The Truth About AMD's Development Cycle

[yoshi314]

so does the linux team have some deadlines for their drivers? i mean deadlines like e.g. 10th of may, instead of 1 release/month :]

hmm this release is a bit dissapointing. i suppose they're focusing on R600 support more than anything else now. if the documents shown are true, that is.

[Michael]

so does the linux team have some deadlines for their drivers? i mean deadlines like e.g. 10th of may, instead of 1 release/month :]

hmm this release is a bit dissapointing. i suppose they're focusing on R600 support more than anything else now. if the documents shown are true, that is.

Interesting question... Well, if the higher-ups at AMD grant Phoronix permission, we will be telling the truth to their monthly driver development cycle.

[d2kx]

What truth? The driver is always older like the Windows' one is from the 2nd of february.

I'm wondering what the 1200 lines of code do...

[Michael]

Okay, here is the scoop. AMD has granted Phoronix permission to publish an article tentatively titled ATI/AMD's Development Method Exposed or The Truth About ATI/AMD's Development Cycle.

This article will tell what really goes into each driver release. With month's where there are not many end-user changes in the official change-log, there really are more changes than what you may actually see. Many users take for granted how much work actually goes into each driver release. In this article we will clear up all of these myths and misconceptions about the fglrx driver and their monthly release plans.

Since last year when AMD announced its intent to acquire ATI Technologies, there have been countless rumors about how this would affect the development of the drivers and what repercussions may be seen as a result. Well, in this article we will be telling you the straight facts about the changes being made. To keep you salivating until this article is published, AMD has made the strategic decision to make Linux part of their unified development cycle. But you'll have to wait for the article to see what this entails :D.

I am still in the process of receiving all of the information from AMD, but hopefully this article will be ready for your reading pleasure by early March.

Whether you are a Windows or Linux user, this should be a very interesting read.

[Svartalf]

I am still in the process of receiving all of the information from AMD, but hopefully this article will be ready for your reading pleasure by early March.

Whether you are a Windows or Linux user, this should be a very interesting read.

Indeed. I'm looking forward to it. :D

[niniendowarrior]

Okay, here is the scoop. AMD has granted Phoronix permission to publish an article tentatively titled ATI/AMD's Development Method Exposed or The Truth About ATI/AMD's Development Cycle.

This article will tell what really goes into each driver release. With month's where there are not many end-user changes in the official change-log, there really are more changes than what you may actually see. Many users take for granted how much work actually goes into each driver release. In this article we will clear up all of these myths and misconceptions about the fglrx driver and their monthly release plans.

Since last year when AMD announced its intent to acquire ATI Technologies, there have been countless rumors about how this would affect the development of the drivers and what repercussions may be seen as a result. Well, in this article we will be telling you the straight facts about the changes being made. To keep you salivating until this article is published, AMD has made the strategic decision to make Linux part of their unified development cycle. But you'll have to wait for the article to see what this entails :D.

I am still in the process of receiving all of the information from AMD, but hopefully this article will be ready for your reading pleasure by early March.

Whether you are a Windows or Linux user, this should be a very interesting read.

You know Michael. I'm very much looking forward to this if only to see how my perception of ATi matches with the article.

[d2kx]

This is going to be VERY interesting, good to hear that AMDs let you do such an article!

[Michael]

It's taking a bit longer than expected as they are preparing slides for Phoronix :)

Though hopefully the article will still be out in a week or two.

[drag]

AMD has made the strategic decision to make Linux part of their unified development cycle. But you'll have to wait for the article to see what this entails

Hopefully it will entail supporting open source drivers along with kernel/X developers, because that is is the only thing that will ever get me to waste my time with ATI hardware.

As it stands right now buying a newer ATI hardware is one of the worst possible mistakes you can make when building a Linux PC and it's going to be like that as long as ATI is depending on closed source FLGLX.

Q. What is the easiest way to install drivers for a r500 series ATI card?
A. Sell it on ebay and buy a Nvidia card. Then install the nvidia drivers.

