Insight Into AMD's Linux Driver Development

Cowards Anonymous writes "It's no secret that ATI Technologies has had a rough time in the past delivering display drivers that met the expectations of their customers. When ATI started out producing a FireGL and Radeon Linux driver they for some time were greatly behind NVIDIA's feature-rich driver. The early ATI Linux driver had lacked essential functionality such as PCI Express and x86_64 architecture support and was also affected by stability and performance problems — not to mention a great deal of bugs."

Comments

rough start

[phrostie]

when i switched from NVidia to ATI, it was a rough start.
for the longest time i couldn't get the driver to build/install, then one day everything just worked!
i can't tell you which version it was, but from then on, i've had no problems or complaints.

an open driver would be nice, but even still, my compliments to them.

Re:rough start

[IllForgetMyNickSoonA]

So you had a rough time gettint the driver to even compile, let alone instal or even (gasp!) work. Then, one day, it just started working.

So far, so good - this is a typical "ATI on Linux" story, but of the happy-ending sort (which are rather rare, from what I saw so far).

What I do not understand is which way do they deserve your compliments for providing such sub-par software? I'd bring the card right back to the shop I bought it, demand my money back, and buy a nVidia! I haven't had a problem with nVidia drivers for years now.

Re:rough start

[Chandon Seldon]

Everything works? So you can use Firefox at a reasonable speed when logged in as a second user now? You can use Beryl now? Those things sure don't work on the X1300 I bought (a horrible mistake) a couple months ago.

It's really absurd - if they'd just release the programming info for their hardware the X.org drivers would support this stuff inside a week.

Re:rough start

[phrostie]

mine is in a HP laptop.

i use 3DDesktop, and other 3D stuff from SearchandRescue and FlightGear to Blender and Ayam3D.

my older desktop had a NVidia card, but some of the later drivers were giving me trouble.

so i gave ATI a try. as i said, it was a rough start, but it's working flawlessly now.

Re:rough start

[loudambiance]

I can run Beryl with no lag on my x1900, so the driver support is there for the x1XXX series, your x1300 should work.

Re:rough start

[alexraison]

I signed up to slashdot specifically to reply to you. I have an X1300 and for months have been unable to get it to work under any flavour of linux so I felt obliged to give someone else suffering the same problem a heads up. I installed the newest Sabayon this afternoon. It just worked straight off with full support for Beryl once installed. I've never been quite so happy to see that funky 3D cube before in my life.

Re:rough start

[lems1]

I couldn't agree more.

I wasted a lot of time and effort trying to post my opinion in the rage3d forums. Reporting bugs; reporting success or failure with various combinations of ATI drivers. Then one day they decided that since using the composite extension breaks their buggy video drivers, they would rather turn off the possibility to even use it!

Then you are faced with the dilemma of either going back to an older version of the driver that had composite support (and you could use Xgl for a certain amount of time), or forget about it and continue the 1-bug-at-a-time fixes that they released every other month.

In the end I decided to cough up $80 for an nvidia card that "would just work" and I stored my ATI in a shelf...

I doubt I'd ever buy another ATI or recommend it to anybody I know. A lot of people feel the same pain.

Re:rough start

[Brotherred]

Everything works? So you can use Firefox at a reasonable speed when logged in as a second user now? You can use Beryl now? Those things sure don't work on the X1300 I bought (a horrible mistake) a couple months ago.

It's really absurd - if they'd just release the programming info for their hardware the X.org drivers would support this stuff inside a week.

DITTO and that is why as a exclusive AMD CPU user and current ATI user I am so pissed of at them. How do they dare tell us how hard they are working when they could make it so much easier on them selves. KISS OFF AMD/ATI! Go dig your self a grave! This was pasted from the Phoronox forums where I have the same login. http://www.phoronix.net/forums/showthread.php?t=31 21&page=6

Re:rough start

[Chandon Seldon]

Did the fix the "2D acceleration in a second X session" bug? That one actually annoys me more than not having 3D desktop support.

Not really surprising

[Endo13]

It's also no secret that ATI has long had problems with their drivers for Windows too.

No, this is not a troll. I use ATI cards almost exclusively myself and I prefer them over NVidia, but I do have to admit that Nvidia's drivers as a general rule seem much better designed and simpler to install.

Re:Not really surprising

[Eukariote]

Sigh, yet another general statement without supporting evidence. I think your post is a sly bit of astroturfing for NVidia. ATI has had WHQL ( http://en.wikipedia.org/wiki/WHQL_Testing ) certified new driver releases for years now. NVidia has only recently been able to get their new releases WHQL qualified. Sure, there is more to drivers, but it indicates that ATI has had a solid development, testing, and qualification regimen in place for a long time.

Re:Not really surprising

[Scoth]

Not sure what kind of "supporting evidence" you might want, but I once bought a Radeon once to replace my aging Voodoo3. I forget the exact model number, but at the time it was about $175 or $200. In some games it was *slower* than the Voodoo3. I gave up and reinstalled Windows clean, but still had the same trouble. Took it back, got a GeForce 3 Ti 200 for about the same price, and it worked beautifully out of the box. Fast forward to several years later and I was looking to replace that card, and I got another Radeon. Installed the drivers, slapped in the new card - poof, blue screen on boot in the video drivers. Stuck the GF3 back in, removed it from the DevMan, and manually installed the VGA driver. Rebooted with the ATI, installed the drivers - back to Bluescreen on Boot. So, another clean install of Windows, and still got bluescreens on boot. Took that Radeon back and got a GF 6600GT again for about the same price, slapped it in, and it's worked ever since. I have a hard time believing I had other bad hardware in there to cause the troubles since in both cases the GeForces worked perfectly. Not to mention the GeForces worked much more nicey in Linux than the Radeons ever did. I genuinely gave ATI two tries now, and both times I was hit with troublesome drivers. I doubt I'll be buying another Radeon anytime soon.

Re:Not really surprising

[Endo13]

Sigh, yet another general statement without supporting evidence. I think your post is a sly bit of astroturfing for NVidia. No it's not. I can't really say much about the Nvidia drivers because I've never owned an Nvidia card personally. But when I have to install them for someone else, they always just seem to work.

I've used ATI ever since I finally gave up my Voodoo2. I have just about every version of drivers ATI released since the first Catalyst came out, and quite a few from before. They all worked, for the most part, but the install process does not seem as streamlined as Nvidia's, and I still keep hearing from people who have a problem with this or that part of the driver install process crashing, not working properly after reboot, etc. I've also encountered a few drivers over the years myself that required some extra work and a couple reinstalls to get them to work properly. However ATI has come a long way, I'll readily agree to that. Their drivers today are much better than they were when I first started using them.

For WHQL certification, we all know that doesn't mean a whole lot. I've used non-WHQL drivers that worked perfectly every time on many different machines, as well as certified drivers that caused BSODs.

Yes I know this is all just anecdotal evidence, and it is only intended as such so please take it with a grain of salt.

Re:Not really surprising

[LWATCDR]

Or that WHQL isn't worth that much? The latest ATI drivers for DX10 seem to have a fair number of issues at least with this Demo http://www.extremetech.com/article2/0,1697,2134729 ,00.asp
It could be a software issue but the Nvidia card works fine with it.

Re:Not really surprising

[Wicko]

The Omega Drivers have always been a good substitute, for both ATI and nVidia. ATI's drivers aren't all that bad, I haven't personally had any issues, but their software is a mess. I have an All In Wonder X800 XT, and I tried their PVR software and it was buggy and unresponsive, as well as poorly laid out. I decided to use a third party PVR and it worked perfectly.

Re:Not really surprising

[Endo13]

It's been a long while since I tried Omega, but the few times I did they ended up being worse than ATI's drivers.

I've personally never had enough trouble with ATI drivers to make a big deal out of it, which I guess is partly why I still use them.

Maybe I should make a disclaimer: I wasn't trying to say that ATI has lots of problems with their Windows drivers - just that they have more than NVidia does, and that this has been the case pretty much ever since there's been an ATI vs. NVidia competition. Which, again, I don't think is news to anyone.

Re:Not really surprising

[Wicko]

Really? What kind of issues did you run into? I'm curious, I've owned mainly ATI cards, a 9700, 9800XT, my current X800XT. I've also installed X1900XTX for a friend, X700 on two separate occasions. I've had one nVidia card, the GeForce 3 Ti 200, and I can't recall having problems with those either. Either I've been extremely lucky in all cases, or the people having problems have had other hardware causing these issues.

Re:Not really surprising

[yourexhalekiss]

This also not a troll... I just figure that the people that read this thread might find this relevant.

I had a hellish time installing ATI drivers for my laptop with a Radeon Xpress 200M chipset on Ubuntu Edgy and Feisty.

I ended up following the instructions at http://ubuntuforums.org/showthread.php?t=321766 and it worked great.

The instructions are written for Ubuntu Edgy, but they also work for Feisty if you use a newer driver from ATI's site and adjust the instructions accordingly. The instructions seem generic enough that they should work for other distros as well, although I don't want to be quoted on that.

Re:Not really surprising

[Endo13]

Well, the thing with ATI cards and drivers is they can be kind of hit or miss. One prime example is my friend's PC that he was trying to use a 9800 Pro in for his kids. Nothing he did could get the card to work. Finally he started doing some searches online and discovered that many other people had the same issues, and that the 9800 simply doesn't work with his motherboard. There's also been many instances where a certain driver version with a certain card model would not work with certain games, motherboards, etc.

For myself, I've had a few issues with the drivers hanging on install or causing a BSOD after reboot. But like I said, never really a big deal. In all cases I was able to sort things out with a reinstall of the drivers, or uninstall and reinstall an earlier version of drivers that worked fine.

Re:Not really surprising

[nschubach]

WHQL does not guarantee ease of use, installation or compatibility. It just means that it tells Windows what it wants to hear instead of what it might need to hear.

Re:Not really surprising

[Gr8Apes]

I've used both for some time now, although not the latest incarnations. (ATI 9800 Pro, nVidia 7900 GS latest two) They both work well in single and dual monitor use, although I like the ATI dual monitor setup better. I should also note that my new Dell laptop can't ever seem to get the dual monitor thing right (nVidia) with something that is modified from a standard configuration. This is not a problem on my Macbook Pro, so it must be a Windows nVidia driver issue.

nVidia definitely has the crown in games currently, even though the R600 has specifications that would make any gamer drool. Hopefully AMD/ATI's drivers will get the boost they need to make the hardware work (I can't imagine the hardware sucks as badly as the tests make it out to at the moment)

Either way, as you can tell, I'm no longer on the cutting edge of graphics, so I can afford to wait until the driver issue is resolved and the prices come down to where it's a small purchase.

Re:Not really surprising

[stewwy]

quote from their latest driver

This release of Catalyst(TM) introduces the second generation ATI Theater(TM) 650. This product will provide support for the new MCE 2006 requirements such as DRM support. Further, it will include features to support ATI All-In-Wonder products. It will also provide improved TV quality and Broadcast Flag support which enables full US terrestrial DTV support.

this is written as if its a good thing!

