on the planet. Yes, they may be better than their Linux drivers but they are still not good. ATI needs people to write drivers for their hardware.

Re:ATI can't write ANY drivers

[ImaLamer]

That's what I don't understand: I wouldn't use an ATI card if it was given to me. After I bought a TV Wonder (laugh it up kids) I knew firsthand that ATI is crap. When you have to install four drivers for once piece of hardware you know something is wrong. Yes I understand that there are audio and video drivers, but three drivers go right to video (you need a driver for their device to device bridge... wtf!)

It's a shame that other companies offer the same exact solutions that work better and cost half as much. If it wasn't for fanboi's there would be no ATI.

Re:ATI can't write ANY drivers

[tech10171968]

I agree with the OP 100%. In my experience ATI has always had superior hardware, but what good is that when the driver situation makes that hardware barely useable? I've been able to enjoy Beryl on my HP Pavillion notebook thanks to the open-source Radeon driver but this graphics unit is only doing less than half of what it's capable do doing, and installing it has been like pulling teeth. Kinda sucks not being able to kick back and relax with my favorite OSS FPS. My next purchase is definitely going to be powered by Nvidia or Intel.

ATI Driver Site

[Anonymous Coward]

On a side note, they should make their driver download site more accessible from a console based browser. I was trying to download their driver and it was next to impossible without a JavaScript/GUI browser. So even if the driver worked well, what's the point if you can't get at it without the help of another computer?

Useless? stupid zealots

[Man in Spandex]

Even if ATI released 100% working, fast drivers, they would be useless if they weren't OPEN and FREE.

You people are all crazy about GPL/Open Source. "VIVA OPEN SOURCE WE WANT SOURCE CODES!". Seriously, how many *nix users contribute to OSS projects? How many contribute code? I bet a good bunch of people contribute because I've seen projects that have died and have been picked up by other developers to keep some applications alive so I won't deny the existence of contributors.

What I will say is that I would honestly prefer closed source drivers like Nvidia but that work correctly and perform great rather than open source drivers with the current ppl @ ATI. Call me crazy but why does every company that develop on Linux have to give in to this OSS idealogy, the idea of giving to the community in order to keep it alive? Seriously the company that do so are great but the companies that don't want to, it's their choice it's their product and it's their software.

Does ATI need to improve their drivers? Hell yes. Is the solution to open source their drivers? Maybe, maybe not. Get a good team of linux developers to regularly work on them at ATI and if the drivers perform great, you will see many happy ATI users who will not need to buy a Nvidia video card to simply to stay on Linux. Sure there will be a few pissed off GPL/OSS/Linux/Penguin zealots who will cry for the source code, too bad for them I guess.

Re:Useless? stupid zealots

[MrCoke]

A few possible advantages of GPL drivers:

- more stability
- AIX support
- proper Xinerama support
- hardware end-of-life cycle when the user decides it, not when ATI decides it

I'm sure you can find a few on your own.

Re:Useless? stupid zealots

[CaffeineAddict2001]

Sure, most linux users won't touch a line of source code. The benefit comes from being able to re-compile the binaries to work optimally with your own system.

Would you buy a car that had it's hood locked shut so that only authorized dealerships could open it?

Re:Useless? stupid zealots

[e2d2]

Would you buy a car that had it's hood locked shut so that only authorized dealerships could open it?

If the alternative was walking? Yes I'd buy it.

Re:Useless? stupid zealots

[CaffeineAddict2001]

I don't get your analogy. Are you suggesting there aren't any open-source alternatives?

Re:Useless? stupid zealots

[Rakishi]

Well if I had to choose between a closed hood car and a yugo (or moped or bike if you want to be nice) I'd choose the former. Open source drivers exist but their performance is apparently not that good.

Re:Useless? stupid zealots

[644bd346996]

Who wants AIX support? Seriously, what AIX systems even have an ATI 3d chip? Or did you mean AIGLX?

Re:Useless? stupid zealots

[MrCoke]

Ugh. Yes :)

Re:Useless? stupid zealots

[Man in Spandex]

Well, most people are putting up with a car with a locked hood seeing how Windows remains the dominating OS for the desktops at home and the workplace so how is that worse than having the same people put up with a car that has an open hood (Linux distro) with a locked compartment (video drivers)? Sounds already better doesn't it?

When the situation is that a game performs better on Linux than Windows using proprietary drivers from the same company and I'm talking about Nvidia, surely somewhere that company has done their homework at making their product work. If ATI/AMD can do the same, that would be a great leap.

I think the problem is because so many companies provide poor support to Linux, the community thinks contributors who are active in the OSS community are the ones who can improve anything but what if you have a company that has programers that know very well what they're doing, should there be a need to open up the drivers? Yeah for enthusiasts it's always good but for the end-users, not really because they have support for their product that they payed for and it works like it's suppose to.

Linux users have been spoiled that's all. It's two extremes, you have the fully customizable Linux and the very closed Windows. Once you get a taste of the openness of Linux, you think it's great and everything should be like that, but then you forget how some things work great on Windows despite being very locked in. It doesn't have to be one or the other. What's wrong with having an Open system like Linux with some closed parts as long as everything works?

Re:Useless? stupid zealots

[edwdig]

Would you buy a car that had it's hood locked shut so that only authorized dealerships could open it?

No I wouldn't, but I also wouldn't insist the dealer give me the specifications so that I could make my own identical engine from scratch.

Re:Useless? stupid zealots

[e2d2]

I may be talking out my ass here but last time I checked I couldn't find any open source drivers for my ATI x1400 mobility. I ended up using the proprietary drivers for Linux. If anyone knows of an OSS alternative please post it.