[ivanovic]

AMD opening the fglrx drivers will not happen anytime soon (http://www.phoronix.net/forums/showthread.php?t=1454). And I don't think they will release any specs either. The only real thing to wait for are the discrete intel cards, IMO AMD/ATI (graphics card wise) is dead!

[drag]

AMD opening the fglrx drivers will not happen anytime soon (http://www.phoronix.net/forums/showthread.php?t=1454). And I don't think they will release any specs either. The only real thing to wait for are the discrete intel cards, IMO AMD/ATI (graphics card wise) is dead!

Yep. And it's too bad.

I like AMD a lot more then Intel, but buying AMD means pretty much that your requiring at least propriatory drivers for the video card (and badly supported hardware usually for audio/network/sata), unless your using those onboard Via stuff (which is fine for most sorts of video playback) which have virtually no 3d performance.

So it's a shitty situation.

You have ATI, which made bad decisions for licensing, then turns around and inflicts those poor decisions on end users with shitty drivers. Which is one thing. But to now have AMD not being able to fix the situation is a damn shame.

Maybe if Dell or HP starts shipping Linux hardware they will turn around. When there starts to be real financial punishment for the state of Linux driver support as AMD/ATI hardware gets passed over again and again for those (relatively) large scale Linux deployments we are seeing more often.


One thing to keep in mind is that Intel isn't releasing specs OR releasing source code. Sortof.

They released specs for 8xx stuff, I beleive. Which forms the basis for those Linux drivers. Then for the 915 stuff they hired out to Tungsten Graphics for driver development (3d visualization company created by Xfree developers). Now for 965 stuff they internalized driver development and hired several prominate X hackers for improving X support for their hardware. (and that is going beyond mere drivers, but going also into the development of better open source OpenGL stack)


So it's not like Intel is perfect or wonderfull. They are doing open source drivers with a very pragmatic approach. Either Nvidia or ATI could trivially be much more open then Intel and still be able to protect their 'IP'.


There already are Free/Open source drivers for ATI hardware. Linux/X.org has open source 2d and 3d drivers for r100, r200, r300, and r400 series cards. So it's not like ATI would have to realy do a whole lot to get well supported hardware.

[Saist]

Q. What is the easiest way to install drivers for a r500 series ATI card?

A: http://www.mepisguides.com/Mepis-6/fglrx/install-fglrx/fgrlx.html

[Svartalf]

There already are Free/Open source drivers for ATI hardware. Linux/X.org has open source 2d and 3d drivers for r100, r200, r300, and r400 series cards. So it's not like ATI would have to realy do a whole lot to get well supported hardware.

Heh... They only have a solid handle on the R100 and R200 series chips right now.

They've got an impressive amount of support for the R300 and up (R400/R500 chips are merely an expansion of the functionalities of the R300 in the form of more pipelines, faster pipelines, etc...)- but it's not anywhere near the amount they have on the R100/R200 chips.

x.org doesn't (Yet!) appear to have FSAA, Hyper-Z, Anisotropic Filter, and a few other things for one to claim that they really have R300 support. It's coming along nicely, but it's not really supported. AMD could conceivably step in and hand everyone in our community a few bread crumbs in this area to make things all better (and gain favor in our community...)- but I seriously doubt they will. Which is a shame.

[drag]

Heh... They only have a solid handle on the R100 and R200 series chips right now.

They've got an impressive amount of support for the R300 and up (R400/R500 chips are merely an expansion of the functionalities of the R300 in the form of more pipelines, faster pipelines, etc...)- but it's not anywhere near the amount they have on the R100/R200 chips.

x.org doesn't (Yet!) appear to have FSAA, Hyper-Z, Anisotropic Filter, and a few other things for one to claim that they really have R300 support. It's coming along nicely, but it's not really supported. AMD could conceivably step in and hand everyone in our community a few bread crumbs in this area to make things all better (and gain favor in our community...)- but I seriously doubt they will. Which is a shame.

The R400 series is a extension to the R300 series. Faster, more pipelines, and PCI express support.

R500 on the other hand is very different. Or at least different enough. There is _zero_ support for it with Free software drivers, 2D or 3D. Vesa driver is all there is for those guys.