Re:Not really surprising

[Andy Dodd]

In addition to that, from conversations with friends who are stuck with ATI cards until they can afford to upgrade, they seem to constantly have issues where Game A requires driver version Y or below, but game B requires driver version X and above, with X > Y.

The end result - having to uninstall and reinstall drivers when changing between game A and game B, or simply not playing one of the games. (I don't remember exact driver versions, but one of the games in question was Civ IV about 6 months after it was released.)

Meanwhile, I have had to downgrade NVidia drivers a total of once, and that's because I was using the most bleeding-edge drivers I could find. Note that not a single piece of software actually NEEDED those bleeding edge drivers, so I didn't have to give up anything when I downgraded.

Long, long ago I had an ATI card with utterly horrific driver support, and haven't even considered ATI since then since I have yet to see any evidence that the situation has improved sufficiently. (They've improved a LOT, but the state of ATI drivers used to be so abysmal that they're still way behind NV in terms of driver quality.)

Re:Not really surprising

[Bert64]

The macbook pro uses an ATI chipset...
When running Linux on the macbook pro i also couldn't get dual head working, so there's definately an issue with the linux drivers.

Re:Not really surprising

[gstoddart]

Sigh, yet another general statement without supporting evidence. I think your post is a sly bit of astroturfing for NVidia. ATI has had WHQL ( http://en.wikipedia.org/wiki/WHQL_Testing ) certified new driver releases for years now.

Hmmm .... "Windows Hardware Quality Laboratories testing" .... is this like buying a PC which says it's "Vista Ready" only to find out that means "well, not Vista with the new GUI stuff, just running with the old GUI"? It's got no credibililty with a lot of people. (eg. , specifally about ATI cards)

A Windows compatibility rating is frequently not worth the paper it's printed on, except to the company who sells it because people think it will work and buy the product. Many of us don't actually take that to mean anything significant in terms of how well the hardware actually works. It merely seems to mean you paid Microsoft for the right to put on that sticker.

And, anecdotally, I've known quite a few people over the years to have huge problems with ATI drivers -- on Windows or any other platform. They may have gotten better, but for some of us, they still have a bad reputation for quality when it comes to their drivers.

Cheers

Re:Not really surprising

[jimicus]

The problem is PC hardware in general terms is such a mash of thngs with varying ability to follow specifications that every once in a while you happen upon a hardware configuration which just does not work.

Old Example: Creative Labs AWE64 ISA sound card, external serial modem, Windows NT 4. Version 5 of the soundcard drivers - the modem stops working. It still dials, still sounds like it's handshaking but it can't complete the handshake.

Solution: Upgrade the soundcard drivers. Took me ages to figure that one out.

I don't think things are substantially better today. There's a reason Macs "just work", and it has just as much to do with a limited hardware platform as it has to do with Apple's amazing ability to manage projects. I would point out that every other OS in existence which has been marketed along with specific hardware (eg. Solaris, AIX, VMS) has "just worked" with its respective hardware.

Re:Not really surprising

[Cal Paterson]

Sigh, yet another general statement without supporting evidence. I think your post is a sly bit of astroturfing for NVidia. ATI has had WHQL certified new driver releases for years now. NVidia has only recently been able to get their new releases WHQL qualified. Sure, there is more to drivers, but it indicates that ATI has had a solid development, testing, and qualification regimen in place for a long time.

This statement is utterly untrue. WHQL is pretty much just a case of;

1. Co-operate with MS on driver releases (and institute their "minimum-standard" level of QA)
2. Pay MS what I'm sure is a large some of money for the privilege

Lack of WHQL doesn't indicate anything about driver quality apart from that certain companies are co-operating with MS to institute a minimum standard; many, many third parties develop drivers over and above this already. That fact that companies do not do WHQL says less about their hackers' development style than it does about their executives attitudes towards unjustifiable costs.

Re:Not really surprising

[Stevecrox]

I owned a Hercules All in Wonder 7500, if I wanted to play games I had to use the Ati 7500 driver (for lord of the rings I had to use the 8500 driver or suffer strange texture effects) for the tv turner I would then have to switch the All in Wonder 7500 driver. If the driver had been consistant I wouldn't have minded for example Uru: Ages Beyond Myst didn't require me to switch driver, Republic and Sims required the 7500 driver and the previously mentioned Lord of the Rings wanted the 8500 driver. Sims 2 ran using the All in Wonder driver, Sims 2 university would crash frequently complaining about the driver.

I've only owned the one ATi card but it was enough to put me off them.

Re:Not really surprising

[grolschie]

It's also no secret that ATI has long had problems with their drivers for Windows too.

This ain't a troll either. I concur with you. I will never ever buy another ATI videocard. I am still yet to get my TV tuner working on my radeon with Windows XP Pro, and I have tried too many combinations of VGA drivers, WDM driver, MMC and MDAC, etc. Crapola.

The featureless Linux drivers when working, kinda worked ok, but when switching to the console via CTLT-ALT-F1 etc, I would later return to a screwed X.

Re:Not really surprising

[TheShadowzero]

Okay, if you wan't anecdotal evidence... My friends (brothers) have matching laptops. They are both Toshiba Satellites and are pretty decent laptops, but they use ATI cards. A year or so ago, BOTH laptops suddenly started to overheat like crazy (the fan for some reason was on the bottom of the laptop) and could literally only be used by propping them up on something or else they wouldnt last 20 minutes without the power simply shutting off. Even while the laptops were propped up, the screen would flicker, the color would get all messed up, it would go to 640x480 resolution and then BSOD. I kind of play tech support for them, and while they thought their windows installations were corrupted and needed to be reinstalled, but when i watched as the BSODing happened, i figured it must be the graphics card considering all the symptoms are graphical ones. So i was repairing the drivers when it BSODed again. right after i had removed the drivers but before i could install the newer ones. d'oh! so that computer HAD to be reformatted (and it was a driver error or something because it works fine now but i doubt a messy windows installation could cause a computer to overheat in twenty minutes). The other one experienced these symptoms more recently and I don't know what they've done with it yet. Sooo yea..ATI drivers killed their laptops.

Re:Not really surprising

[Jarik_Tentsu]

I have heard this a lot, but being an owner of a 9800Pro manufactured *by* ATI, and having a Gigabyte manufactured 9600XT, we've noticed no real problems with the Catalyst Drivers (granted, the drivers that came on the driver CD for both of our cards were unappealing and not great...but after installing Catalyst we both increased our benchmark scores by up to 500).

In fact more of my friends with nVidia cards seem to have the problems with stability. =S

Is this problem just apparent in newer cards? Or am I just insane?

~Jarik

Re:Not really surprising

[Hal_Porter]

ATI has had WHQL ( http://en.wikipedia.org/wiki/WHQL_Testing ) certified new driver releases for years now. NVidia has only recently been able to get their new releases WHQL qualified.

So this must be about ATI then ;-)

Re:Not really surprising

[codeguy007]

While generally I will concede that Nvidia has better drivers however I have had issues at times with Nvidia video cards and drivers. There was an issue with my GeForce FX 5200 and the newest drivers: Some games worked great, some would hang partway through the game. So I ended up having to install 6 month old drivers to get all the games to work. It was a performance hit though on some of my games.

My newest card is an x700 (well if you don't count the 73002go in this laptop) and I find it to be and adequate card and really don't have a complaint with it. It's not the fastest but it does the job good enough. I have had no driver issues with it since it was installed.

Sometimes it's not necessarily the drivers as much as the hardware. ATI has a tendency to make oddball cards that might even have names similar to good working cards but just don't work. It's always best to check out the reviews and comments on any card before buying it.

I think the worst case of this is my TVWonder. The card works great under windows, albeit an old card. SVGA, Composite capture under Linux works very well. But I can't for the life of me get the tuner frequencies right in Linux. That version of the card miss identifies itself as a PAL tuner instead of NTSC. Now what baffles me most is that I have even tried to create my own frequency table and I just can't sync up more than 1 or 2 channels. It's never been a big deal as I just use SVGA in from my cable box. Other version of this card work fine under Linux.

Re:Not really surprising

[Gr8Apes]

I'm aware the MBPs use the ATI chipset, I probably should have mentioned that. I cannot comment on any Windows ATI driver issues with dual head, because I've never had a windows laptop with an ATI chipset. (I downloaded bootcamp but haven't installed it yet as my need for windows ever shrinks:) I have used a desktop windows system with dual head, and I like their implementation a little better than nVidia's.

As for Linux, all my linux installs except for 1 are headless and that one only has 1 very old CRT attached.

Re:Not really surprising

[Andy Dodd]

I have (like you) had a few rare occasions where NVidia had a driver version that had some stability problems requiring me to backlevel them (to be honest, those were both clearly marked as betas...)

I have not had any problems where some software requires newer drivers and other software requires older drivers, requiring either frequent driver switching or not using a certain software package, a story which seems all too common with ATI cards.

So ?

[BESTouff]

When ATI started out producing a FireGL and Radeon Linux driver they for some time were greatly behind NVIDIA's feature-rich driver.

And they still are.

Re:So ?

[drgonzo59]

I second that. fglrx in Linux on x86_64 still sucks. And where the hell is Composite? Why is that not working yet? That should have been done a long time ago. They keep releasing new versions (such as 8.37) but they never fix the outstanding issues. Installation is still no fun if you happen to run a non-RPM based distro (you know... Ubuntu or Debian). It sucks that my laptop came with X1400 and I'm stuck with AMD/ATI...

The best way...

[twoboxen]

to build a customer base is to alienate your existing customer base. I bought an R200-based laptop a couple years ago. ATI decided to just not support those cards in their fglrx driver package one day. Why would I buy from a company who won't continue support for their own products for more than a couple years? I will make every effort to never support them again until they get customer/product support in order. NVIDIA, bravo.

Re:The best way...

[the_humeister]

If you're using some sort of Linux/*BSD/etc, you shouldn't have to worry because X.org has had mostly full and useable R100, R200, and recently R300 open source drivers for quite some time now. They're decent. I've been playing Unreal Tournament (and variants) without problems. The only issue is visuals with Doom 3 do to S3 Texture compression being patented. If you're using Windows, well good luck!

Re:The best way...

[ajanp]

Yeah, I had a horrible experience with ATI support a couple years ago and decided to make the switch to Nvidia and picked up a 6600GT. I had problems with the Nvidia card almost immediately and initially regretted the decision to switch, until I started looked at the support that was available from a very active Nvidia community. It took maybe 15 minutes to fix the problem, and even though I've moved on to another card since, I still keep one of the support threads http://forums.nvidia.com/index.php?showtopic=1029 bookmarked mainly because of all the information in the thread, but also because more than 2 years after the thread was initially created, there are still posts being made.

Is it a bad thing that people are still having problems with the card 2 years later? Maybe, but at the same time it's an outdated card (multiple newer-gen cards have been released since), yet people continue to offer support and help people to resolve issues and newer driver releases still have updates for old cards. I'd prefer that situation after dealing with ATI's non-existent support anyday. Granted that almost two years have passed and I certainly hope that ATI has improved support across platforms nowadays and would hope the ATI community is as helpful for supporting old (and new) products, but until I can confidently assume that ATI will offer equal or better driver and customer support than Nvidia, I don't see myself going back.