Now if you mean at a higher level regarding my overall approach to OSS - hell yes. I'll take open source over proprietary any day. I really meant in this specific case, with ATI drivers, there isn't an alternative beyond basic vesa support (AFAIK, again I could be talking out my ass)

Re:Useless? stupid zealots

[Bill_the_Engineer]

Would you buy a car that had it's hood locked shut so that only authorized dealerships could open it?

No, since most motorists won't touch a line of source code. The benefit come from being able to recompile the engine to work optimally with their own driveway.

I kid.

A graphics driver is a part of a PC system and not *the* system, therefore your car analogy doesn't really apply. Take the emission control system as an example, it is part of the automobile and it affects performance. However, other than being able to set some parameters in ROM (for some vehicles) they are black boxes that must be replaced outright when they begin to fail. There are lots of other parts that are not modifiable at all. Sure you can replace the parts with custom made parts, but the analogy was that someone wouldn't by an automobile that had closed source parts, when in fact millions do.

The car analogy supports the closed driver argument better. The car is open source since it allows the driver to work under the hood, yet it contains closed source parts that leaves the driver with no choice but to make a trip to the auto parts store and purchase a replacement.

Personally I would love to have open-source drivers, but I am a realist and live with closed source binaries. As long as the vendor supports the product, I can live with closed source driver. However, when they don't....

Re:Useless? stupid zealots

[Millenniumman]

If said car cost nothing, worked well, and was the only car that fit in my garage?

Why not?

Re:Useless? stupid zealots

[narfbot]

Avivo is my guess for OSS driver, but you're talking about something that's really only a month old.

Re:Useless? stupid zealots

[narfbot]

Today, cars are still fairly open. Enough information is available for someone who buys a car to change things (i.e. specs). Historically the PC was open, and specs were made available to program the hardware. But before PCs became ingrained in the market well enough, Microsoft came along and created a black box that sits between your software and your hardware. Eventually that black box was called windows. Because Microsoft got into PCs very early, everybody got their software through them and to play along, you had to conform to their black box methods. With video cards, it's called Direct3D. So up to the late nineties, video cards were traditionally open. So the market has become accustomed to the Microsoft product, and it's a closed product. To get support for other systems, even open support, has been pretty much impossible as of lately.

So if a car manufacturer released a closed car, it would have a hard time selling because that's not what the market expects. On the other hand in the computer industry, a closed video card is what the market has come to expect so that it gets Microsoft support.

Vista only makes things worse:
http://www.cs.auckland.ac.nz/~pgut001/pubs/vista_c ost.html

Re:Useless? stupid zealots

[PeterBrett]

Well if I had to choose between a closed hood car and a yugo (or moped or bike if you want to be nice) I'd choose the former. Open source drivers exist but their performance is apparently not that good.

Open source drivers where the specs are available are very good. Look at the Intel, Matrox or Radeon R200 drivers, for example.

Open source drivers where the developers have had to painfully reverse engineer the hardware interfaces from scratch are not very good (yet). This is surprising why?

Re:Useless? stupid zealots

[Rakishi]

I never said it's surprising. I simply stated that right now for high end hardware OSS drivers are yugos if you're lucky. You can give excuses and reasons all you want but that's the truth and trying to ignore it when someone says why they use closed source drivers is idiotic.

Re:Useless? stupid zealots

[Peter La Casse]

Who wants AIX support? Seriously, what AIX systems even have an ATI 3d chip? Or did you mean AIGLX?

While the parent poster did mean AIGLX, this does display an advantage of an open source driver: not every computer is an x86, and there do exist small communities with the resources to port an open-source driver to their platform.

Re:Useless? stupid zealots

[PeterBrett]

I never said it's surprising. I simply stated that right now for high end hardware OSS drivers are yugos if you're lucky. You can give excuses and reasons all you want but that's the truth and trying to ignore it when someone says why they use closed source drivers is idiotic.

OSS drivers for high-end hardware suck? Oh right, I'll go and tell the physicists who use our university's eight million pound supercomputer that they'd better switch to Windows.

How well does Windows run on a Cell processor, by the way?

In response to your ad-hominem: the only set of closed-source drivers I've used on Linux that worked well have been the Nvidia ones. All others have been buggy and unstable (some, such as the ATi drivers, unusable), not from lack of effort on my part to get them working properly. At least the less-featurefull open source drivers don't kill my system. My opinion of closed-source drivers is based on hard-won experience. And this makes me an idiot, does it?

Re:Useless? stupid zealots

[Rakishi]

OSS drivers for high-end hardware suck? Oh right, I'll go and tell the physicists who use our university's eight million pound supercomputer that they'd better switch to Windows.

How well does Windows run on a Cell processor, by the way? Well since we were talking about video hardware guess what sort of high performance hardware I was referring to. I mean your response it actually hilarious. In an attempt to ignore my argument you posted something that makes you look like an idiot (who can't comprehend simple arguments or meaning from context).

So how well does that supercomputer play Quake4, by the way?

In response to your ad-hominem: the only set of closed-source drivers I've used on Linux that worked well have been the Nvidia ones. All others have been buggy and unstable (some, such as the ATi drivers, unusable), not from lack of effort on my part to get them working properly. At least the less-featurefull open source drivers don't kill my system. My opinion of closed-source drivers is based on hard-won experience. And this makes me an idiot, does it? Of course since the Nvidia closed source drivers give you better performance than any video card+OSS drivers using the later for high performance graphic applications would make you an idiot. And yes ATI drivers suck because they don't give a damn about linux and have said so themselves (less than 1% of the market is simply not worth the cost to them). That's why you use nvidia and closed source drivers.

I'm still not quite sure if you're an idiot or are trying to side track the argument (by taking what I say out of context, etc.). Either way I doubt I'll reply again, zealots and morons aren't worth my effort.

Re:Useless? stupid zealots

[PeterBrett]

I'm still not quite sure if you're an idiot or are trying to side track the argument (by taking what I say out of context, etc.). Either way I doubt I'll reply again, zealots and morons aren't worth my effort.