I don't think that you are going to see any of the advanced features of the r300/r400 chipsets anyways. There is a very limited amount of developers in teh Linux community that are realy capable of reverse engineering hardware and they've seemed to have turned their attention to Nvidia.(I've read at least one of them say that Nvidia hardware is more pleasent to work with and much less buggy then ATI.. which is a huge problem for reverse engineering. (even if you do get something right, it can still be wrong because the hardware may be broken in special cases))

edit:

In terms of compatability with Linux desktop the free software drivers are superior to the flglx. For example they support AIGLX for easy acceleration of things like compiz or beryl.




Q. What is the easiest way to install drivers for a r500 series ATI card?
A: http://www.mepisguides.com/Mepis-6/f...lrx/fgrlx.html
That's a happy fantasy. In my own personal experiance I couldn't get Flglx drivers to work at all with my hardware.
I tried official Debian packages. I tried the Ubuntu packages.
I tried the official releases from ATI. Multiple different versions.

And this wasn't very long ago.

And I've compiled my own custom versions of X.org, the kernel, and mesa development libraries, from scratch, and installed them in parrellel with my existing distro-provided software in order to play around with XGL when it first came out. And I've been using Linux for years.

If I can do that well and yet not be get those stupid ATI drivers to work correctly then it's just dumb luck weither or not it works for anybody.

Meanwhile my Intel graphics had 3d acceleration the first time I booted up my machine from a fresh install with absolutely no configuration or intervention from me.

[yoshi314]

well i've been using gentoo for a couple of years.

when i tried ubuntu or debian i was clueless about how to enable ati's opengl in xorg :]

one distro did specify fglrx libs explictly in xorg.conf

like Load "/usr/lib/xorg/modules/dri/fglrx_dri.so" in Modules section of xorg.conf

gentoo does symlinks :
/usr/lib/opengl/ati/extensions -> ../xorg-x11/extensions
/usr/lib/opengl/ati/include -> ../xorg-x11/include
/usr/lib/opengl/ati/lib
/usr/lib/opengl/ati/lib/libGL.la
/usr/lib/opengl/ati/lib/libGL.so -> libGL.so.1.2
/usr/lib/opengl/ati/lib/libGL.so.1 -> libGL.so.1.2

i guess one way or another, it should eventually start working. (proper kernel config is also important).

[Svartalf]

R500 on the other hand is very different. Or at least different enough. There is _zero_ support for it with Free software drivers, 2D or 3D. Vesa driver is all there is for those guys.


I'd buy "different enough". I've reliable info that they're intrinsically the same architecture- same rules, etc., but perhaps differing register layouts. The R600, however...


I don't think that you are going to see any of the advanced features of the r300/r400 chipsets anyways. There is a very limited amount of developers in teh Linux community that are realy capable of reverse engineering hardware and they've seemed to have turned their attention to Nvidia.(I've read at least one of them say that Nvidia hardware is more pleasent to work with and much less buggy then ATI.. which is a huge problem for reverse engineering. (even if you do get something right, it can still be wrong because the hardware may be broken in special cases))


I'd hesitate to call the ATI parts "buggy"- it's more their drivers than it is the silicon from what I am led to believe. And the drivers situation is more due to a lack of resources than anything else.


In terms of compatability with Linux desktop the free software drivers are superior to the flglx. For example they support AIGLX for easy acceleration of things like compiz or beryl.


I'd have to concur. Now, having said this, I will observe that ATI's driver support is slower in many areas than the Free Software Drivers. I shouldn't have a situation where I can play a game with an R200 and not with an RS370. But, I do on many games- things like Unreal Tournament 2k3/2k4...


That's a happy fantasy. In my own personal experiance I couldn't get Flglx drivers to work at all with my hardware.
I tried official Debian packages. I tried the Ubuntu packages.
I tried the official releases from ATI. Multiple different versions.


Heh... I've managed to get them working and keep them working- mostly due to the Livna repository RPMs of the drivers. This is in 64-bit mode to boot. I will say that making the fglrx drivers go without something pre-packaged for the distribution is NOT for the faint of heart or anyone but the most skilled in Linux system work. Not a good thing, if you must know.