Re:The best way...

[twoboxen]

You're right... i have been using the radeon driver for several months. Out of principle, though, I'm pissed. :)

Re:The best way...

[linuxrocks123]

More information on the S3TC software patent stupidity:

http://dri.freedesktop.org/wiki/S3TC

Re:The best way...

[udoschuermann]

I've got a 4-year old R250-based laptop. The proprietary ATI drivers ceased to be supported and 3D "performance" was never worth talking about, nor was it ever stable enough for serious work. I switched to the open source radeon driver and got quite an awakening: glxgears went from 120fps to 1500fps; even at full resolution (1600x1200) glxgears still ran at 450fps.

I bet I get significantly better performance out of my old R250-based system with the open source driver than I'd get from an expensive brand-new card with the shabby proprietary drivers. So, I wouldn't bother with any ATI card that doesn't have 3D support in the open source driver. End of story.

What is the point of the QA process!

[Anonymous Coward]

What is the purpose a two months long QA process, when the result is broken drivers either way?

The long QA process *guarantees* that the drivers break on new kernels.

The long QA process isn't worth jack shit. The recent 8.37 drivers locks up my computer 100% of the time.

Did ATI need 3 months to QA a version check patch for the xserver-xorg-core version number, just to release drivers that no longer work?

With such shitty drivers, I would think it would be better to at least release early and release often.

what a joke

[radarsat1]

While ATI/AMD is working steadfast in addressing all of these issues and further enhancing their level of Linux support, many of their customers do not realize all of the work that goes into these drivers.

Whatever. They don't need to do any work. All they need to do is open up the specs, and people will do all the work for them. People aren't bitching that the drivers don't work, people are bitching because they aren't allowed to improve them.

There's a whole community out there willing to do all the software work from scratch, but they don't have the resources to create the hardware. The hardware developers somehow see this need to provide the software themselves, instead of taking advantage of the community, but then go and do a shoddy job of it. That's why people are annoyed by the whole thing. It could be so much better, with very little effort from ATI, but they steadfastly refuse to play nice, forcing developers to resort to reverse engineering. Same goes to Nvidia by the way, but at least they seem to be a bit more competent in Linux/X.org driver development.

This whole argument is just a big excuse. We don't want excuses, we want some damn drivers.

--- someone who's been buying Nvidia since he realized that ATI doesn't work as well on Linux.

Re:what a joke

[harrypelles]

Amen, brother.

Re:what a joke

[sammy baby]

People aren't bitching that the drivers don't work, people are bitching because they aren't allowed to improve them.

Well... I mean, I'm bitching that they don't work.

This whole argument is just a big excuse. We don't want excuses, we want some damn drivers.

Yes. Personally, I don't care who writes them, as long as they're functional and not encumbered by something redolent of evil.

Re:what a joke

[Kadin2048]

The hardware developers somehow see this need to provide the software themselves, instead of taking advantage of the community, but then go and do a shoddy job of it.

Bingo.

When hardware companies try to make software, the result is almost inevitably shit. There are some exceptions, but big hardware companies tend to see software development as a 'cost center,' an afterthought to be minimized as much as possible, rather than a critical and major part of their product.

Look at scanners if you want. I've used some great film scanners in the past; brilliant hardware engineering, but coupled with the absolutely shit software that came in the box with it, it was practically a doorstop. To get anything else done, you had to get VueScan or Silverfast -- addon software written by people for whom software is their primary focus.

Re:what a joke

[michrech]

When hardware companies try to make software, the result is almost inevitably shit. There are some exceptions, but big hardware companies tend to see software development as a 'cost center,' an afterthought to be minimized as much as possible, rather than a critical and major part of their product.

There is absolutely nothing wrong with this, SO LONG AS THEY OPEN UP THE DAMNED SPECS SO THE COMMUNITY CAN WORK WITH IT PROPERLY. I can only assume I'm not the only one who thinks so, either.

I'm sure "the community" would be quite content to have a sample available to them before the hardware comes out (so they can write a driver/software package for it) to help the hardware company. Hell, I'm sure the developers (and the community) would be HAPPY to PAY for the development hardware (don't most console game companies doe this?). I know I'd chip in where I could, and I very much doubt I'm not the only one.

Re:what a joke

[plague3106]

Well maybe they would be happy to do that as long as every ATI linux user agrees not to bother them when things don't work right or perform well.

Re:what a joke

[sanyacid]

The hardware developers somehow see this need to provide the software themselves, instead of taking advantage of the community, but then go and do a shoddy job of it. So what is the reason why won't they let community do their own drivers as well? Would hardware manufacturers lose something on that?

Re:what a joke

[revengebomber]

The community COULD do their own drivers, but the specs aren't available. Everything about how to interface with the card would have to be found via reverse engineering.

I'm fairly sure the reason the specs aren't open, is because it would disclose some "secrets" about how the companies optimize their cards.

Re:what a joke

[Wicko]

They don't need to, but they really should. I mean, does it not seem logical that the best people for the job are the ones that actually designed the hardware? While this seems hardly true, at the moment, it could be ATI's way of thinking.

Re:what a joke

[jimstapleton]

Or because they use the OpenGL and DirectX specs to develop their card interfaces, both of which cost money and NDAs to attain, and thus they can't release the specs without risk of violating some copyright or patent, setting the leagal war machines of two companies upon them.

Re:what a joke

[John Whitley]

When hardware companies try to make software, the result is almost inevitably shit. There are some exceptions, but big hardware companies tend to see software development as a 'cost center,' an afterthought to be minimized as much as possible, rather than a critical and major part of their product. Actually, even when the problem isn't one of a cost-center mentality, the technical staff and technical management in hardware companies often doesn't really understand software development from any of design, lifecycle, or team process viewpoints. I've seen this cultural assumption that software somehow just naturally derives from the existence of the hardware.. that because hardware design can be difficult, that software must be trivial. This blindness in turn costs the company money for two reasons.

The first is hardware engineers who don't see their work as providing clean, semantically sound "APIs" against which the software must operate. Worst case I've seen of this was a chip maker that had to scrap a set of new masks and tape-out again. Oops. We didn't need that 1,000,000 USD anyways.

The second, a generalization of the above, is that the product of a hardware company is almost always a synergy of hardware and software. This applies even when the hardware company isn't the producer of the software. E.g. a general purpose CPU design won't succeeed if it has flaws that make it difficult to compile correct or performant sofware; a video card without drivers and/or specifications is useless; and so forth.

Re:what a joke

[644bd346996]

Your sig is ironically on-topic. Lotus Notes is a horrible piece of software.

Re:what a joke

[jZnat]

It's also possible (and likely) that they are violating patents they don't have a license for, so giving the specs out might bring light to this. A stupid legal reason rather than a technical reason...

Re:what a joke

[westlake]

They don't need to do any work. All they need to do is open up the specs, and people will do all the work for them.

and when the non-Geek calls technical support or returns a card under warranty because he doesn't have a fully functional driver - who fields the call and pays the bill?

Re:what a joke

[Kadin2048]

Company 'N' produces chips. Some come out great when tested. Some have minor failures when tested. They use software to disable features of the hardware so that the broken bits don't fail since they're not being used. Then you have a premium card and a 'value' card that's cheaper. But sometimes stuff works and they disable it in software anyway so they can still sell 'value' cards and premium cards at different price points. If you were to look at the source, you'd see this and just re-enable those bits to see if they worked, and possibly get a 'better' card than the arbitrarily-disabled one you'd paid for.

I agree this is an issue. However, I think it's more of a marketing problem than anything. E.g.: Foo, Inc., makes a device called the Widget. The produce two versions; the Widget Jr. and the Widget Pro. The Pro model is advertised as performing x, y, and z; the Jr. only does x. If Foo, Inc. opens its hardware interface to outside developers, they just need to make it clear that even if the Widget Jr. can do y and z, this isn't a supported feature, just an accident. Might they lose a few sales here and there, to hackers who'll purchase the Jr model, with the intention of enabling the unsupported features in hardware? Sure. But most companies aren't kept alive by the 'hacker' market. People want support; they want to know that what they bought is going to work.

As long as you differentiate your produces in sales and advertising, and give software developers a way to sense what version of the hardware they're working with (so that they can enable the right set of features by default), I don't think that the dilemma you're talking about is world-ending.

Just to continue more generally: an example of hardware design that I like, would be older serial modems. They had a nice clear line between the hardware and the software, and they interacted over a well-known and widely-supported interface. The hardware guys could make the modem work however they wanted; but it was going to interface with the computer over RS232, take ASCII commands ("ATDT", etc.), and interface to the POTS system on the other end. It's not the greatest example -- the standard command set was just a de facto standard because of the popularity of Hayes-brand modems, and lots of models had different command strings, but you rarely had to worry about actual driver or firmware issues, like you might for a "Winmodem" PCI card today. (And of course, sometimes the modems couldn't agree on the communication protocol for the POTS side of the equation -- anyone remember K56flex vs X2?) But in general, by abstracting the hardware at a fairly high level, and using a standard communications interface, using different types of modems was easy and seldom required actual recoding. That's the goal that hardware designers ought to be aiming for.

Re:what a joke

[swv3752]

The whole reason both companies do not wish to open up is that it is the software that mostly differentiates the different card lines. When card A is merely Card B at a different clock speed. And Card C is a slower clock and fewer pipelines, who is going to buy card A when they could use card C and get virtually the same performance at half the cost. The community has already come up with overclocking utils. Think of what they could do with full specs?

Doubt me? Look at the default Linksys router Firmware and DD_WRT.

Re:what a joke

[jimicus]

There is absolutely nothing wrong with this, SO LONG AS THEY OPEN UP THE DAMNED SPECS SO THE COMMUNITY CAN WORK WITH IT PROPERLY. I can only assume I'm not the only one who thinks so, either.

You are correct. But nowhere in computing is the idea of technology taking second place to marketing more prevalent than high-end graphics cards. So you wind up with two cards which are virtually identical, but one is artificially crippled in the driver.

Or, alternatively, they're just so damn embarrassed of all the screwups in the hardware that they just cannot bring themselves to release public specs. I've got strong reason to believe the same is true with a lot of hardware.

Re:what a joke

[michrech]

The whole reason both companies do not wish to open up is that it is the software that mostly differentiates the different card lines. When card A is merely Card B at a different clock speed. And Card C is a slower clock and fewer pipelines, who is going to buy card A when they could use card C and get virtually the same performance at half the cost. The community has already come up with overclocking utils. Think of what they could do with full specs?

That isn't our problem, and it isn't something ATI/nVidia *needs* to do (nor do I think they *should* be doing it). They only do it because it's easier to create one card, then cripple it in it's software/onboard BIOS than it is to actually create two cards. This has several side effects that I don't like, including the creation of a "one size fits all" drive that, as nearly all of us have experienced, doesn't work as well as it could (Jack of all trades, master of none? I guess hardware companies skipped that lesson...) *dreaming of a day when hardware vendors will start giving two flying fucks about their customers again*

Doubt me? Look at the default Linksys router Firmware and DD_WRT.

