Talk:AIGLX

Untitled

Started up the article, linked the Fedora Wiki. Someone might want to add some links, but all I can find are little snippets in blogs which are hardly relevant (and say nothing more than I've described in this article.) My work here is done, so someone else should start fixing all the things I've got wrong. :) LainOfIICHAN 16:22, 27 February 2006 (UTC)

BTW, does this work with/replace Xinerama for running GL (like xscreensaver hacks) across multiple displays?

Inaccuracies

"hardware-driven usability improvements, such as smoothed fonts, accelerated window-drawing" -- This is a bit misleading because, compared to Xgl which does font smoothing and accel through GL, AIGLX does _NOT_ do so, instead using traditional acceleration.

The first paragraph also suggests that AIGLX has similarities to compiz, which doesn't make sense. AIGLX is analogous to Xgl, but compositing managers like compiz and modified Metacity work one layer above AIGLX/Xgl, and should in theory work with both.

Will hopefully find time to clean up the article, but in the meantime, readers should be aware of these inaccuracies -- lstanley

Hardware Support

Hardware support for the product is limited to hardware with free drivers due these changes: currently R200 generation and earlier ATI Radeon cards (R300 drivers are currently in development) and a few Intel embedded onboard video modules.

This is outdated. VIA Unichrome works now, and probably several others as well. In addition, the latest nVidia drivers do support XOrg 7.1, though they don't work with AIGLX.

ATI drivers support X.org 7.1 now. Not sure about AIGLX.

ATI NPOV Dispute

What is the problem?

I'll state my case.

I run Linux and have an ATI card. The facts are quite accurate, there is no AIGLX support, the DRI API used is obsolete, it has been going on for over a year and there is no timelime for its introduction. The development process of the driver has not addressed these fundemental issues. Combined with numerous other flaws:

A Look at the State of ATI Linux Drivers (2005) Posted by timothy on Sat Jul 30, '05 03:53 PM from the better-than-nothing dept. http://linux.slashdot.org/article.pl?sid=05/07/30/197259

The State of ATI Drivers on GNU/Linux (2006) Posted by timothy on Thu Jul 20, '06 11:22 AM from the all-I-want-is-some-hot-3d-action dept. http://linux.slashdot.org/article.pl?sid=06/07/20/1433230

...and end-user feedback of the development process:

Insight Into AMD's Linux Driver Development (2007) Posted by CmdrTaco on Monday June 04, @12:07PM from the they-should-all-wear-hats-while-they-code- dept. http://linux.slashdot.org/article.pl?sid=07/06/04/1532215&from=rss

It is quite clear that there is a serious problem with the skillsets available to ATI. It is not something that has happened recently, nor is it something that is new.

Therefore, I do not feel that there is anything wrong with the point of view, it is an expression of the facts. It just happens to be that the facts paint ATI in a bad light.

Whoever made the dispute would like me to change the facts of reality, not my personal opinion.

I also feel that this article needs to be updated, the url above shows the exact responsiveness of AMD to ATI's lack of Linux support. I think the opinion expressed is bias at this point, and not ALL FACTS have been taken into account. 66.41.10.101 15:10, 4 July 2007 (UTC)

a) Whose case? Please sign comments with ~~~~. b) Wikipedia is not a place for people to present cases, it's an encyclopedia, and all material must be sourced. Contributors are editors, not originators, but this material smacks of original research, which is not allowed. See WP:OR. Talk about what is fundamental to next-generation Linux distributions is editorializing, and doesn't belong in this article. It certainly doesn't belong in a section about ATI, as it has nothing to do with them or their drivers, other to increase the negative impression of them (I'm not saying it isn't richly deserved; it just isn't encyclopedic). -- Jibal 22:10, 18 July 2007 (UTC)

(this blank title was breaking this page)

I don't know why this is being disputed; I've run AMD/ATI on both Linux (gentoo) and Solaris 10 within the last 24 hours and I can attest to the fact that this is not a biased view, but rather the reality of ATI drivers on *nix. GLX will work on gentoo (sabayon 3.2 only- 3.3, 3.3b, 3.4loop3 all drop the screen like a hot rock) at low refresh, and just barely... the screen 'shakes' and wobbles. AIGLX will not work. I came across this page while trying to find if I can write to the VFB since ATI's lack of driver support has made my solaris development box a brick. I've never commented on a talkpage before, but I'm dealing with this very issue at the moment and thought I'd put in my $.02. 64.9.41.146 01:23, 5 July 2007 (UTC)