Meanwhile my Intel graphics had 3d acceleration the first time I booted up my machine from a fresh install with absolutely no configuration or intervention from me.


Well, that should be a hint to NVidia and AMD. If they want a piece of that, they're going to have to step up to the plate like Intel did.

[niniendowarrior]

It's taking a bit longer than expected as they are preparing slides for Phoronix :)

Though hopefully the article will still be out in a week or two.

Michael, I hope you haven't forgotten about this. Is it already posted somewhere?

[Michael]

Michael, I hope you haven't forgotten about this. Is it already posted somewhere?

Nope, haven't forgotten about it. AMD is making some new slides for me, so it's taking a while.

[yoshi314]

are you sure THEY haven't forgotten about this?

i can already see it. ati pushes out the slides and says "our developers were busy preparing the slides. no drivers this month" :D

[Michael]

are you sure THEY haven't forgotten about this?

i can already see it. ati pushes out the slides and says "our developers were busy preparing the slides. no drivers this month" :D

Let me assure you, no they haven't forgotten about the slides and they have definitely not forgotten about this month's driver. :D

[yoshi314]

that was just an attempt for a brief comic-relief ;-)

any hints what's coming up with this month's drivers?

[Michael]

any hints what's coming up with this month's drivers?

You'll have to wait for the Phoronix article, like every month.

[Svartalf]

You'll have to wait for the Phoronix article, like every month.

No, I want it RIGHT NOW, Dagnabit! ;)

I'm curious what they're going to say in this thing, to be honest with you. :D Guess I'll have to work on that "patience" thing my mother always chided me about, eh?

[Michael]

No, I want it RIGHT NOW, Dagnabit! ;)

I'm curious what they're going to say in this thing, to be honest with you. :D Guess I'll have to work on that "patience" thing my mother always chided me about, eh?

It should be worth the wait :D

[yoshi314]

too much waiting results in extremely high expectations. let's hope it does not get to us.

[Michael]

too much waiting results in extremely high expectations. let's hope it does not get to us.

Information from AMD has now been delayed to the end of the month, but I assure you it will be worth the wait :)

[Moustacha]

end of the month isn't too far away, goody

[glussier]

One thing I read, is that AMD intends to use the ATI control center for more than the video card. With-it you'll be able to set your sound, chipset etc...

Hopefully there won't be too much AMD advertisements in the Control Center, otherwise I'll stop using ATI/AMD on my Intel systems.

[d2kx]

Can't wait, too.

@qlussier: When they do this in a year or two, be sure that they most likely will not do that in their linux drivers.

[yoshi314]

One thing I read, is that AMD intends to use the ATI control center for more than the video card. With-it you'll be able to set your sound, chipset etc...it would be nice if it had the avivo settings tab, like the windows version does.

[emparq]

some interesting tidbits:


ExtremeTech: The Vista driver, beginning with Catalyst 7.1, contains a "re-architected" OpenGL driver component. Exactly what is "re-architected" about it?

Rogers: Everything. This is a completely new driver, written from scratch in the last three years. ...

...

ExtremeTech: Why make the switch in the Vista driver, and not the XP one? Can XP users expect to get this new re-architected driver in the future?

Rogers: The new OpenGL driver has been designed from the ground up for Windows Vista, Windows XP and Linux and internally we are running and testing it on all three (or six if you count the 64 bit versions).
...
The XP and Linux versions of this new OpenGL driver will ship later this year.


---------

http://www.extremetech.com/article2/0,1697,2103633,00.asp

---------

Looking forward to your pending article, Michael. Hopefully, more questions will be answered (like how much later this year?.. :))

[Malikith]

That is definitely interesting, I want it now.. hehe. ;)

[DarkFoss]

Nice read...It's about time ATI re-vamped their OpenGL Nvidia has been killing them for years even in Windows..

[emparq]

Hi Michael,

Any updates from AMD on when they'll release those slides? Did they push back another month?