Errr.. I don't understand what you are getting at here. I know the 3rd party firmware offers features that Linksys decided, for whatever reason, not to include. I run tomato on my own WRT54g. I don't do anything fancy with it, but I liked several of its features over the stock firmware (including the ability of it to be a wireless client for a collection of wired PC's).

Re:what a joke

[fistfullast33l]

many of their customers do not realize all of the work that goes into these drivers.

A PR shill or a shitty developer had to say that. The very first large programming assignment I did for my CS degree involved a CLI. I wrote it so that it handled all input as long as it was typed correctly. IF the user typed something wrong, had too much whitespace, or whatever, the app crashed with a seg fault. But it worked perfectly if the user typed the right input. My professor flunked me. I told him that it worked perfectly if the input was correct, and that it took me so long to write (at the time, a week of pure programming was a lot of work). My professor said, when you get a real job, people are not going to care how long it takes you to write something. They'll want it to work perfectly. No bugs and delievered as quick as possible. A truer statement cannot be said. If an engineering organization hasn't learned that as long as ATI has been in business, then they need to go into a different line of work. That's why they get no pity from their users, and in my opinion, they don't deserve any.

Re:what a joke

[Hal_Porter]

Nvdia said something about "doing object orientation in hardware - it's very neat" as a reason for keeping the code secret. Maybe it's bullshit, but I can imagine something like that. Consider, the graphics card implements a bunch of interfaces, a contigous bunch of hardware registers analogous to a C++ interface.

The hardware contains information to say which interface is present at which offset. You could use GUIDs, COM style but these are probably overkill - a simple table of integers would be ok, e.g. 0x01 for Framebuffer, 0x02 for Cursor, 0x03 for BitBlt, 0x04 for TriangleRenderer.

So the code which tries to set up a framebuffer can search for a Framebuffer interface, write in the timing values and it's good to go. Or the code which handles Cursors or BitBlts can search for the interface. If the card has 3D, it will have some version of the TriangleRenderer interface.

The cool thing is that a universal driver becomes quite natural - once some cards support and interface, the software needs to support it to. But if you run the code on a board which doesn't support an interface you can just ignore that.

In a Windows driver incidentally (at least up to Win2k&XP), this is quite easy to handle. The minimum functionality is a framebuffer, any hardware you support above that is done by passing function pointers back to Windows at init time. If you provide an implementation of hardware cursors for example, it will be used but otherwise the GDI will use a software cursor. The designer can choose to only accelerate the most common cases of cursors and so on too - the function can 'punt' stuff back to the GDI if it can't handle it.

Re:what a joke

[Xlucid]

You implying that this situation doesn't exist at the moment? If it does exist at the moment, then whoever currently pays the bill would continue to.

If it doesn't exist at the moment, this whole article and thread would not have arisen.

Nvidia is not the competition

[crush]

The article is a long excuse explaining why AMD/ATI are unable to release decent GNU/Linux drivers. That's interesting enough as far as it goes: AMD/ATI and Nvidia both have crap closed, proprietary drivers which don't work well, make kernel updgrading difficult and are unauditable for security. So why bother with them? Further ATI have a history of dragging their ass and blocking the release of Free drivers,

Why bother with this crap? Just get an Intel GMA X3000 integrated motherboard and save time, power, money and hassle due to Intel "getting it" and releasing Open Source drivers and full specs. (You'll probably also be able to benefit from their free wireless drivers.

If you're into hardcore gaming then you're probably running a PS3 or an Xbox on the side anyway.

Re:Nvidia is not the competition

[twoboxen]

Intel needs to devote some extra resources to proper development of a dedicated video option. Integrated can never fully meet many customer-groups' needs. Additionally, the way they own the processor market they can afford the risk to branch into this sector. AMD is competition, yes, but hardly a threat.

Re:Nvidia is not the competition

[PenisLands]

Free wireless drivers? The wireless in my laptop works with linux, but according to Ubuntu's driver management thing, it requires proprietary drivers. So, maybe you are mistaken... or are only some of intel's wireless adaptors free and open?

Re:Nvidia is not the competition

[dc29A]

If you're into hardcore gaming then you're probably running a PS3 or an Xbox on the side anyway.

Let me know when good RTS, MMO and strategy games come to consoles. Until then, I'll stick with my PC!

Re:Nvidia is not the competition

[oliverthered]

I don't think you should be looking at ATI or NVidia for the changes in the kernel that break the ABI.

The kernel is intentionally difficult to upgrade or use for people who want to use binary drivers.

Re:Nvidia is not the competition

[the_humeister]

Or you could use an R100, R200, or R300 based ATI card. They're not hard to find, relatively inexpensive, and still powerful enough for a casual gamer (at least R300s are, possibly R200). Oh, forgot to mention that they have mostly full open source drivers written already.

Re:Nvidia is not the competition

[crush]

The kernel will always be a problem for closed-source software that stupidly relies on a stable ABI. The kernel intentionally stays flexible and changes rapidly in order to keep innovating. FL/OSS software has little problem rebuilding against it and staying innovative. That's why ATI and Nvidia will never be able to produce satisfactory drivers for Linux.

That's exactly why it makes most sense to go with fully open hardware supported by FL/OSS drivers unless you want to either stick with old kernel versions and not benefit from improved security and functionality, or else you have some hardcore gaming or 3D-modelling need.

Re:Nvidia is not the competition

[crush]

Only some. On the newer motherboards. See here. Make your life easier and more productive. Sell the crap hardware from proprietary companies that haven't seen the new market conditions before it's too late.

Re:Nvidia is not the competition

[crush]

Sure. Run Windows and play those games on it, but that's irrelevant to a discussion of Linux drivers.

Re:Nvidia is not the competition

[plague3106]

Ya, because building software against a stable API is SOO boring! We don't care about getting things done, its how fast we can change it!

Re:Nvidia is not the competition

[crush]

Yes, that's another good option. Similarly the nouveau project for reverse-engineering Nvidia's closed, proprietary, probably-infringing-patents hardware is coming along nicely. But ... if buying new hardware then the chance for the market to reward an OPEN piece of hardware and simultaneously save on power is too good an opportunity to miss. Intel are really doing the right thing right now and it would be good to see the market confirm their strategy.

Re:Nvidia is not the competition

[cortana]

None of them are fully free and open--there is always some combination of non-free, binary firmware and/or a regulatory compliance daemon of some kind involved.

Re:Nvidia is not the competition

[moro_666]

I second that. I'm a devoted fan of my nvidia powered laptop, unlike the integrated intel laptop that i had before, this doesn't stall at weird moments nor do i lag behind the graphics. The dedicated graphics card and it's private memory kick ass already in the usability. I have chosen my next laptop to be ati powered, just to see the difference (amd guys, fix that driver, now !).

  And as it may come as a surprise to the thread starter, some people actually do play games with their laptops while on the move, sometimes just to relax, sometimes to beat the highscore of a friend in a public 2D flash game. You don't drag your Xbox on with you everywhere you go. Laptops are portable and therefor there when you need them, but this doesn't mean that they have to lag as hell on an integrated intel chip. Also keep in mind that there's other stuff than games that needs the power of a proper graphics chip with loads of ram.

  Thumbs up for intel going public with their drivers, but totally thumbs down on their hardware. Dudes, duuuudes, seriously, you're not even near to nvidia/ati when you have kde running with 25 windows open. Just a scroll somewhere will make the intel cough blood while the ati or nvidia analogue just kicks on like nothing ever happened.

  Lagging behind graphics will cut your productivity. Even if it's open source lag.

Re:Nvidia is not the competition

[crush]

Not true anymore. There is one blob of FCC regulatory compliance stuff that used to be in userspace, but has moved out of it. Intel have been busy opening up all the details on all this stuff to hackers.

Re:Nvidia is not the competition

[Mad Merlin]

You think you want a stable kernel ABI, but you really don't.

Re:Nvidia is not the competition

[DaveWick79]

Also keep in mind that there's other stuff than games that needs the power of a proper graphics chip with loads of ram.
Unless you're doing 3D CAD on your laptop I don't see any other application of this. And last I heard most if not all of the major 3D CAD softwares don't run natively on Linux.
I don't ever expect Intel to take over the graphics market - they simply want something they can use for cheap onboard use that doesn't require exorbitant licensing fees, and does a decent job for the average user.

Re:Nvidia is not the competition

[bcrowell]

Well, if all you want is onboard, 2-d graphics support, then pretty much anything on the market will work just fine with Linux, so that's not a strong technical/practical reason to buy Intel. OTOH, if you want 3-d video for gaming, then Intel isn't an option. Of course, you might want to buy an Intel mobo just because you want to support OSS-friendly hardware on moral/political grounds.

Re:Nvidia is not the competition

[cortana]

Forgive me but these drivers still seem to require non-free firmware.

Re:Nvidia is not the competition

[twoboxen]

Intel is already beginning to TRY to take on the graphics market (see: Santa Rosa and beyond). If ever they had motivation and/or a reason to do so it would be if one of their principal competitors merged/purchased a major graphics company. Yeah, like THAT would ever happen! ... O, wait.

Re:Nvidia is not the competition

[niko9]

Why bother with this crap? Just get an Intel GMA X3000 integrated motherboard and save time, power, money and hassle due to Intel "getting it" and releasing Open Source drivers and full specs. (You'll probably also be able to benefit from their free wireless drivers.

I agree. I just wish that little Intel GMA X3000 chip was available on a separate PCI-Express card. My Athlon socket 939 motherboard suites me just fine for my computing needs. Would be a shame to have to buy a new motherboard, CPU, ram (DDR2 vs DDR), just to get free drivers for video. I know Intel has plans for some type of discreet video cards, but no has from their camp has stated that these cards will be supported as freely as their current offerings.

Re:Nvidia is not the competition

[arodland]

Doesn't apply to 2100, 2200, or 2915, which means there's what, three people who would actually benefit from this? The rest still have to deal with firmware licenses. :)

Re:Nvidia is not the competition

[mhall119]

They're talking about ABI, not API. The Linux kernel API is extremely stable.

Re:Nvidia is not the competition

[PeterBrett]

And last I heard most if not all of the major 3D CAD softwares don't run natively on Linux.

Cambridge University Engineering Dept does all 3D CAD teaching using Pro Engineer. On Knoppix boxes with NVidia graphics hardware. It runs very fast, even with moderately complex models.

Re:Nvidia is not the competition

[plague3106]

Nice. The only 'solution' that article provides is for you to give up your driver to someone else. Excellent idea. Sounds really 'free' to me.

Of course this is handled in the .Net world rather nicely, and you don't even need to worry about some developer that can't bother to read the documentation to figure out which interface is obsolete. (Which is apparently a very big problem in the linux driver development).