Well, I guess that makes two of us, because I have very little time for people who resort to name-calling rather than reasoned argument. Good day to you.

Why not open up their drivers?

[tomshaq]

Hmm.. maybe I am missing something, but why would ATI not want their drivers in linux, whether open or not? I know in my recent purchase of a laptop, if they had an ATI gpu, I ignored it as a choice, as I run Ubuntu, and I want graphics that actually look decent and are able to support 3D. Hopefully now that a big hitter like Dell is urging them on, they can stop ignoring the growing Linux community.

Legal restrictions (esp DRM?)

[Maxo-Texas]

Perhaps part of the issue is that open source drivers could be written to do anything.
Various legal restrictions currently require that hardware does less than it is capable of.
With open drivers, the hardware could do everything- even if it broke the law.

I have a music player-- it won't let me copy songs off it to the hard drive.
OTH, if i take out the memory card and put it in a card reader, I can copy the songs off.

So the restriction on the music player is really stupid. But they did it anyway. Probably for legal reasons.

Re:Legal restrictions (esp DRM?)

[ianare]

For wi-fi, yes, there are legal restrictions set forth by the FCC which prevent a device from operating at a certain frequency. This is controlled by firmware/drivers, so that the same hardware can be used in different countries with different legal requirements. For music and video, they are not legal restrictions, only the greed of the entertainment industry forcing hardware vendors to lock down their devices to prohibit fair use (under the pretense of combating piracy).

Re:It's hard to write drivers for a crap OS.

[Anarke_Incarnate]

So that explains VISTA problems, but we're talking Gnu/Linux here....

Re:It's hard to write drivers for a crap OS.

[Cnik70]

Score: Linux - 1 Vista - 0 MOD Funny

Good heavens!

[Petersko]

"And dismissing a binary simply because it's a binary, without even considering where the best option lies, seems like a fool to me."

My english done gone busted itself all up inside.

Message to Dell: Don't waste your time

[CaffeineAddict2001]

Use nVidia. They make a better product, have better support and happier customers. This is especially true for linux users, but I find it true in general.

I really don't care...

[jessecurry]

I really don't care if the drivers are closed-source as long as the company updates them and responds to bug reports. I hate to see politics creep into Ubuntu where I have to explicitly enable NVidia drivers, that makes it more trouble for me to use the drivers than it should.

Re:I really don't care...

[Anonymous Coward]

> I hate to see politics creep into Ubuntu where I have to explicitly enable NVidia drivers, that makes it more trouble for me to use the drivers than it should.

Technically you have to explicitly enable those drivers on Windows too. You just do it with an installer. On ubuntu, it's even easier -- I got a little "restricted hardware" icon, did a couple clicks in the wizard, and I was going. Didn't even have to reboot.

Re:I really don't care...

[turnip torrent]

I agree. I'd rather have a good, fast closed source driver than an old, poorly-made open source driver.

Re:I really don't care...

[Ant P.]

It's not politics, it's legality. The Kororaa livecd tried shipping preinstalled binary drivers and got a nastygram from one of the kernel devs as a result.

Re:I really don't care...

[jessecurry]

the politics in that case would be those of the kernel devs. It's somewhat annoying having a message that talks about drivers that are nearly imossible to maintain. It's easy to see the open-source ideology being screamed in that statement.

Re:I really don't care...

[WNight]

This is ATI. We've all be wishing for a stable driver since they started, it's not open source holding them back...

If it was open, perhaps someone else would do it for them.

Dell

[erko]

Ah Dell, You must be new here.

1650 pro 512mb AGP cheap (no linux)

[Odinson]

I bought a 1650 in early May. 3D has never funtioned in Linux. Just crashes the machine. Many distro, hardware combo's tried. Works fine in Windows. $50 + shipping takes it. $150 retail.

Yea ATI's drivers are great....

BTW I'll give it to any developer making a serious effort to write open source drivers. I'll even pay shipping.

Re:1650 pro 512mb AGP cheap (no linux)

[ltmon]

The Avivo guys (open source r500 drivers) would appreciate it I would think:

http://gitweb.freedesktop.org/?p=avivo/xf86-video- avivo.git;a=summary
http://www.phoronix.com/?page=project&q=AvivoDrive r

Re:It's hard to write drivers for a crap OS.

[sa1lnr]

After having a 9800 pro and a x800 pro running under Windows XP I agree with your statement. ;)

A little bit of a kick to Creative too?...

[Anonymous Coward]

I'm currently running Ubuntu on a Dimension 9200 - It's a great machine, really powerful, NVidia graphics, runs multiple guest VMs happily, Compiz spinning away on my desktop...a good all-round showcase example to show my colleagues that Linux is now desktop ready.

"Great" they say. Wow! And walk off with their shiny Ubuntu Cds. Then they send me a youtube link to something, and all they hear from my office is "Beep". Oh yeah, fully supported hardware except the Creative X-Fi audio, which has literally Zero linux capability. Creative have been promising drivers for 2 years now, but have repeatedly pushed the ship date as "It has taken more resources than expect to redesign our software and drivers for Vista." so at current look the beta will now not even be until Q4 (http://opensource.creative.com/soundcard.html#X-F I)

So until then, if you buy a top-end Dell box, you won't be getting any sound out of it except some Spectrum-era beeps. Shameful.

For some amusing reading (and some trolls, which isn't helping I guess), here's a (currently 73 page) thread on their forums complaining about this:
http://forums.creative.com/creativelabs/board/mess age?board.id=soundblaster&message.id=31220&view=by _date_ascending&page=73

"please type the word in this image: obstruct" :-D

Re:A little bit of a kick to Creative too?...

[PenGun]

Never never never buy Creative's stuff. A bigger bunch of thieving assholes you will have to look hard to find.