[Michael]

Hi Michael,

Any updates from AMD on when they'll release those slides? Did they push back another month?


The slides are now completed and they are now going through the internal processes to get authorization so that Phoronix can publicly write about it. I hope it will all be ready shortly.

[emparq]

Good to hear. Thanks!

[DrCR]

Looking forward to both the article and better ATI drivers. :D

[yoshi314]

it looks that everything going through "internal process to get authorization" takes the same amount of time (thinks about airlied's r500 driver proposal) :D

[Svartalf]

Looking forward to both the article and better ATI drivers. :D

I'm looking forward to the article. Better drivers? Maybe. Maybe not. With all that's went on, I'm going to sit on the fence on that one- I'll believe it when I see it and it may be a while yet before we do.

[DarkFoss]

I think that they can't help being better drivers by default. It's pretty evident that the entire architecture the original drivers are based on is completely outdated. There has been zero performance increases in how many releases now.. 10? No AIGLX support.. it most likely can't be supported with the drivers in their current state
They remind me of ATI's Windows drivers pre-catalyst.. Cobbling on support for newer cards any improvement(s) seemed to break other parts of the drivers. They best drivers came from Cyborg and Omega mixing previous and leaked beta's rather than ATI's Official releases. All that changed with Catalyst.. I'm hopeful that it will be the same way for their Linux drivers once the new architecture is implemented.

I'm also looking forward to the day that I can participate in a thank you thread for the Linux Catalyst Drivers like there is for Windows. Right now it must be kind of discouraging to MTippett and the rest
of AMD's Linux staff, to work on the releases and see very few positive responses.. I'm sure it's hard work just keeping these functional.

[Svartalf]

I think that they can't help being better drivers by default. It's pretty evident that the entire architecture the original drivers are based on is completely outdated. There has been zero performance increases in how many releases now.. 10? No AIGLX support.. it most likely can't be supported with the drivers in their current state


Just because they've redesigned them doesn't make for better drivers for us. We might get better drivers. We might not. To presume that they're going to be better because they re-worked them is setting yourself up for disappointment.

[emparq]

Just because they've redesigned them doesn't make for better drivers for us. We might get better drivers. We might not. To presume that they're going to be better because they re-worked them is setting yourself up for disappointment.

While I agree that being overly optimistic about anything can lead to disappointment, why would you doubt that these drivers (when and if they ever get released), wouldn't be 'better' at least in some regard (performance, stability, or added features)?

Development time costs money, and AMD/ATI wouldn't be allocating developer-hours to completely re-work an underlying subsystem of their drivers just to realize no improvement on any front. What would be the point in that?

At the very least, one would have to expect that with these new drivers, a brighter path towards higher performance, more features, or less instability problems would emerge.

[cornelius]

Any updates on the slides Michael? :)

[Michael]

Any updates on the slides Michael? :)

I asked AMD about it yesterday with it taking sooo long. Apparently their internal control process for releasing public material takes longer than the Linux people expected and they have already had to remove a few slides. I am hoping for it soon...

[rolz]

Another week passes...


In two weeks time it would have been 3 months ago when the slides were first mentioned . Every time I return to this page, I feel a bit more like a horse that's chasing a carrot which is being dangled in front :)

yet still very interested in seeing those slides. Not as much as the next driver release, but still pretty damn curious ;)

I don't want to come of all mean and bitter, bit ironic though how a slideshow "the truth about our monthly development cycle" is taking 3 months :)

[Svartalf]

Another week passes...


In two weeks time it would have been 3 months ago when the slides were first mentioned . Every time I return to this page, I feel a bit more like a horse that's chasing a carrot which is being dangled in front :)

yet still very interested in seeing those slides. Not as much as the next driver release, but still pretty damn curious ;)

I don't want to come of all mean and bitter, bit ironic though how a slideshow "the truth about our monthly development cycle" is taking 3 months :)

You've little idea about how IP leakage and bad PR adverse this bunch is. Honestly.

[Michael]

Great news folks :) The article should (finally) be out by month's end.

The AMD VP last in the chain has apparently signed off on the slides yesterday.