Personally if I were ATI (or NVidia), I'd simply abandon support for Linux completely, since it seems as the changes in Linux are there to intentially hinder a company that doesn't have a 'compatible' mindset. Hmm, I wonder where I've heard that complain before..

Re:Nvidia is not the competition

[ceswiedler]

Not for drivers, it's not. The Linux userspace API is extremely stable, and the ABI somewhat less so. But anything inside the kernel can change drastically. That's why Linux hackers want all drivers in the kernel tree, so they can find anything which breaks due to an API change, and fix the problems.

Re:Nvidia is not the competition

[jZnat]

The Fire Emblem series are great strategy [RPG] games. Available on Famicom, Super Famicom, GBA, Gamecube, and Wii.

Since you already mentioned RTS, I figured you meant something else by "strategy games". Check them out.

Re:Nvidia is not the competition

[mhall119]

Drivers are programmed against the kernel API, and as such will not need to be rewritten unless the kernel API changes. That said, drivers are COMPILED against the kernel ABI, which means that whenever the ABI changes, the drivers have to be recompiled.

Linux hackers want driver code in the kernel tree so that they will be automatically recompiled against the new kernel, instead of the user having to download and compile the driver source every time they get a new kernel. They also want them in the kernel tree because so many drivers require the same functionality, and it's better to re-use and improve common code, rather than each driver containing their own implementation.

Re:Nvidia is not the competition

[MROD]

Except when the kernel driver is ophaned, such as happened with the Advansys SCSI host adaptor. All that happened was that it was removed from the standard build configuration and the code didn't compile anymore.

The problem with a kernel driver interface which changes between major versions is the problem of having to udate huge numbers of drivers AT THE SAME TIME possibly adding typographical bugs in that codebase. The more you have to change, the more likely such errors will creep in.

I'm not saying never change the kernel driver interface but only change it with major kernel version change.

(And no, I don't buy the arguments in the document referenced in the grandparent posting.)

Re:Nvidia is not the competition

[Creepy]

almost all high end and mid-tier OpenGL based CAD packages that I have contact with support Linux now (UGS, Pro-E, AutoCAD).

Re:Nvidia is not the competition

[psavo]

You know, people should just quit with this R100-R300 fucktarded bullshit.

ATI lost all the goodwill generated by R200 opened specs when they withdrew those opened specs. They even try and slow any progress done with R500 support (look in David Airlie livejournal blog). If they don't give full HW specs, then just fucking screw them. Don't buy, don't recommend, just forget they exist. Open-sourced thing, although it has working (albeit much slower that fglrx) 3D, does not have tv-out support.

And on top of that, whatever they (ATI) offer is utter and unforgivable CRAP. It crashes and it just plain does NOT WORK (Tv-out, multiple heads, FUCKING SLOW).

Granted, nvidia hasn't done even that partial HW-opening-up (unless one counts the TNT2-obfuscated code for something, AFAIR it was a one-off affair).

And yes I write this on a computer with Radeon 9800XT. Whatever I buy next will have (in order of preference) Intel or Nvidia graphics. Screw ATI, no value, no money.

Re:Nvidia is not the competition

[the Hewster]

Why bother with this crap? Just get an Intel GMA X3000 integrated motherboard [wikipedia.org] and save time, power, money and hassle due to Intel "getting it" and releasing Open Source drivers and full specs. (You'll probably also be able to benefit from their free wireless drivers.

I tend to agree. ATI is one of the worst offenders by actively blocking open source development (still no open source r500 2D drivers just before the r600 comes out) _AND_ having crap proprietary drivers (no compositing support, low performance). I recommend the same: buy an Intel motherboard with integrated graphics an sign my (shameless plug) open3d pledge. There are 100 signatures and 900 to go!

Re:Nvidia is not the competition

[TheRaven64]

Hardly anyone is asking for more than royalty-free distribution rights on firmware. It doesn't run in kernel space, or even on the host CPU, so there's no difference between non-Free firmware and hardware for which you don't have the HDL.

Re:Nvidia is not the competition

[Mad Merlin]

Of course this is handled in the .Net world rather nicely, and you don't even need to worry about some developer that can't bother to read the documentation to figure out which interface is obsolete. (Which is apparently a very big problem in the linux driver development).

.Net is userspace, and the linked document is not concerned with userspace. The userspace visible kernel APIs have been stable since before Linux 1.0, as the document points out.

In Linux, obsolete internal kernel interfaces are removed when they become obsolete, and anything using the obsolete interface is ported to the new interface by the authors of the new interface. I'm sure you'll agree that it's difficult to accidentally use an interface that no longer exists.

Now, for a proprietary kernel that needs to retain binary compatibility with every driver ever developed for it, simply because the old drivers cannot or will not be fixed, it becomes exceedingly easy to accidentally use an obsolete interface that's hanging around just to support some cruft from the mid 90s.

Re:Nvidia is not the competition

[DaveWick79]

At last check, AutoCAD still does not support Linux and it seems that Autodesk has no plans to. Same for Solidworks. I've not seen an engineering or design shop using anything but those two in a long time.
I guess I haven't looked in a couple of years, it looks like a few have started marketing Linux versions of their software in the past 2 years. Kudos to them.

Re:Nvidia is not the competition

[DaveWick79]

Also while UGS NX supports linux now, SolidEdge still does not.

Re:Nvidia is not the competition

[kscguru]

That's why Linux hackers want all drivers in the kernel tree, so they can find anything which breaks due to an API change, and fix the problems.

And that's exactly why hardware manufacturers DON'T want their drivers to live in the kernel tree. They don't get:

• Bug fix backports: Kernel hackers apply a patch to the latest vanilla kernel, then say backports are the distro's problem. Two distros do, fifteen don't, users of those fifteen scream at the hardware companies for having buggy drivers. With an out-of-tree driver, you just need to update the driver. Windows and OSX are light-years ahead of Linux here.
• Community contributions: Kernel hackers only add GPLed code; very few even allow dual-license, and certain vocal hackers are zealously GPL-only (to the extent of rewriting code JUST to make it GPLed). Which means the hardware vendor can't take a fix from a linux driver and integrate it into a Windows driver without GPLing that too. The vendor has to either GPL all drivers (Windows included), or maintain two separate trees (GPL and non-GPL), or not open drivers at all. Guess which is easiest / cheapest?
• Driver API stability: a modern 3D graphics driver is a full OS in its own right, with internal threading models, schedulers, memory management, context switches, etc.; a modern driver needs more than just bugfixes. Every good developer knows the way to keep two large codebases manageable is a stable API between them; the only people who don't seem to get this are otherwise-intelligent Linux kernel hackers.
• Kernel API freedom: Kernel hackers like stable userspace APIs (for good reason). But hardware vendors don't need to provide stable APIs if they have a shim library that actually talks to the cards (e.g. atioglxx.dll, the ATI OpenGL implementation). It's a lot easier to let the API change rapidly and only commit to a stable API at the library interface (the OpenGL API).
• Easier work. The Linux kernel development process is optimized for making the kernel hacker's life easier at the expense of the driver developer's (hint: saying "we'll update your driver for you" clashes very badly when the HW vendor is simultaneously making changes). If kernel hackers want to see better device drivers, they need to stop treating drivers as second-class citizens. Microsoft is very good at courting driver developers; Linux is the definition of arrogance.

Re:Nvidia is not the competition

[jhol13]

And, most importantly, they do not get testing. Those who change API/ABI cannot test.

Re:Nvidia is not the competition

[jhol13]

Thank god OS X, Windows, Solaris, etc. developers can innovate without breaking everything in every change.
Thank god ATI and NVidia test their drivers instead of giving source to those who cannot.
If ABI/API changes too often it means decreased security, not improved. Security fixes extremely rarely need to change ABI. Most importantly, stable ABI/API means increased reliability.

Re:Nvidia is not the competition

[oliverthered]

There's no reason that they should change the ABI between major releases breaking compatibility with the code left over from the 90s, just like very other major os in the world

Re:Nvidia is not the competition

[oliverthered]

your right I don't think I want a stable ABI, I know I want a stable ABI.

That document assumes that the only way to get a stable driver (personally I don't care if the drivers flaky so long as it works 90% of the time) is to put it in the kernel tree.

The only reason I can see that you couldn't produce a stable driver outside of the kernel tree is that the kernel API/ABI keeps changing all the time. That's not a driver stability problem it's a kernel stability problem.

Re:Nvidia is not the competition

[oliverthered]

which box, certainly not the box the hardware comes in because they all seem to be Windows and Mac only.

Re:Nvidia is not the competition

[mhall119]

which box, certainly not the box the hardware comes in because they all seem to be Windows and Mac only. Here's a fun little game, try running Windows on a PPC Mac box, and Mac OS X on a Windows box. Guess what happens? Now try Linux on either box, guess what happens?

Re:Nvidia is not the competition

[mhall119]

That's why Linux hackers want all drivers in the kernel tree, so they can find anything which breaks due to an API change, and fix the problems.

And that's exactly why hardware manufacturers DON'T want their drivers to live in the kernel tree. They don't get:

• Bug fix backports: Kernel hackers apply a patch to the latest vanilla kernel, then say backports are the distro's problem. Two distros do, fifteen don't, users of those fifteen scream at the hardware companies for having buggy drivers. With an out-of-tree driver, you just need to update the driver. Windows and OSX are light-years ahead of Linux here.

I don't think I've ever heard of someone blaming the hardware company when there was a patch already available, just because their distro of choice hadn't incorporated that patch. Is this a straw man, or have you actually encountered it? Even so, how would having an open-source driver prevent them from providing patched binaries directly to the user?

Community contributions: Kernel hackers only add GPLed code; very few even allow dual-license, and certain vocal hackers are zealously GPL-only (to the extent of rewriting code JUST to make it GPLed). Which means the hardware vendor can't take a fix from a linux driver and integrate it into a Windows driver without GPLing that too. The vendor has to either GPL all drivers (Windows included), or maintain two separate trees (GPL and non-GPL), or not open drivers at all. Guess which is easiest / cheapest?

Code added to the Linux kernel MUST be GPL, that's the licensing requirement. However, contributors are still free to release their code under other, even proprietary licenses. There is nothing to stop nVidia from providing GPL and non-GPL drivers that share code. If someone makes a patch to their driver, and releases it only under the GPL, then of course it can't be re-licensed without permission. However, if there is no GPL code, they don't get any user submitted patches anyway, so they don't gain anything with non-GPL. An alternative would be to foster development on a BSD-style license, or ask for a copyright grant from the contributor so they can relicense it.

Driver API stability: a modern 3D graphics driver is a full OS in its own right, with internal threading models, schedulers, memory management, context switches, etc.; a modern driver needs more than just bugfixes. Every good developer knows the way to keep two large codebases manageable is a stable API between them; the only people who don't seem to get this are otherwise-intelligent Linux kernel hackers.

In Linux, drivers are a part of the Kernel, not outside programs. It is, in a sense, a single large codebase, not multiple large code bases. nVidia wants to be part of that codebase, without actually contributing itself to that codebase.