About time..

[TheHawke]

A company the size of Dell finally made a request for improved Linux drivers.

Should have seen it coming though when Dell started bundling Ubuntu with their systems. Since they have an outstanding contract with ATI/AMD, it's only good business sense to request improved drivers.

Any takers that Dell will be making the request that ATI improve their technical support for at least Ubuntu?

Dude, stop pimping out your mom

[Anonymous Coward]

You sick little monkey. No one wants to see your mom's droopy meatbags.

No no, you go to Apple for that!

[Anonymous Coward]

Stevie 'blow' Jobs and all that youkno!

And them macses giving you a job while doing their stuffz and all...

ATI does produce drivers

[Skapare]

ATI does produce drivers. The problem is, they don't release the source code so we can fix them and make them work without exposing ourselves to certain draconian laws.

Linux Driver != Open Source Driver

[keithjr]

The discussions regarding GPL and open-source drivers are irrelevant to the point Dell (and ATI+Linux users over the years) have been trying to make. There's more to making drivers work in linux than opening up the source code.

The more a piece of software makes use of a certain OS's API and specific device control structure, the harder it is to make it portable. Everything to do with how the software interacts with the operating system, and optimizations made therein, have to be re-written, and linux has a very very different device node structure than windows! There is a great deal of effort required to make the same functionality, and the same performance. Nvidia has historically shown more dilligence on this front. The fact that a so many it-won't-work cases exist for the ATI drivers implies they've cut a lot of corners. Yet they continue to release updates. I wonder how many people at ATI are actively working on this...

ATI has had proprietary linux drivers for quite some time now, and as somebody who's used them for about 4 years, I can say they've come a long was in terms of performance. However, dropping support for fairly recent cards is rather troubling, and nothing Dell can say would make a difference there (no market for cards that aren't being sold). And still no AIGLX. Outside pressure might help with that one...

Re:Linux Driver != Open Source Driver

[Slashcrap]

ATI has had proprietary linux drivers for quite some time now, and as somebody who's used them for about 4 years, I can say they've come a long was in terms of performance.

Are you sure you mean performance? Because the performance of their Linux drivers has remained absolutely static for the past two years at around 50% of their Windows drivers. Have a look at some of the Phoronix.com ATI year in review articles where they benchmark all of the releases from the last 12 months. You will never see a flatter set of graphs in your life. That has always seemed highly suspicious to me. Surely the performance should change a bit, even if it's just by accident?

If you mean that the performance has improved, in the sense that you can now get them to work the majority of the time, then I have no arguments.

The more a piece of software makes use of a certain OS's API and specific device control structure, the harder it is to make it portable. Everything to do with how the software interacts with the operating system, and optimizations made therein, have to be re-written, and linux has a very very different device node structure than windows!

My understanding is that the Nvidia proprietary OpenGL driver code is pretty much the same for Windows and Linux, with a relatively small compatibility layer that deals with the different kernels. What does the OS have to do other than provide DMA functions, do the PCI setup and generally just getting out of the way and letting the driver talk directly to the card? I'm betting that's orders of magnitude less code than the OpenGL part of the driver. And what exactly do device nodes have to do with it? Do you think the driver just pipes its output to /dev/agpgart and lets the kernel sort it all out?

Re:Linux Driver != Open Source Driver

[turing_m]

"You will never see a flatter set of graphs in your life. That has always seemed highly suspicious to me."

It does seem very interesting that this is the case. And it would make a good deal of sense for MS to make an under the table deal with them to limit performance under linux especially if it means that more computers are installed with Vista. It hurts a graphics card manufacturer very little to shut off linux, but it means a lot to MS to be able to cripple future competition and mindshare. Gaming rigs last a long time, even if not used by their original owners.

Re:Linux Driver != Open Source Driver

[TheRaven64]

The drivers link to the kernel, so there are some legal issues arising from the GPL if you distribute the drivers and the kernel but not the code. Distributions typically get around this by requiring the user to download the driver separately. The user is not bound by the GPL (since they are not distributing), and the manufacturer isn't either, since they are not distributing Linux (and you can't copyright an interface, so the GPL doesn't apply to the driver until it is linked). Dell, however, will want to distribute both the kernel and the driver. The problem here is that as soon as they distribute the driver, they lose the rights to distribute the kernel (GPL violation).

Couldn't agree with you more

[Anonymous Coward]

I don't even use Linux (BSD). But I was happy with Dell's decision for the very reason you mention, and plan to buy an Ubuntu laptop and install OpenBSD on it.

its good to see

[darth_linux]

a major company like Dell using its enormous influence and... i'm sorry. i can't finish that with a straight face.

Better Drivers?

[Mysticweed]

ATI hasn't produced a decent driver for any platform in their existence, why would they break the streak and change now? I believe it would be a win win for every graphics and sound card to go open source, then we might actually have some decent software that won't crash your computer. (are you listening Creative Labs?)

The foolishness of binary-only anything

[SgtChaireBourne]

When it comes to closed systems like video cards and their drivers, I think only a fool would turn up his nose at a binary simply because it doesn't come with source code.

Haven't learned our lesson regarding security or portability have we?

Popular binary drivers had some unresolved, severe exploits and couldn't be bothered to address them for about two years. That's just an anecdote, but illustrates that the problem is real and not just theoretical. Anecdotes aside, there are inherent problems with binary-only drivers (or binary-only anything). For the obtuse, the interview with Theo de Raadt interview with Jonathan Gray and Damien Bergamini go into more details.

Production mistakes and design flaws aside, happen. That's why we get the effect that "given enough eyeballs, all bugs are shallow". But with binary-only that also means that nearly anything, from back doors to monitoring, can be piggybacked into the blob. You'd be hard pressed to find out. And depending on the vendor for the binary also leaves you dependent on their choice of architectures - not yours, and their lifecycle timeline - not yours.