[Svartalf]

Heh... Finally. :D

[Raven3x7]

Great. Lets hope the info they are going to give you was worth the wait.

[emparq]

That is good news. However, I think this also demonstrates a truth about AMD/ATI's management and their decision-making policies.

How long ago did you first contact AMD/ATI, Michael? I imagine that this was sometime back in February? And now, only in the middle of May (5/18), do they finally get the final sign-off?

I really hope that this delay period was just an extreme case of nondecision/bureaucracy within AMD/ATI's management structure.

[Michael]

That is good news. However, I think this also demonstrates a truth about AMD/ATI's management and their decision-making policies.

How long ago did you first contact AMD/ATI, Michael? I imagine that this was sometime back in February? And now, only in the middle of May (5/18), do they finally get the final sign-off?

I really hope that this delay period was just an extreme case of nondecision/bureaucracy within AMD/ATI's management structure.

Actually it was AMD that came to me initially. The delay has been in part due to the engineering manager that initially contacted me, having to go through all sorts of hurdles to get this information released -- since he isn't in their public relations department, he had to go through and get this information cleared with management, legal, VPs, etc...

[niniendowarrior]

Heh... Finally. :D

You can say that again! :D

[emparq]

Actually it was AMD that came to me initially. The delay has been in part due to the engineering manager that initially contacted me, having to go through all sorts of hurdles to get this information released -- since he isn't in their public relations department, he had to go through and get this information cleared with management, legal, VPs, etc...

Cool, I'm psyched to hear that (that AMD/ATI proactively approached you.. not so much the part about the engineering management having to go through so many hurdles).

That underscores my point though, that there's certainly room for improvement in whatever process is controlling their public relations- at least as it concerns their *nix-related development.

What I don't understand though, is why the secrets in the first place? Even if they don't have anything to show us just yet, why not let their developers or marketing people blog or interview about their near-term driver development goals or roadmap?

As you pointed out on Chris Blizzard's blog here (http://www.0xdeadbeef.com/weblog/?p=288), AMD/ATI Marketing guy Henri Richard said that they "committed to fixing the ATI problems with open source." That certainly sounds great, but what does that mean exactly? Better binary (fglrx) drivers? Opening up hardware documentation so that open-source developers can actually write truly open drivers?

This is exactly what I'm talking about. Why keep this type of information secret? Why not just tell us, what they're planning on doing, what they've done so far to get there, and when/how often we can expect to hear news or updates on the development process? That right there would already be a giant first step towards rebuilding their good-faith with the *nix community, even if they aren't delivering on their promises just yet.

But at any rate, I suppose that these slides are essentially just that- AMD/ATI trying to be more open about what they're doing and their near-term goals. I really hope so.

But I guess we'll (finally) find out soon.

[Michael]

Cool, I'm psyched to hear that (that AMD/ATI proactively approached you.. not so much the part about the engineering management having to go through so many hurdles).

That underscores my point though, that there's certainly room for improvement in whatever process is controlling their public relations- at least as it concerns their *nix-related development.

What I don't understand though, is why the secrets in the first place? Even if they don't have anything to show us just yet, why not let their developers or marketing people blog or interview about their near-term driver development goals or roadmap?

As you pointed out on Chris Blizzard's blog here (http://www.0xdeadbeef.com/weblog/?p=288), AMD/ATI Marketing guy Henri Richard said that they "committed to fixing the ATI problems with open source." That certainly sounds great, but what does that mean exactly? Better binary (fglrx) drivers? Opening up hardware documentation so that open-source developers can actually write truly open drivers?

This is exactly what I'm talking about. Why keep this type of information secret? Why not just tell us, what they're planning on doing, what they've done so far to get there, and when/how often we can expect to hear news or updates on the development process? That right there would already be a giant first step towards rebuilding their good-faith with the *nix community, even if they aren't delivering on their promises just yet.

But at any rate, I suppose that these slides are essentially just that- AMD/ATI trying to be more open about what they're doing and their near-term goals. I really hope so.

But I guess we'll (finally) find out soon.