Exactly, me and hundreds of forum questions around the Web are trying to deal with the same issue. AMD/Ati's almost complete lack of regard of *nix users. It is really frustrating. Especialy for Mobility Radeon users - my 3D performance is really bad and I can't even use compiz or beryl. 128.86.161.200 13:53, 7 July 2007 (UTC)

I couldn't agree more. This section is "not neutral" because of choices made by ATI and not the editors of this page. I can't think of how to express that ATI does not support Linux without making it sound negative; however, that is the case.

I agree as well. Can we remove the marker? 88.217.40.170 01:32, 18 August 2007 (UTC)

Wow. AMD's lack of regard for *nix users? I've heard some bull manure in my lifetime, but I think that took the cake. Source is here:

http://www.mepislovers.org/forums/showpost.php?p=67268&postcount=2

Byco AMD/ATI only releases crippled Linux drivers.

Incorrect. ATi/ArtX made no secret that they only viewed Linux as a workstation OS. Their drivers were optimized for multi-pass computational rendering typically used in OpenGL software renderers... something you might find at LucasArts or Dreamworks.

It was not until 2005 that ATi started to treat Linux a serious desktop OS. It also was not a secret that AMD mandated a complete driver rewrite given the existing driver architecture's... mess.

AMD/ATI is able to release drivers for Windows Vista on time.

also incorrect. ATi was/is a developing partner for the Xbox 360. This meant that they had earlier hands on access to the developing DX10 API. ATi was also one of the early developers who worked on Windows XP 64 drivers.

This isn't to say that other developers didn't have hands on with the DX10 API... Nvidia launched a DX10 card with the 8800 series, long before ATi released a DX10 card.

When Vista launched, ATi collectively renamed their long in beta drivers as being stable. Unlike Nvidia, ATi has a clear and concise roadmap for driver releases. As is ATi is still fixing critical bugs in the Vista drivers.

Nvidia offers Linux drivers that are on par with Windows drivers.

... in the same way that the Fiat Panda is on par with a Hyundai accent. That ain't saying much when the driver set has major issues with any code that makes Pixel Shader 1.4 calls, and the driver teams refuses to fix known memory leaks literal years after they are reported.

Nvidia drivers for Vista are delayed.

Nvidia doesn't have a driver roadmap, at least not one they'll talk about. Nvidia also had major problems with Windows Xp 64bit drivers. What is interesting about this is that Nvidia had x86 DX10 capable hardware on the market long before ATi... who was working with the DX10 API on Xbox 360 hardware.

There isn't any excuse for this from Nvidia.

Nvidia GeForce 8 series cards crippled on Linux. http://www.phoronix.com/scan.php?page=article&item=781&num=1

Crippled? Ha. Nvidia can't code drivers. Plain and simple. Like Vista, they finally got caught with their pants down around their ankles going "What do we do now?

What do you wanna bet that Nvidia is able to release drivers on time for the next Windows release.

Couple other items. First. Always use Microsoft Windows, or just Windows. Please do not use AOLspeak or l33t speak when referring to these products. The quickest way to get your opinion or editorial dismissed is to refer to the product by a derogatory name or title.

In a case like this where you are trying to take the ... "higher ground" ... act like it. Do not play down to the Halo Players.

That being said, I think you have this screwed around the wrong way. AMD has been quite clear about what they thought of ATi's Linux drivers. AMD code engineers have also been very supportive of OpenSource in general... both with x86-64 and OpenBios/LinuxBios.

However... rewriting a new driver architecture from scratch isn't exactly an easy task. I'm not a coder, but I've heard the horror stories about driver development for the Intel HDA Azelia Audio... ATi's driver matrix is /worse/.