Some, like the GP, may prefer the GPL, others may prefer other open source license. Whatever. Any of them is a far cry better than no source code.

Also, remember the open source is not just a license, but a development model. Popular hardware will gain development speed and quality for the drivers. It's not like the drivers have any inherent value without the hardware. Opening up the drivers would most likely boost the sales of the hardware they use.

Re:The foolishness of binary-only anything

[3.1415926535]

I know this thread is old, but it bugs me every time somebody drags out the old "NVIDIA knew about it for 2 years" line. Rapid7 didn't do their homework and mistook a similar bug in the X server (which, I might add, is open source). See http://nvidia.custhelp.com/cgi-bin/nvidia.cfg/php/ enduser/std_adp.php?p_faqid=1971

ATI is behind NVidia

[Britz]

and they have a hard time catching up. Maybe they will finally "get it" and release the specs. I think the Open Source Windows drivers will also be a favourite for their stability and lack of bugs. Game developers could also check much better what went wrong and will send in bug reports. Open Specs would help Ati in the Windows market as well.

Re:ATI is behind NVidia

[Anonymous Coward]

Sadly they did "get it" in the past and everyone still bashed them then. Realizing they suck at writing drivers and a growing market of linux users, they released the specs for r100 and r200 chips to dri developers to write open source drivers instead of writing proprietary drivers themselves for linux. The open source drivers written are actually better than the binary drivers ATI released for those chips and are still being developed to this day.
(side note: ATI dropped support for r100 and r200 in later binary driver releases.)

So what happened? Everyone raved on about how Nvidia is linux "friendly" by writing proprietary drivers that work well and ATI are lame since they didn't want to even write drivers themselves (and instead opened up the specs to open source developers).

They thought they were being linux friendly and got burned for it. So what does ATI do?
"Hmm Nvidia gets all this praise and business for their locked down hardware specs and binary only drivers. Screw it, lets do it to." So consequently the specs for r300 and later gpus were never released and ATI started to write binary drivers for linux. Unfortunately the quality of their binary drivers lags behind Nvidia's to this day. Can't put too much blame on them though for not putting in more effort; they tried to embrace the open source community before and now have a bitter taste in their mouth and a bad reputation to match.

Re:ATI is behind NVidia

[Britz]

I bought r200 hardware exactly because of this and I am still running it.

Ati Radeon 9200 or 9250. Cheap and passiv cooling. The computer I am running right now has r300 (free drivers) though. The Radeon 9550 is also passivly cooled. The reason I am running it is because a friend dropped it.

ATI Driver performance

[nelsonal]

Why has ATI struggled with driver development for as long as I can remember. I'm curious why they've never been able to hire better developers.

Re:ATI Driver performance

[JustNiz]

because ATI management are too stupid to think bigger than cutting quality to save costs.

Re:ATI Driver performance

[nelsonal]

That sure seems to be the case, I could understand if it was a first round of products, but it's been pretty obvious since 2001 or 2002 that their drivers have been enough of a detrement to performance (and sometimes operation) that customers are choosing competitors' products.

Re:ATI Driver performance

[elFarto the 2nd]

Oooh, so close to the 20,000,00th comment.

Regards
elFarto

Re:ATI Driver performance

[Anonymous Coward]

Why has ATI struggled with driver development for as long as I can remember. I'm curious why they've never been able to hire better developers.

My guess? It's all down to cutting corners on the development team's size, and just pasting in bugfixes and new hardware has left them with a colossal mountain of buggy spaghetti code in their driver stack. The not-big-enough team is too busy trying to [get new hardware to work | fix major outstanding bugs] to really do the necessary architectural rewrite, and management only adds new team members when it becomes clear that the drivers are currently borked, meaning they can't ever get ahead of the curve. Even good programmers can turn out awful code if the deadlines are strict enough and the workload too high. (Remember E.T. for the Atari 2600? Heh. Yeah. Classic example.)

  The bright spot on the horizon is that AMD took over, and from the rumors I've heard, they were appaled by the state of the opengl drivers (which suck on windows, too) and set up a separate, parallel team in charge of a total ground-up rewrite of the opengl driver, which will run on Vista, Mac, and Linux when it's completed. Remember when that VP of sales said something about a new "open" driver that was in the works? Yeah, I think the dummy meant "opengl".
  ATI employees have said there's a new opengl subsystem in progress, and it should be ready by the end of the year. I don't know, I hope AMD hired some good programmers for that job, otherwise it might be a while. Heh.

Re:ATI Driver performance

[Anonymous Coward]

Congrats on the #200000001 post.

Here's a poem Cmdr made for you.

When the niggers raped the women I remained silent because I wasn't a woman
When the niggers raped the children I remained silent because I wasn't a child
When the niggers raped the other niggers I remained silent because I wasn't a nigger
When the niggers raped me, I couldn't speak because there were 4 black cocks in my mouth.

Ow, my ass hurts and my jaw is broken.

Yes, he's missing a finger

[Anonymous Coward]

He's missing the middle finger, as his entire post attempts to describe a gesture for which that finger is very useful.

Hmmm.

[nowhere.elysium]

Well, I had an abnormal experience of ATI/nVidia - I had an nVidia 7300LE (fairly low end, I know) in my 64-bit Ubuntu box (which was also, in an attempt to get a working driver, a Debian, Fedora, Mandrake and even Gentoo box, in various stages). 3D definitely did not 'just work' - the 3D acceleration drivers had this ugly habit of making xorg take 6 and a half minutes to start up. The relevant thread on ubuntuforums can be found here: http://ubuntuforums.org/showthread.php?t=484455 I'm now running an ATI X600 - it's massively superior, in terms of functionality. The control panel leaves a lot to be desired (the general impression of being in control would be a good start), but it works, and it's acceptable. I do fluxus-based 3D livecoding, which takes a lot out of the graphics card: as such, I need it to be able to cope with some hardcore abuse. I'm not massively pro-ATI or pro-nVidia: the most functional product will get my money, as far as I'm concerned. I don't really care if it's GPL'ed or not, truth be told. Binary blobs suck, but shit happens. Get over it. Until someone out there can keep up with the graphics card industry in terms of writing decent GPL'ed drivers, I'll happily use the blobs.