AMD needs to remain competitive and one of the ways they do that is by not showing all of their cards at once. By saying such a thing as they will be delivering a new OpenGL driver in August followed by providing the source code to their 2D components, that would give NVIDIA several months time to respond with a new driver feature or what have you (NOTE: the statements used in this sentence are for example purposes only).

And as for secrets for other things like new products, as Advanced Micro Devices is a publicly traded company they have regulations that they must conform with such as with press releases, etc.

As far as your other questions, they will hopefully be addressed in this upcoming article.

[Svartalf]

That is good news. However, I think this also demonstrates a truth about AMD/ATI's management and their decision-making policies.

How long ago did you first contact AMD/ATI, Michael? I imagine that this was sometime back in February? And now, only in the middle of May (5/18), do they finally get the final sign-off?

I really hope that this delay period was just an extreme case of nondecision/bureaucracy within AMD/ATI's management structure.

Not an extreme case- the norm, unfortunately.

[emparq]

AMD needs to remain competitive and one of the ways they do that is by not showing all of their cards at once. By saying such a thing as they will be delivering a new OpenGL driver in August followed by providing the source code to their 2D components, that would give NVIDIA several months time to respond with a new driver feature or what have you (NOTE: the statements used in this sentence are for example purposes only).

And as for secrets for other things like new products, as Advanced Micro Devices is a publicly traded company they have regulations that they must conform with such as with press releases, etc.

As far as your other questions, they will hopefully be addressed in this upcoming article.

Sure, but this is software and drivers support we're talking about- not their bread-and-butter new products that will be hitting the market soon. I understand the need to keep those types of secrets, as new products are what keep them afloat. But sadly, even here, they're doing a better job of keeping the public informed of what's in the pipeline.

At any rate, I understand what you're saying, but I still disagree that they cannot/should not share more information with the public about their *nix-driver efforts.

If anything, their public image in the free/open-source community would improve, and people would badmouth ATI products less. This might lead to healthier future sales as their brand-value improves with a growing market for linux PC's (as demonstrated by Dell's recent announcement to start selling Ubuntu-based PC's).

[joshuapurcell]

At any rate, I understand what you're saying, but I still disagree that they cannot/should not share more information with the public about their *nix-driver efforts.

If anything, their public image in the free/open-source community would improve, and people would badmouth ATI products less. This might lead to healthier future sales as their brand-value improves with a growing market for linux PC's (as demonstrated by Dell's recent announcement to start selling Ubuntu-based PC's).While I also agree that more information would be better for those of us consumers who are interested using AMD products under Linux, I also think that this company has proved they are much more interested in the open source community than their competition due to their promise to provide open source drivers. I think this step goes much further towards helping their image than any document they will be providing (although the doc is still relevant of course).

My next card will no doubt be an AMD/ATI card... in fact I may replace my current card once the drivers are released. I'll still have to take performance into account, so I'm looking forward to future benchmarks.

[Huenengrab]

I also think that this company has proved they are much more interested in the open source community than their competition due to their promise to provide open source drivers.
Well, I'm the first to get stuck in the roof for jumping around like an idiot, once these drivers have been released, but somehow, I kind of doubt that they'll open source their drivers. "Open graphics drivers" (e.g. what got announced at the RH-conference) could mean anything.

[joshuapurcell]

Well, I'm the first to get stuck in the roof for jumping around like an idiot, once these drivers have been released, but somehow, I kind of doubt that they'll open source their drivers. "Open graphics drivers" (e.g. what got announced at the RH-conference) could mean anything.
Not only the definition of "open source drivers" could be different than what people expect, but the open source drivers could come out and we find out that ATI cards still perform worse than Nvidia. It's a huge step forward and I'm excited for what it could mean, but it hasn't happened yet and nobody knows exactly what it will do for Linux gaming. Still, anything is better than nothing... and we may all be pleasantly surprised (I hope).

[Michael]

The Truth About ATI/AMD & Linux is now finalized and is waiting a final review by AMD just to make sure I didn't expose their deeply inner secrets. Expect it to FINALLY go live this coming Friday!