In recent months the change has been dramatic. Each Windows XP / Vista driver release for the past ... literal year... has been matched by a Linux Driver release : http://ati.amd.com/support/drivers/linux/radeonprevious-linux.html

within the past couple of months one of the driver releases was pulled, and a new "critical" bug was fixed in a matter of days... the first time ATi had ever done anything like that on the Linux platform.

Yes. The graphics situation isn't where we would personally like it to be. But, it's much better than it was even 2 years ago.

All that being said.. I am beginning to feel a bit like Dr. Breen from Half Life 2 every time I counter point somebody who is out to spread F.U.D. about graphics in Linux.

I've been banging on ATi for months... but... I don't see any Radeon HD support, several Radeon x1950 cards still won't work... and AIGLX support isn't anywhere to be seen.

Yes... I know that all of that will come with the new OpenGL driver. But... I don't know when that driver will hit... and each month that it doesn't get here makes another month that it is harder to hold my own position.

So... lets look at another analogy from somebody else who likes to take their time.

Nintendo. Now... compare Metroid Prime to Halo for a second. One is dull, lifeless, more bland than Brittish cheddar, and is known for collecting idiots who can't speak proper english. The other is held as a paragon of perfect gameplay, the ultimate FPS adventure.

PSo... why the difference? Well, aside from the fact that Bungie codes games about as well as Nvidia codes drivers, Bungie is on a time limit from parent Microsoft to get a product out the door. And it shows.

Nintendo... however.. takes their time. The product is finished... when it is finished. And it shows as well in all of their games.

In the same manner... I'd much rather AMD/ATi take their time and get the support right... as they did with Vista... take the time to properly test and retest under all conditions..

then to pull an Nvidia and release drivers that earn a reputation for being junk.

Doesn't make me feel any better though...

Source : http://www.mepislovers.org/forums/showpost.php?p=55345&postcount=2

... *whistles*...

suddenly Intel's code release makes a LOT of sense.

I guess I should expand on this. AMD's OpenSourcing of ATi's drivers isn't new, nor is ATi's support of OpenSource. Specifications by ATi are why the X.org Radeon Driver supports R100 (Radeon 7500) Graphics cards... cards that ATi never wrote drivers for.

It hasn't been a secret that ATi had long term plans to OpenSource their drivers, but where unable to do so because of patents and licensed technology. Previous timelines have set OpenSourced drivers at being a year or two out. This was one of the extreme concerns by Intel after the AMD/ATi merger, that AMD would move the timeline up. After Henri Richards comments, this seems to be that the OpenSource Drivers are on a fast track now with AMD.

Something else that has mainly just been theory so far is that the new OpenGL driver is supposed to be free of Licensed technology, and therefore candidate for opening the sourc.

Original Thread about Henri Richards speech is located here : http://mepislovers.org/forums/showthread.php?t=7254

Intel's response to AMD here : http://mepislovers.org/forums/showthread.php?t=7299

Source : http://www.0xdeadbeef.com/weblog/?p=288

ati + open source

The ATI marketing guy on stage (Henri Richard) at the Red Hat Summit just committed to fixing the ATI problems with open source. To paraphrase “most people are worried about what they will lose…IP, etc…we’re worried about what we can win.” They know it’s a problem and they are committed to fixing it.

The fact is, AMD is a lot more serious about Linux and Unix than you obviously bothered to research. Have you even bothered to drop by ATi's Linux Driver page? It's here : http://ati.amd.com/support/drivers/linux/linux-radeon.html Compare it to Nvidia's Driver page here: http://www.nvidia.com/object/unix.html Notice anything... different? The ATi page lists both an Off site wiki, it also lists a bug feedback form. Now visit the driver history pages. Here's ATi's : http://ati.amd.com/support/drivers/linux/radeonprevious-linux.html And here is Nvidias : http://www.nvidia.com/object/linux_display_archive.html Notice... anything different there? Oh yes. ATi has released a Linux driver every single month for over a year. Nvidia? Well... Look for yourself.