Re:Hmmm.

[El_Oscuro]

I had similar experiences. I have used both nVidia and ATI cards in my Linux boxes. While the nVidia cards were supported fine in all distros I used, the cards themselves kept frying and always had to be replaced. I replaced my fried BFG card with an ATI and it worked fine. On my arcade cabinet, I use an ATI based ArcadeVGA card to drive the Wells/Gardener arcade monitor, and it works fine with Lincade.

It's two generations newer!

[Anonymous Coward]

[My Mobility Radeon X1600] easily bests my Geforce 6800. Well I'd fucking hope so; the X1600 is two generations newer than the 6800!

Shall I help out here?

[Anonymous Coward]

Since ATI has been swallowed by AMD, ATI has gone downhil[l]. [With p]aper launches of products and m[i]ssed sh[i]p dates[,] [w]ha[t] makes you thin[k] ATI will make an effort or do the[y] even have the means to do {do - delete} it? I was originally going to make a statement saying that ATI hasn't been too terrible lately, but after all of the corrections I had to make just to comprehend your post, you instead left me wondering if you're missing a finger. I can give him the finger for you, if you want.

--Mr. Helper

Re:AMD *will* soon deliver open graphics drivers

[derrida]

As Henri Richard announced (some time ago...).

Closed drivers aren't pragmatic

[Anonymous Coward]

When it comes to drivers, closed isn't pragmatic. Closed drivers don't get audited. Closed drivers don't work if you upgrade your kernel (or need to run an old one), or want to run on an arch that the vendor didn't bother to compile for. And then worst of all, closed drivers enslave humanity!

Standards

[jaweekes]

I'm not trying to be a troll, but why can't we just have a base standard for talking to peripherals? Have a "It will do basic functions" standard driver that will work, and then have propitiatory drivers for extended functionality. I'm still amazed that printers require different drivers in XP for even simple operations, but USB drives do not.

Just my 2cents worth.

Re:Standards

[TheRaven64]

We do. It's called VESA. It lets you set display mode and set pixels on the screen. For everything else, you use a custom interface. For printers, we have plain text. Most printers will function as teletype devices and output plain text in a monospaced font. Better printers have a PCL, PostScript or (sometimes) PDF interpreter embedded, so you just send data in a well-known format and they work. Cheap printers do the rasterisation in software, so they need complex drivers[1].

For USB and Bluetooth, there are a few device classes. These must implement a standard interface. It's not much different to something like IDE or SCSI in this respect. The interface standard includes low-level things like voltages, signalling frequency, etc, but it also includes high-level commands, like 'store block at address x'. USB profiles include things like mass storage devices and human interface devices. As long as the device implements the commands specified, it will work.

The problem with doing this kind of thing for graphics is that the requirements change too quickly. Hard disks are still doing pretty much the same as they were 20 years ago; they load and store blocks from specific locations. The addressing has changed slightly (linear, rather than cylinder/head/sector), and there are some extra bells and whistles like command reordering, but the basic functionality is the same. The same is true of human interface devices; they provide digital signals for buttons and either absolute or relative motion on axes, and that covers most devices. Ten years ago, most graphics cards were frame buffers with a few simple commands like block transfer and draw line. The next generation did depth buffering, compositing, texturing, and a few other things. Then came transform, clipping, and lighting. Then came pixel, vertex and geometry shaders (spread over a few generations). A specification detailing how to do any of these things would be obsolete in a few years. You could say 'the GPU should run OpenGL commands.' That would be really nice, but the problem is that then you've got an interface that is a superset of what any card can do. Everyone does some things in software, so you need a fall-back path. You could have the driver query the card for every single feature, see if it supports it, and then use the software path for everything it doesn't, but the complexity of such a driver would be horrendous. And the interface specification would be obsolete within three years.

[1] I'm not completely against this practice. My laser printer has a 50MHz MIPS processor for running the PostScript engine, and this is the bottleneck for printing. If I send it PCL, which is much simpler to render than PostScript, it prints much faster. It uses more CPU time on my PC, but my PC can spare 2-3 times the processing power of the printer without my even noticing.

I don't get it. They make the damn hardware.

[Sark666]

I know writing drivers that are opengl compliant, and do proper video modes/timings and all the niceties like tv out is not a trivial thing, but they made the damn hardware! It's not like parts of it are a secret to them.

You could say they don't put enough man power into writing drivers for linux, but even on the windows side, ati has been shakey in the past. I remember ati having to always release some patch because a new game exposed some other problem in their incomplete drivers.

Again it's their hardware, instead of taking years of flak in the media, why not just write up to date drivers?! What? Does it take 100 software engineers working around the clock for 5 years to have a working driver?! Couldn't they lay out a unified driver spec (similar to nvidia) put 10 of their guys on it for a month and be done with it? I don't know the inner workings of driver development, so maybe someone can enlighten why this seems to be the insurmountable hurdle that ati can never get past.

Radeon 7000

[Anonymous Coward]

I just upgraded my Matrox Millennium II to a Radeon 7000. I find that if you stay 5 or 10 years behind the current technology you get good, Free software.

It's great being able to view some of the more resource-intensive GL screen savers!

Broadcom as next target?

[Culture20]

I don't know what percentage of Dell laptops might use broadcom wifi, but it would be nice for a big name to pressure them into making usable linux drivers.