Kernel API freedom: Kernel hackers like stable userspace APIs (for good reason). But hardware vendors don't need to provide stable APIs if they have a shim library that actually talks to the cards (e.g. atioglxx.dll, the ATI OpenGL implementation). It's a lot easier to let the API change rapidly and only commit to a stable API at the library interface (the OpenGL API).

It's even easier to let those people who are changing the API make the changes for you, which is what the Kernel devs are asking for.

Easier work. The Linux kernel development process is optimized for making the kernel hacker's life easier at the expense of the driver developer's (hint: saying "we'll update your driver for you" clashes very badly when the HW vendor is simultaneously making changes). If kernel hackers want to see better device drivers, they need to stop treating drivers as second-class citizens. Microsoft is very good at courting driver developers; Linux is the definition of arrogance.

Re:Nvidia is not the competition

[kscguru]

I have other replies, but only one point here is really important.

Driver API stability: a modern 3D graphics driver is a full OS in its own right, with internal threading models, schedulers, memory management, context switches, etc.; a modern driver needs more than just bugfixes. Every good developer knows the way to keep two large codebases manageable is a stable API between them; the only people who don't seem to get this are otherwise-intelligent Linux kernel hackers [kroah.com].

In Linux, drivers are a part of the Kernel, not outside programs. It is, in a sense, a single large codebase, not multiple large code bases. nVidia wants to be part of that codebase, without actually contributing itself to that codebase.

Fundamentally, totally, and irreconcilably wrong. (With respect to various kernel devs, who disagree). Drivers as part of the kernel work if - and only if! - the driver is trivial. 80% of the drivers out there are trivial - bus drivers, serial ports, chipsets, cheap network cards, VGA, generic USB devices. 20% are not trivial - 3D graphics, enterprise-class network cards (the ones with TSO, checksum offloading, huge ring buffers, etc.), VMware/Xen/KVM, advanced filesystems (most of which exist in FUSE). The key thing all these nontrivial drivers share is a large API with other components (usually user-level) - advanced filesystems expose other types of metadata, virtualization drivers expose a whole virtual CPU, 3D drivers expose an entire array of GPUs plus a huge texture memory, and all the APIs to control it.

The difference between a trivial driver and a non-trivial driver is the stability of the API the driver provides. Most drivers present in Linux today (including the misleading "supports more devices than any other OS" mantra) have a stable, simple API, and only need bugfixes. The drivers Linux is conspicuously lacking (i.e. the drivers that are present in Windows, Solaris, sometimes Mac, but not in Linux) have complex, evolving, independently-versioned APIs - APIs with huge dependencies on a codebase that exists outside the Linux kernel, code that is under active development even after the hardware ships. It's fine that kernel devs would happily update the APIs between a driver and the kernel - who is going to update the APIs between the driver and the codebase that uses the driver? For all the griping, the kernel-driver API doesn't change that much ... the driver-userlevel API changes much more rapidly for all the drivers I named above, and keeping that up-to-date is harder than maintaining out-of-tree drivers. And thus, the driver logically belongs in the codebase to which it is more tightly coupled - the companies' codebase, NOT the kernel's codebase. Moving the driver to the kernel makes the kernel dev's life easier, but the companies' life much harder.

I'll summarize: Linux does not have complex drivers for complex devices (like 3D graphics) because the Linux kernel is not set up to interface with complex drivers. The Linux kernel assumes drivers need only bugfixes or integration with new kernel APIs; complex drivers also need integration with user-level APIs, and the Linux kernel development model makes such integration extremely difficult to impossible. The real Linux device support mantra is this: "Linux supports more trivial devices than any other OS, but can't handle non-trivial devices."

I don't particularly blame Linux - it evolved as a Unix replacement in a world where there were no complex drivers, where the drivers-are-in-kernel-tree convention worked. Now Linux is trying to jump to the modern desktop world where there are complex, out-of-tree drivers. If Linux wants appeal outside of the Unix-replacement crowd, Linux will have to grow out of this drivers-in-kernel-tree myopia.

Allow out-of-tree drivers, and all the other problems in the original post disappear.

Branding problem..."FireGL?"

[xxxJonBoyxxx]

Now, I don't pay much attention to video cards, but when I saw "FireGL" I thought "why would you need to optimize OpenGL graphics for Firefox?".

ATI and Fedora 7 / X.Org 1.3

[harrypelles]

I made the same mistake as many Fedora users - jumping (to Fedora 7) before looking. I'm not poking at Fedora here, on the contrary, I am a loyal Fedora user. It's ATI I'm upset with. ATI released a new fglrx driver (version 8.37) since Fedora 7's tests and final release that also does not work with X.Org 1.3. We're all sitting around waiting for the 8.38 which ATI claims will be compatible. And don't even get me started on ATI's absent AIGLX support for Linux. My next card will nVidia.

Re:ATI and Fedora 7 / X.Org 1.3

[pembo13]

Yah well, Nvidia is the lesser of the two evils... learned that before I left Windows. Talk with your dollars, it's the only language these companies understand.

Re:ATI and Fedora 7 / X.Org 1.3

[El Lobo]

Interesting. People on /. people use to bash Vista for the same thing. The difference is that here Vista is to blame, while on Linuzz everyobody magically undarstand that this is a driver's problem. Not a troll, BTW, just talking to myself.

Re:ATI and Fedora 7 / X.Org 1.3

[drgonzo59]

The difference is that here Vista is to blame, while on Linuzz everyobody magically undarstand that this is a driver's problem.

Perhaps because 'Linuzz' is open it is easy to see where the problem lies. With Vista you get huge binary blob and if it's broken you don't know if it is the drivers or Vista -- you can't debug it and look at the source so you call MS tech support and wait 6 months for a service pack or MS tells you to call ATI/AMD and you wait 6 month for a fix. Binary drivers suck that's the problem here...

Re:ATI and Fedora 7 / X.Org 1.3

[DoofusOfDeath]

And don't even get me started on ATI's absent AIGLX support for Linux.

OK, you talked me out of it.

Nothing to see here.

[BESTouff]

Basically, TFA says that "ATI has a release cycle". They even have an unofficial bugzilla and an unofficial wiki. Oh, and they'll drop R200 support too. And all that's supposed to make better drivers for Linux one day. I really wish they'd go the Intel way: hire some top-notch developers, give them specs and make them do Free drivers.

Three years of problems

[Tribbin]

I've had three nvidia videocards (one that is onboard) and one ati.

I've always had, and still have problems with them when I use the proprietary drivers under linux (Ubuntu/Debian/Slackware, both packed and from the nvidia site). A few months back I just gave up and stopped blaming it on the videocards and drivers since I seemed to be the only person whose screen froze up upon switching to terminal mode and back.

The onboard videocard gives the same problems and I have exactly zero problems when using the slow open source drivers from x.org. That's why I excluded the posibility of my motherboard being broken.

Is there anybody; a single person who had the same problems or could anybody smarter than me tell me what I overlook?

Re:Three years of problems

[lp_bugman]

I had huge problems with all my AGP nvidia cards. The video use freeze. After a couple of minutes using X in my AMD Athlon 64 with a VIA chipset. The problem was VIA implementation of AGP. There is a setting in the closed source drivers to disable the NV AGP implementation.

Section "Device"
  Driver "nvidia"
  Option "NvAGP" "0"
EndSection

Hope it works for you :).

Re:Three years of problems

[metamatic]

I have a ThinkPad T42p with a FireGL T2 in it. With the slow open source drivers, everything's fine. With the closed ATI drivers, textures don't load properly, and eventually the machine locks up. ATI's drivers suck. Their OS X drivers suck too, the OpenGL rendering of antialiased lines and polygons is utterly broken on my Mac with ATI card. I'm planning to buy a new Mac soon, and I'm currently planning on getting a MacBook rather than a MacBook Pro specifically to avoid the ATI hardware in the Pro. I'd rather have Intel integrated graphics that work, than fast ATI graphics that don't. (Opinions mine, not IBM's.)

Re:Three years of problems

[ThatFunkyMunki]

You should try using your Distro-packaged nVidia drivers. Normally, they've got scripts and other mechanisms to install things that the official packages from nVidia don't have.

Re:Three years of problems

[digitalhermit]

I've noticed this with some Nvidia cards and it has always been solved by disabling the framebuffer mode for text console. This kills the high-res console, but at least makes it stable.

Re:Three years of problems

[overlordmead]

Hear, Hear.

I have similar issues that choke my ubuntu home server up. onboard hypertransport(AMD64) combined with software cpu throttling guarantee a complete system lock(no Alt-SysRq) in a few minutes. I can get further with software cpu throttling disabled, but switching to a terminal is almost certain to cause a complete system lock. Even with the hypertransport disabled ATI binaries cause the system to lockup in a similar fashion. So far the only reliable configuration(it is a home server) has been a generic or x386 kernel with BIOS hypertransport disabled and vanilla vesa or ati X drivers. How many restarts, shitty FAQ pages, and binary driver versions( some that JUST DIDN'T WORK, 8.24 I'm looking at you!) did it take to figure that out? I don't want to think about it and with current setup everything works and I rarely have to. Next board is onboard Intel GFX.

I also seem to be the only person experiencing a bug with the ehci_usb module with cheap mp3 players? disable that module and a lower entity catches it as a flash drive, otherwise it kills my usb bus(all usb devices fail) and I have to alt-sysrq my way outta the situation.

Re:Three years of problems

[Workaphobia]

That's not the optimal solution, as that disables AGP altogether and gives you slower performance.

I've had stability problems with my AGP GeForce 6200 and VIA chipset for a long time. I'd get random crashes when doing anything remotely graphics intensive, always accompanied by nvidia error messages in my log. In the beginning I used to be able to still ssh into the system, or even switch to a virtual console for a little bit before I lost all system functionality, but later on I stopped trying and just used SysRQ to sync, unmount, and reboot my system.

Finally, it became frustrating enough for me to do something about it. I'd dabbled in the README and nvidia's forums before, but I looked longer and harder and isolated the problem to that setting fairly easily. It was embarassing that all that time all I needed to do was change my AGP settings; knowing nothing about the nature of the problem, it was previously all just a giant black box to me, and if it didn't work there was nothing I thought I could do about it.

Anyway, instead of getting rid of all AGP, you can set the option to 2 to use AGPGART instead of NVAGP. I've had no stability issues since then, and I still get reasonable performance for my crappy card. Of course agpgart has to be selected in the kernel as either a module or built in.

I don't think this problem is related to the grandparent's post though, because it occurs for him when he switches terminals, not at random while using hardware accelerated applications.

Re:Three years of problems

[Workaphobia]

Generally when you have problems involving the console, the first thing to consider is the framebuffer. Disable it in the kernel configuration or in your boot parameters to the kernel. As someone else said above, you'll get a less-pretty console, but if you can't trigger the bug then you'll be closer to solving the problem.

I had a similar issue, where switching to a framebuffered console after going to X would cause my consoles to become corrupted until I rebooted. The machine would function just fine except for displaying the console.