[Moustacha]

huzzah!!!!!

[Vakilik]

The Truth About ATI/AMD & Linux is now finalized and is waiting a final review by AMD just to make sure I didn't expose their deeply inner secrets. Expect it to FINALLY go live this coming Friday!

About time AMD put it through, should be a great article man. Can't wait.

[niniendowarrior]

The Truth About ATI/AMD & Linux is now finalized and is waiting a final review by AMD just to make sure I didn't expose their deeply inner secrets. Expect it to FINALLY go live this coming Friday!

About time. I do wonder if AMD will want this article spun in a particular way... but oh well... that's just me thinking aloud.

[emparq]

As you pointed out on Chris Blizzard's blog here (http://www.0xdeadbeef.com/weblog/?p=288), AMD/ATI Marketing guy Henri Richard said that they "committed to fixing the ATI problems with open source."

Linux Weekly News has posted an article here (http://lwn.net/Articles/234590/) in response to the statements made by above-mentioned Henri Richard and the recent flurry of ATI buzz that followed.

In short, it underscores the fact that AMD/ATI have a long road ahead of them if they are seriously looking to rebuild their good faith with the Linux community.


As it happens, the developers in charge of making graphics work on Linux systems are pretty much unanimous in their lack of enthusiasm. This is not the first time that ATI has made promising sounds, but, so far, the corresponding actions have not been forthcoming. Graphics hacker Dave Airlie is particularly unimpressed (http://lwn.net/Articles/234592/), noting that ATI has not yet bothered to communicate its intentions to the developers:

As for working with the community I'd expect they'd at least try talking to the ppl who maintain the ATI open source driver if they intend on doing something with it...

[Michael]

As it happens, the developers in charge of making graphics work on Linux systems are pretty much unanimous in their lack of enthusiasm.

I am just going to say it now... that there is pure bull shit.

[Svartalf]

I am just going to say it now... that there is pure bull shit.

I'll heartily concur... The malaise is definitely NOT due to a lack of enthusiasm or caring on the part of the developers.

[yoshi314]

there should be another article - what's the root of the problem with fglrx?

i mean, you said that there are devoted developers. and there are some betatesters (no idea how manym though), and even community bugzilla which seems to have some of ati's attention. that mix shouldn't give bad results.

but it somehow does.

the only thing we can be sure about fglrx are X/kernel fixes, although not always timely.

other than we get some obligatory bugfixes, some new (also obligatory) bugs.
new features are far and between, and they're usually different from what majority of people would expect, at least lately.

the last two features i was glad to have were : introduction of opengl2.0 support in january 2006 and textured video support (no idea when it was implemented).

[Michael]

there should be another article - what's the root of the problem with fglrx?

i mean, you said that there are devoted developers. and there are some betatesters (no idea how manym though), and even community bugzilla which seems to have some of ati's attention. that mix shouldn't give bad results.

but it somehow does.

the only thing we can be sure about fglrx are X/kernel fixes, although not always timely.

other than we get some obligatory bugfixes, some new (also obligatory) bugs.
new features are far and between, and they're usually different from what majority of people would expect, at least lately.

the last two features i was glad to have were : introduction of opengl2.0 support in january 2006 and textured video support (no idea when it was implemented).

Some of these questions have been answered in this upcoming article.

[rolz]

whaa i was so hiped to see that michael had the last post when i checked :)

but for now i'll just keep going, "almost there almost there"

[Michael]

I am very happy to report that AMD has signed off on the article last night. There is now nothing else holding us back from publishing the truth about ATI/AMD Linux.

The article is in the publishing queue for June 1st (tomorrow) by mid afternoon.

[Moustacha]

awesome, can't wait to read it!

[emparq]

I am very happy to report that AMD has signed off on the article last night. There is now nothing else holding us back from publishing the truth about ATI/AMD Linux.

The article is in the publishing queue for June 1st (tomorrow) by mid afternoon.

Good to hear, looking forward to finally reading it.

[Michael]

It's out!

Read all about it: http://www.phoronix.com/vr.php?view=10083

This thread is being closed and a new one will open for the discussion.