Re:Broadcom as next target?

[Slashcrap]

I don't know what percentage of Dell laptops might use broadcom wifi, but it would be nice for a big name to pressure them into making usable linux drivers.

It would be better for them to provide some financial support for the people developing the existing in-kernel OSS drivers for the Broadcom chipsets. It would be a lot easier to make those perfect than it would be to get any support from Broadcom.

They seem to be doing extremely well as it is for a reverse engineering project.

A Thought.

[Brothernone]

This has occured to me more than once. It seems that ATI isn't really focusing on the PC Market. With alot of the consoles usint ATI powered graphics anymore it seems to me their support for much of anything on PC is slipping, I imagine, due to the vast ammount of orders they fill by supplying the likes of Nintendo with graphics. This is assuming that they are supplying an actually chunk of hardware as opposed to simply helping with design. Still, an interesting thought.

Another Technical Insult

[Anonymous Coward]

Good luck, folks. I've owned two ATI video cards in the past and I've always had to resort to user-modified drivers (Omega) for normal operation.

This led me to two empirical truths:

A) Always build your own computer
B) Never buy ATI video cards

GO AWAY MORMON SHIT

[Anonymous Coward]

THOUGHT YOU DIED :(

Psh...

[tsmit]

ATI can't even code Windows XP/Vista drivers.. You expect them to make Linux drivers?!

smart business decision?

[sh3l1]

smart business decision? if ATI didn't release the drivers for linux it was because they wouldn't make very much money. Giving a small percentage of their customers a driver that could cost thousands or millions to develop isn't what many people would consider a smart business decision. Just ask netgear who decided not to support AMD chipsets, which make up a measly %60-70 of the market, with the WG311v3. This is not flamebait, I am making a point.

i believe the adage states

[Anonymous Coward]

you get what you pay for!

Good one!

[PingXao]

15 years ago some of us were asking ATI for OS/2 Warp drivers. *rim shot*

It would be poetic justice if ATI put Dell on hold for an hour every time they called to check in on those drivers. Then transferred them to 3 different parties before cutting them off.

In fact, if ATI promises to do that I will forgive them for the OS/2 lies and bogus promises they made.

Far from useless...

[orospakr]

... source code for your drivers (or indeed any other software you run) is very useful on several counts:

• They permit upstream to merge the code. Proprietary drivers aside, installing Ubuntu on a modern machine is far easier than installing Windows, simply because you don't have to fetch (and keep up to date) a host of drivers and applications from third party sources.
• Merged code (for instance, Free Software drivers in upstream Xorg) can be kept consistent and up to date with API changes in the common infrastructure code. Common code between multiple drivers can be factored out. As it turns out, Windows and Linux have both had three different USB stacks. The only difference? Windows still keeps the first two around to support so-called "legacy drivers", whereas all the drivers in Linux have simply been updated
• Maintainership of the driver beyond the product life cycle. As Greg KH himself has said, even if your hardware is only used by ONE person in the world, it is welcome in the kernel so long as it is in good condition and someone (not necessarily anyone with any special privilege; all that is required is someone able with available time) is willing to help maintain it. This is very good, because generally proprietary drivers (for proprietary systems) become useless as kernel evolves.
• The transparent and flatter development process for Free code is very useful: you can actually speak to the person who wrote the code, and retrieve their work as it gets done. Public version control, mailing lists, bug trackers, and IRC channels for the win.
• Real computer professionals actually know how to program. Despite your remarks to the contrary, myself and many people I know do contribute to upstream Free Software projects. Even for quick hacks to make things work right now, the source code is immensely useful. Again, I and many others have made small fixes to drivers to get things to work in a pinch.
• Free code usually is subjected to rather rigorous quality standards before inclusion in upstream. Linux is a fine example; in fact, they are notorious nitpicks.
• In summary, not only does Free Software promote a non-tiered development model, but it also simplifies things for the user and technicians who simply want to make the computer work without silly restrictions confounding the process.

this is a good thing

[Anonymous Coward]

my neighbor just bought a dell 1521 with ati graphics and the ubuntu install was flawless, except for the video driver;-(, he should have went with the intel model 1520.

Re:Poll Troll Toll

[mrsteveman1]

LUNIX

i got five on it.......

nm

Re:AMD *will* soon deliver open graphics drivers

[Disfnord]

Uh, yeah, no. That blog is bullshit. He said they will work on better Linux support, and they've been saying that for years. He never said anything about open source drivers.

That will be a first for any industry

[Master of Transhuman]

"ATI will make the smart business decision and give customers what they want."

Never seen it happen yet in my 58 years of breathing.

Humans ALWAYS make the wrong decision. It's a wonder the species still exists - must be the ability to breed out of season...

A race?

[sworoc]

Who stands a better chance?

A) Nvidia, with the lead, but still no completely open drivers?

B) Intel, with completely open drivers, but nowhere close to the lead or high end graphics cards?

C) ATI, with moderately capable hardware and struggling drivers? If they open their drivers up completely and Dell turns to them for their supply of graphics cards, could we see a surge in popularity amongst Linux users?

Long-term, either Intel will catch up on the hardware side, or Nvidia and ATI will have to open up their drivers. The first of the latter two to open up their drivers completely will probably get the edge they need in the POSIX user crowd.

I guess we can always hope that AMD wises up and makes this a 3 horse race!

Re:A race?

[Anonymous Coward]

Intel will win. They are all in a race to make high end specialty multi-processor graphics cards. That's why AMD bought up ATI and why Intel started up a similar project. Those cards are going to start out on high performance computing which is dominated by Linux. It is likely that Intel will push their solution as a general programmable architecture upon which applications (like graphics layers) can be built and will leverage open source to get their drivers up to speed faster.

Re:A race?