I remember reading that nvidia conflicts with one of the framebuffer drivers in the kernel to produce the corruption problem that I had, but I forgot the names of the good driver and the bad one. The good one that it works with is supposed to be more modern and support more features, and might have a T in the name. Like tgfb or something. You should be able to find it in your kernel configuration menu.

Good luck.

Help from open source?

[G3ckoG33k]

While I would love to help out in 3D graphics, but am in no way capable, I am also curious - how many competent 3D graphics hardware developers are there, really, outside of AMD, nVidia and a few other companies? What do they expect? A handful, dozens, hundreds, or even thousands of developers? Is it really that 'easy' so that they would benefit from from the Argus eye?

--

Misleading summary title?

[RingDev]

WTC does this have to do with AMD?

-Rick

Re:Misleading summary title?

[bouchecl]

WTC does this have to do with AMD?

ATI is now part of the big AMD family. Where have you been in the last few months?

Re:Misleading summary title?

[plague3106]

I dunno, maybe because AMD bought ATI?

Re:Misleading summary title?

[RingDev]

Yeah, in the same way that Mazda is a part of the big Ford company. But you wouldn't submit and article called "How Ford Builds Cars" and then cover only the Mazda factor's stereo installation.

-Rick

Re:Misleading summary title?

[tomstdenis]

ATI isn't a company anymore? AMD owns it, as in, AMD owns the IP to the cards, as in it's an AMD Radeon now.

OMG you can't be this stupid can you?

Tom

Re:Misleading summary title?

[plague3106]

Wow, you are dense aren't you? I don't know the answer to your first question, but the next two questions, the answer is YES. That's kinda what happens when you buy something, you take control & responsibility of it..

Re:Misleading summary title?

[KillerBob]

OMG you can't be this stupid can you?

Never ask that question. You know the answer, I know the answer, and he probably knows the answer... the bitch of it is, if you bring to light just how stupid he really is, somebody's going to take steps to correct it, and somebody's going to be arrogant enough to come up with an "idiot-proof" designation. And of course, you know, the minute something gets described as "idiot-proof" they come out with a better idiot.

Re:Misleading summary title?

[SIR_Taco]

AMD bought ATI dumbass

It's like one big acronym family ;)

(it's a joke... don't take it to heart)

Re:Misleading summary title?

[RingDev]

Atleast there is still room for improvement. Let me know when you buy an AMD brand graphics card, I'd be interested in hearing about it.

-Rick

Re:Misleading summary title?

[jonasj]

Yeah, in the same way that Mazda is a part of the big Ford company. But you wouldn't submit and article called "How Ford Builds Cars" and then cover only the Mazda factor's stereo installation.

Not at all in the same way. ATI is just an AMD brand now. Compare http://www.mazda.com/ to http://www.ati.com/ and you will see why the article's title is perfectly valid.

Re:Misleading summary title?

[plague3106]

You'd be guessing wrong. "Upper echelon" at ATI isn't really upper echelon if they must now answer to another corporate structure. In other words, they are irrelevent. There's no reason for AMD (or any company) to buy ATI and not start merging them into the fold.

AMD slapping their logo on the graphics card boxes is the first step in getting their name synonymous with graphics. It doesn't take as much as you think. For Westwood, I think it was one title, and the next title they produced was under the EA brand (Generals).

Re:Misleading summary title?

[Xlucid]

Ignorance != Stupidity

Where we're at

[TheGreatOrangePeel]

I have to say I'm not finding anything insightful in the linked article. It's just a long winded way of saying, "The drivers aren't very good, but AMD/ATI is working hard on it." which we most of us likely already knew. It would have been good to see some insights on what AMD did to improve the driver development process, what impact the open source announcement made, etc.

Translation: ATI fails to release OSS drivers

[dwheeler]

Wow, a lot of text to paper over the obvious problem: ATI is still failing to release OSS drivers for Linux. The paper describes all the ways that ATI tries to avoid releasing the drivers, and how they all fail to solve the problem. ATI has testing processes, etc., sure - but later on, when X.org and the Linux kernel change, there's no way for me to update the driver - so I have to hope that they will EVENTUALLY do so (leaving me vulnerable to any security problems) OR throw away the ATI card.

Dell has solved this problem by including the Intel stuff instead for their Linux offering. It's time for ATI to release their drivers as OSS.

Does AMD just not yet it?

[ArcRiley]

I think I speak here for a fairly wide swath of GNU/Linux developers and distributors: While the quality of the driver and it's release is of some importance, the license of that driver is the deal breaker. Give us a poor driver with a free (as in freedom) license and the community will make it great. Give us an excellent driver with a proprietary license and only a minority of users will use it. Why? Many distros (I'll use Fedora as an example) will not package proprietary drivers. Ubuntu, which I believe is the most popular right now, is on the fence and, at the very least, warns the user. This is part of the fundamental nature of GNU/Linux: It's about freedom. Does AMD understand this yet? It doesn't seem so. Moreso, because most free software developers ignore ATI's drivers as proprietary, there is little GNU/Linux software that takes advantage of the higher end cards. The DRI drivers on an r200/r300 card work just fine for almost everything. Why would a user, then, pay $100+ for a higher-end video card when a $35 Radeon 9250 is better supported? For users it's thus a choice of price/benefit if nothing else. If AMD wants to work better with our community they need to join our community. Break the closed development loop in favor of integrating their paid developers and "volunteer" driver developers, there's a number of skilled developers with DRI that I'm sure would be very willing to help should AMD do this.

Re:Does AMD just not yet it?

[mrtom852]

That article took ages to get approval from the management at ATi (~two months) - if they're that paranoid about releasing information on their release cycle then I don't think we stand a chance of getting any open source/specs from them.

apologist article

[lib3rtarian]

The article basically says "thanks for these power point slides, AMD/ATI, I'll kiss your F**** asses in my article" Seriously, that article sucked.
Also, when they say that customers don't realize how much work goes into drivers, is that an excuse? I don't care how much work goes into drivers, I know it's hard to do. It's hard to develop the cards to begin with, and to engineer them. The entire process is hard and full of work. The bottom line is that if you can't produce working drivers for a product that you created and manufacture and sell, that you are in the wrong business and wasting my time.

The wonders of summarizing summaries

[jd]

"The early ATI Linux driver had lacked essential functionality ... not to mention a great deal of bugs."

I'd have thought lacking a great many bugs would be a good thing. (Yes, I know what was meant, but it's monday morning, I'm at work and I feel like pulling the legs off the English language.)

Intel has not release docs

[dmoore]

> Intel "getting it" and releasing Open Source drivers and full specs.

Actually, Intel has not released docs for their GMA X3000. Their current stance is that the driver is the documentation. That's fine and good, except the driver is still very incomplete (missing OpenGL features, no XvMC, no tv-out, etc.). See here:

http://lists.freedesktop.org/archives/xorg/2007-Ma y/024582.html

Re:Intel has not release docs

[crush]

That's interesting. Thanks for the link. Is there some productive way to pressure Intel and help them make the final step?

TFA...

[Dancindan84]

TFA mostly talks about the ATI/AMD development cycle and how it should/will improve the driver release timing and quality. -Some- mention of that in the post would have been nice, rather than essentially copy/pasting from the first page.

Linux or Open-Source drivers?

[mi]

It seems, most of the critics of the closed-source drivers happily shut up, when they get drivers for their platform.

Manufacturers have learned this long ago — they release binary drivers for Linux/i386, and the criticism all but disappears. NVidia has gone farther than most by releasing Linux/amd64 and even FreeBSD/i386 binaries.

But FreeBSD/amd64 is not there... Nor are Open|NetBSD... Nor Linux/ppc.

I know, each additional platform costs plenty. But it is the source, I'm asking, not binaries. If — as the case may be — the most modern version of NVidia drivers can't be built on FreeBSD/amd64 due to feature FOO missing on the platform, I could #ifdef that feature out myself. Earlier versions of the driver weren't using FOO even on i386...

Re:Linux or Open-Source drivers?

[Anne Thwacks]

NVidia may have released FreeBSD/i386 drivers, but has anyone actually got the damned things to work?

What the hell use are drivers that don't actually work?

I dont give a stuff if the drivers are binary closed source, if they work. However, in my experience, nether ATI nor NVidia can write drivers to save their lives. Hell the ATI ones can't even manage text mode properly on FreeBSD/UltraSparc. The NVidia ones wont even load on FreeBSD/I386 with xorg 7.2.

I get better performance from a five year old 4MB S3 card, because it doesn't need half baked closed source crap. Its not about dogma, its about bloody working software!

Even drunken students could write better software on friday night than these idiots are pretending is a saleable product.

Re:Linux or Open-Source drivers?

[mi]

NVidia may have released FreeBSD/i386 drivers, but has anyone actually got the damned things to work?

I had no problems with the drivers as installed by the /usr/ports/x11/nvidia-driver . Used it on FreeBSD-4, -5, and 6. Maybe, they have not updated for Xorg-7.2 yet — have not tried that...

If you were trying to install directly, rather than via port (pick the right one, to get the version of the driver, that supports your card), you ought to give port a try...

Just make it work

[SanityInAnarchy]

From TFA:

This schedule does also explain why new kernel and X.Org support isn't generally added the same month as its release. If a new kernel at the start of the month breaks fglrx support, that month's driver is already far into the validation and beta stages, which prevents engineers from appending support to the branched driver.

This is why you should follow the new kernels and X.Org while in the development stage. Follow the CVS (or GIT or SVN).

It's true, that kernel may take longer to get out the door than you'd like -- but your new drivers should be backwards-compatible by at least one or two minor kernel versions, right? That way, even if your release is a month or so ahead of the official kernel release, you probably already support it by the time it does roll out.

It is also important to keep in mind that while the AMD release notes in every driver may not be as long as an end-user would like, the developers are actually working on things all of the time.

The end-user doesn't care about the release notes, probably doesn't read them. I know I don't read the release notes on my nVidia drivers, I just know that they just work, and just about as well as they do on Windows.

It is not that we expect you to be doing some set amount of work for us every month. It is that we expect you to not be so insanely, stupendously far behind your competition (Intel and nVidia) that I simply have to tell everyone building/buying a computer for use with Linux to get nVidia if they want games, and Intel if they don't, but never, ever buy ATI.

The entire year we have seen only one formal Linux display driver release from NVIDIA for the GeForce series, along with two beta releases (100.14.03 and 100.14.06) and finally another legacy release (1.0-7185) for their older graphics cards.

And all of them worked.

On the AMD side, however, there have been five Linux drivers this year with another seven expected by year's end.

All of which are lacking certain features I've taken for granted elsewhere (AIGLX, for one), perform worse, and generally suck more.

It's been a LONG time since I tried ATI/AMD for video, and it will take a lot more than a pathetic "It's not our fault, really!" article to make me try them again.

As far as the NVIDIA Linux driver itself goes, it is separated into independent x86 and x86_64 packages and no distribution-specific scripts are included with the mainstream driver. NVIDIA Corporation does not rely upon distribution vendors and community maintainers for allowing the end-user to generate distribution-specific packages, which AMD relies upon heavily.