[TheRaven64]

Intel. They don't need to own the high-end. Every year, the size of the high-end market shrinks, while the size of the commodity market grows. Remember SGI? The new Intel chips are probably good enough for 60% of computer users. Next year they will be enough for 80%, then 90%, etc. Even if ATi and nVidia keep pushing the performance envelope, and Intel continues to lag behind, their potential market will increase.

By the way, one of the best selling graphics products (made by Intel and a few others, under license) is a slightly tweaked version of the PowerVR chip found in the DreamCast. You might not have seen it on sale, but there's a chance you own a device containing one. They're found in PDAs, mobile phones, and other devices of that form factor. Considering how many more mobile phones are sold than PCs even now, that's the market I'd want to be in.

Re:I don't get it. They make the damn hardware.

[springbox]

Sometimes you'll see poorly written drivers if the people designing the hardware also write the software. I would hope that this is not the case for a company like ATI.

Re:I don't get it. They make the damn hardware.

[the_greywolf]

Again it's their hardware, instead of taking years of flak in the media, why not just write up to date drivers?! What? Does it take 100 software engineers working around the clock for 5 years to have a working driver?!

They're Canadians. This surprises you?

This is news how?

[hitmanWilly1337]

C'mon here, anyone who's ever tried running ATI cards under GNU/Linux knows the pain of a keyboard to the forehead. We've been asking for better drivers from these guys for years. Now Nvidia, while they are in fact proprietary, they at least work. ATI's are closed source and broken too. Which is why I believe that the open-source ati driver is actually ahead of the open source nv one. Necessity is the mother of invention.

Re:GPL or nothing [Amen]

[gnat79]

I will just add my comment that I am a stickler for the GPL (or other Free license). Software is useless to me if I can't have the four freedoms. I don't like to hear all the people here on slashdot brushing it aside; if any group of people should understand the GPL's impact, you should.

Good hardware is one that "just works" ...

[Anonymous Coward]

When I buy hardware, I am only going after hardware that "just works" with most of the OS-es running on a Personal Computer.

Pay attention, Intel and ATI: Good hardware is the one who runs on almost any OS. Want to stick with MSWindows only? This sends a very wrong message to the end-users. Imposing such limitation will back-lash sooner or later.

For the past 15 years the most reliable ( for me, reliable means the hardware who never let me down ) hardware that I voted for with my wallet was/is AMD / Nvidia. It never let me down with any MSWindows OS version, it went ok with Linux (Debian/Linux), prices too were allways reasonable.

I never buy "bleeding edge" hardware, I prefer those that have at least one year of testing by the users. "Drivers in a can" don't allways "just work", software needs time to include new, usually "buggy" drivers.

It is sad when we, end-users purchase hardware like ATI and Nvidia (so You got our money) but often the software drivers for those products are just NOT WORKING with an Operating System of our choice.

Wake up, video-card manufacturers! Your products will only become successful when they will "just work" no matter what Operating System the Personal Computer of the end-users runs.

It is US, the people who make your business that are asking you to promote quality, diversity and cooperation with Linux, BSD, Sun, as well and not the developers of those Systems. Respect and honor the one who purchase your products.

Complete interface specifications

[Schraegstrichpunkt]

What we really want is complete interface specifications, so we can develop our *-licensed drivers ourselves, and, y'know, actually do something new every few years.

AMD/ATI reputation

[Sits]

See AMD/ATI graphics drivers Slashdot greatest hits for some further details. I would love to see better (and ideally Free) Linux support from their drivers (yes there is some Linux support out there but I'm told if you have the very latest ATI/AMD HD X2xxx cards then you will have no 3D on Linux). This is not too surprising because for around 6 months after launch there was no X1xxx Linux support. The drivers have DEFINITELY improved over time (the number of crashes due to binary ATI drivers has fallen) but while AMD/ATI are still doing stuff like refusing to get back to developers or AMD/ATI are allegedly misappropriating developer code I feel very uneasy and currently I would be wary of buying a machine with a new AMD/ATI graphics card for use with Linux.

I KNOW WHO YOU ARE

[Anonymous Coward]

I SHOULD POST YOUR IP LOL

But you can break the "legal requirement"

[Anonymous Coward]

with the closed code:

a) Move to an area where your current power spectrum is illegal
b) Install the firmware for an area where their power specrum is illegal

neither require the code to be opened.

So if closing the code is because of legal repercussions about people overriding the legal limits, then how come they aren't already in the chokey?

"Slashdot requires you to wait between each successful posting of a comment to allow everyone a fair chance at posting a comment.

It's been 14 minutes since you last successfully posted a comment"

Re:M$ DRM and DX is what is keeping open drivers..

[One Childish N00b]

DX are keeping open graphics drivers away from Linux? RMS oughtta suit up in some spandex and lay the smack down on them in the ring! Oh, *Direct*X - my bad.

ATI? Hopefully Broadcom as well.

[ihop0]

Dell AMD notebooks generally come with ATI graphics cards, and Broadcom based wireless cards which are poorly supported for Linux (except by the reverse-engineered bcm43xx, which is making decent progress).

So far the only Dell Linux notebooks are E1505s, which ship with more linux friendly intel wireless & nvidia graphics cards.

Hopefully if Dell is trying to influence ATI for better drivers, they'll be pushing Broadcom for better wireless drivers as well.

Same here - for now

[Lonewolf666]

I got myself a new PC several weeks ago, with a NVidia 8600 GT mostly because I was not so convinced with ATI's Linux drivers in the past.

Considering my future purchases, there are good news and bad news for ATI.
Good: I might reconsider if they actually release good Linux drivers, especially if those are Open Source.
Bad : As I'm quite happy with the performance of my current PC, the next upgrade might be 5 years in the future ;-)

Re: DAMMIT err I mean ATI

[rkaa]

True. And you need a new keyboard and eyeglasses :)