Irrelevant. My nvidia drivers shipped with Ubuntu, so I really couldn't give a damn who it was that took the x86_64 package and turned it into an Ubuntu package.

In fact, while I do encourage testing on various distros, really, just provide one easy, reasonably transparent installation package -- in the form of a tarball or something close to it -- and have the license allow distros to repackage them for you.

Also, you're forgetting Intel. Intel is going to be the really ugly competition here, because they are the only game in town with a fully supported, fully open driver. If Intel ever releases a video card capable of taking on ATI or nVidia in my price range, I'll take it. (In the past they have done integrated crap that doesn't really compare to the $100-200 range of nVidia/ATI cards, but they've told us that's changing.)

This is the perfect opportunity for AMD, by the way: Just open up your code, and I'm there. The performance difference in hardware isn't significant, but better drivers are, and open drivers would be better.

R300 not supported.

[Benanov]

Those drivers actually have firmware in them--lots of binary blobs. Not exactly the "preferred form for editing the work"; the R300s are not supported, only the R200 cards are according to your link. Someone not update that page in a while?

nVidia blob isn't crap.

[gukin]

I know that lots of folks are vexed that nVidia won't open their 3D driver, saying "If nVidia stops supporting Linux you're all doomed." Well just who do you think supports the open source nv.c driver? How about the nVidia SATA driver? Yes, it's nVidia, so even though the 3D stuff is closed source, they're still supporting OSS.

Next about the "Crap" drivers from nVidia, I've ordered a bunch of new Linux PCs, each will have a low-end nVidia video card added when it arrives? Why? Well I need dual headed support and that can be spotty with other video card vendors. I also need to run them in 8-bit color (don't ask, I just need to.) and my experience with the glorious wonderful OSS Intel video drivers is a nasty little box that follows the cursor around on the screen. I don't have ANY issues with nVidia's "Crap" drivers, everything looks great and works great.

I applaud ANY vendor who makes efforts at supporting OSS but I buy stuff from vendors who support Linux. Every system I own is either an nForce mobo with an nVidia video card or is simply sporting an nVidia card. When asked about what to buy, I recommend nVidia products. They have the best quality 3D support and performance of ANY vendor (which isn't much), they make it possible for me to play games under Linux that I'd otherwise have to play under Windows and that is worth a lot to me.

Re:nVidia blob isn't crap.

[tixxit]

Disable hardware cursor support in xorg.conf. I've had that issue on a few old laptops (various video cards), and it always fixes it right up.

Re:nVidia blob isn't crap.

[kungfujesus]

Seriously, i have no idea why people bitch about the nvidia 3d drivers for linux. The nvidia driver works flawlessly for this 8800GTX on linux.

Re:nVidia blob isn't crap.

[makomk]

Of course, last time I looked at the "open-source" nv driver, it was totally and deliberately obfuscated - no comments, lots and lots of magic numbers, and absolutely no way of knowing what was going on unless you were an NVidia developer with access to the hardware documentation. (It didn't start out that way - I think it was the price of NVidia's co-operation.)

Re:nVidia blob isn't crap.

[Rambo]

Without a doubt, nVidia makes top-notch drivers and tries to stay in sync with the rapidly developing kernel and distributions. I've seen some nice ATI cards and been tempted, but every time I look I shrug it off and buy another NV card. Their installer is pleasant, quick, and text-mode based so I can even set up my wife's computer remotely. Their openness in allowing you to (easily) expand the installer files and potentially patch specific files (hibernate support was problematic at one point), then run the installer from that set of modified files is to be applauded. Having a unified driver core makes for a excellent experience for users of both Windows and Linux.

Larry

Re:ATI driver quality

[Anne Thwacks]

In that case, Id liken having an NVidia card to having a death in the family. SInce I instaleld a NVIdia, having become exceedingly fed up with ATI, I have not been able to boot my FreeBSD machine into X even once.

All of these people are scum, and I really dont care if some are more scummy than others.

I want support for Sun supplied ATI Rage on UltraSparc in FreeBSD NOW!!!!

So what do you tell them

[baggins2001]

I have had NVidia cards before and just gave up on them. Recently I had a whole motherboard and just gave up on it because the onboard card wouldn't work correctly with the distribution of linux.
I have had other Nvidia cards where I have recompiled the driver for each new version of the kernel. I don't want to recompile drivers everytime I upgrade a kernel. Especially on server systems, but heh whatever for now.

You know if we could get every linux user (assuming 1 million) to invest 3,500 - 4,000 dollars in AMD, we would have controlling interest and could tell them to do it however we wanted, but then we would probably want to protect our investment and not let our competitors know how our devices are built or work.

Hey but wouldn't it be cool, we would have our own linux cpu and video company, that could be used to leverage motherboard manufacturers and wireless cards to make compatible equipment

Hey, got to go my bongs empty.

AMD just held there stockholders meeting so it would be a year before we could have a stockholder protest.

problems and bugs

[shish]

stability and performance problems not to mention a great deal of bugs.

You mean stability and performance problems are features?

Support was pretty bad...

[dascandy]

I have a fairly recent ATi card (X1300, ok, I'm a cheapskate) and I tried to install my card when I got it.

First try, nope, no default driver since it's brand new. Second try, ATi drivers - nope, the card was out 3 months, but no drivers. Screwed - SVGA only at a bad non-native resolution without any 3D support.

About 1.5 months later I tried to download their drivers again, this time it would support it. Wouldn't install. There are a few undocumented things you have to turn on in your kernel or it won't work, not to mention some symlinks and devices that are assumed, not specified or even given in an error message.

Two weeks later I finally figured out which parts I didn't have that broke it (shmem support, mostly) and got it to work. Now, it's a breeze. One in every 40 movies I watch makes mplayer occasionally do something that completely whacks the driver into triple fault - IE, reboot. It supports two displays but if you use that, both of them will occasionally forget for 1/25th of a second that they were on and just display garbage instead. I'm an operating system developer and if I want to do any development on the card, I have to either copy / rewrite the X driver (which, needless to say, didn't support the card) or just be screwed.

No, that's not the track record I know AMD has. I'm hoping strongly that AMD will change the ATi corporate direction into the right direction.

Laptops cursed with ATI cards

[Richard_J_N]

Unfortunately, my (otherwise excellent) T60p thinkpad is crippled by having a FireGL 5200 ATI graphics card.
The fglrx driver works well enough mostly (although no AIGLX at all), but about once a week, doing something like scrolling a page will crash the machine. I've gone back to the VESA driver - at least it doesn't crash.

Unfortunately, the vesa driver doesn't support any 2D acceleration, such as copy-rectangle. This means that scrolling large pages is CPU-limited (on a very fast core-duo machine!).

AIGLX

[xNstAble]

I think that if in more than one year they have not been able to deliver drivers supporting AIGLX/Beryl/Compiz, than probably there's something really wrong with their process.
Maybe it's just they are not involving enough people in the development of the drivers, but in this case open sourcing the drivers would be a great idea. In any case at the moment I would never buy another ATI card, since I use almost only Linux and since Beryl/Compiz are one of the most compelling reasons for me currently to buy an accelerated graphic card.

Re:Does AMD just not get it?

[ArcRiley]

Like McDonald's board of directors meeting over poor sales in India, clueless to the ethical views of that market. Sales VP: "They don't like our beef" Marketing VP: "So we'll give them better beef, fresher!" Technical VP: "We could slaughter the cow on site if it helps" Sales VP: "That could be a great slogan, 'fresh from the cow'" [cue standard nods of agreement and voicing of support from around the table]

My experience with ATI

[king-manic]

I had 2 identical systems a while back. old P4 1.8gz with asus mother boards I got from a dealer friend who couldn't moev them. They were unopened cpu's and new, just not cuttign edge. We needed 2 comps one for me one for my brother. At the time I had been Nvidia exclusive since about the first GeForce. I was advised that for that generation the ATI 9600 was the best bang for the buck. While my brother already had my hand me down GeForce MX. My system was crashing consistantly at least twice a day and the error message ont he windows systems traced it back to the aTI drivers. My brothers computer was stable and fine. So I went back to nvidia after that. It might have had great bang for the buck but crashign twice a day is not acceptable.

PS. And no windows in general does not crash twice a day on my systems. So it's acceptable.

Blast from the past

[G3ckoG33k]

I think a Carmack + the Torvalds could make some serious inroads here, obliterating the scenario you pose. Unfortunately, the Carmack I think of is apparently into rocketry in his spare time... :(

-

Perfect time to boycott.

[nbritton]

Now that Intel's back in the drivers seat we can leverage AMD into releasing ATI chipset documentation. Send a message that we won't stand for this bullshit anymore... boycott them.

Let the president of AMD know why you won't buy his products: dirk.meyer@amd.com

I didn't get it. When the problems will be solved?

[anton_kg]

I have several issues with ATI drivers and my thinkpad t42p. All of them are listed in http://www.thinkwiki.org/wiki/Problems_with_fglrx ... and all of them are critical. I mean like screen flickering and hang on loggin out. I simple can't use the driver with such bugs!
These bugs are filed for 1.5 year by now. I don't care if ATI has better developing circle or not.
When all problems will be solved?
When I can use TV-OUT with my ATI card under linux again?.. And don't tell me about IP. IBM, for example, released drivers as open source and it's comes with xorg by default.
I simple don't recommend buying ATI cards to anyone. The better alternatives are exist.

It will happen.

[ichigo 2.0]

It's time for ATI to release their drivers as OSS.

Don't worry, they will. Right after the planned x86-GPU extensions make drivers obsolete.

Windows logo on products

[Jac_no_k]

There is some usefulness to the MS-Windows logo for systems in a locked down environment. After installation by someone with admin privs, the software or hardware will work without local administrator privileges.

I'm convinced they pay this guy at Phoronix

[Hohlraum]

to periodically release this BS just to keep ATI regularly on the Linux sites. Ever heard the saying 'There is no such thing as bad press'? Its a freaking joke, I've stopped reading these site all together.

nVidia

[Z34107]

All they need to do is open up the specs, and people will do all the work for them

Yes... if ATI opens up their specs, their people will do all the work for nVidia's people. And vice versa.

I, for one, can understand why there's some animosity towards releasing the blueprints of your state-of-the-art 5-hojillion-manhours-in-the-making video card to all the tubes on the internets.

Granted, it's not the same as giving nVidia a briefcase of trade secrets, but you have to be careful when your company's existence depends on that extra frame per second your hardware gets in Doom VII 1/2

Well duh!

[xtracto]

this is written as if its a good thing!

It is written as if it is a good thing beacuse the issue is that if they do not support DRM, you cant play such X or Y content, if they do not support "Broadcast flag" you cant see X or Y channel.

They did not make DRM they are just making their cards able to *read* such stream of data...

even the number of developers is not open-sourced!

[davFr]

from the article:

While the number of actual developers working on the Linux driver is confidential, there are over 100 people involved with the closed beta program.

Confidential meaning here "so few we are ashamed to tell you"!!...
But I am not so sure about this one : does "100 people involved with the program" means "code monkeys, managers, and coffee robots included"? ;o)