The fact is, whoever wrote the article to begin with did absolutly no research what-so-ever. Did not attempt to contact AMD. Did not ask on Phoronix. Did not ask on Rage3D. And you get the idea. Now, if you want, I'll be happy to re-write the blurb on AMD's support of AIGLX.Liaison-Rocks —The preceding unsigned comment was added by Liaison-Rocks (talk • contribs) 00:51, August 23, 2007 (UTC).

updated article notes

never mind, I went ahead and updated the information. Split the blurb into 4 separate sections; one for AMD/ATi; one for Nvidia; one for Intel; and a short rewording of the final blurb on XGL. Left the text above the section notated as contested in place, although some, if not all, of the information is probably duplicated in the lower section. Retitled Contested Section to indicate a split in the page. Added in multiple external links citing various facts that were grossly overlooked in the original blurb. Explained why there is no AIGLX support in AMD/ATi's drivers, something the original article did not bother to attempt. Left one Phoronix citation unlinked, specifically the performance comparisons of the X.org Radeon driver against the ATi Fglrx driver. The 3 external screenshots linked in were taken in Mepis 6.5.02 and Mepis 7 Beta 1. Intel opening statement could be classified as opinion, Intel has never officially stated in an online, print, or broadcast media that they opened sourced their drivers in retaliation for the AMD and ATi merger. I am of the understanding that unofficial statements are considered hearsay and to not be taken as fact. Left uncited the clarification that Intel does not have a policy in place on opening the source code to their graphics drivers and that future GPU's from Intel are may not have open sourced driver code made available; uncited because Intel's Press Room does not contain the information and other statements could be considered hearsay, off the record, or nonfactual. Left multiple points in Nvidia section uncited, particularly the numerous forum postings about bugs in the Nvidia drivers and problems with Pixel Shader 1.4 calls. Nvidia's lack of an official or linked unofficial bugzilla means that most problems can only be taken as hearsay and not official fact. It is my opinion that this is intentional on Nvidia's part. Clarified each companies involvement with Open Source in each section in order to provide background information the drivers. The IBM link refers to the original FireGl cards which used technology from the IBM Oasis GPU. IBM's technology was also used in later GPU's. The SGI reference is uncited due to SGI having filed for bankruptcy. The status of technology licensed from SGI and used in ATi products is in a state best described as unknown.Liaison-Rocks —The preceding unsigned comment was added by Liaison-Rocks (talk • contribs) 03:32, August 23, 2007 (UTC).

fixed the page layout so that See Also aligned properly with the edits. —Preceding unsigned comment added by Liaison-Rocks (talk • contribs) 22:04, August 25, 2007 (UTC)

Graph

I'm planning a nice color graph on the AIGLX - links between protocols, directions of communication, languages, etc. Be free to suggest your additions. The 1st draft:

 User program  
     |
 X server
     | 
 Graphic card

Konstantin Kosachev (talk) 16:23, 15 February 2008 (UTC)

Screenshots

I think this article has a couple good screenshots (and descriptions) of the problem AIGLX fixes: http://smspillaz.wordpress.com/2010/05/21/beware-the-benchmarks/. The author is Sam Spilsbury, primary compiz developer. —Darxus (talk) 06:35, 3 January 2011 (UTC)

It is now 2012

I took the liberty of removing sections that, although useful 4-5 years ago, are misleading and unmaintainable here in "the future". Assistance documenting AIGLX from a historical perspective would be greatly appreciated. I'll leave the tags for someone else to remove. Thanks! Krushia (talk) 00:03, 20 September 2012 (UTC)

I got the impression that AIGLX was a current thing from reading the article. Thankfully there is a "needs updating" tag on the top of the article, but I initially missed it. I'll do some editing on other articles to make it clear that this is legacy technology. 104.228.101.152 (talk) 17:22, 18 August 2020 (UTC)

Reliable media coverage

If you are intimately familiar with this subject and are aware of examples of non-trivial coverage from reliable third party publications, could you please add them to the article? I'm not having any luck. AMFMUHFVHF90922 (talk) 18:50, 28 April 2013 (UTC)

AIGLX is a display server?

AIGLX replaced Xgl, which was a display server. Therefore, AIGLX should be a display server too, I guess. Am I wrong? If not, then it should be part of Category:Display servers. The term "project" at the beginning of the article is too generic. Should be more specific: library, API, display server or something like that. The product's wiki is not exactly helping in understanding what kind of product it is exactly. It uses the term "technology", which is also too generic. —  Ark25  (talk) 02:56, 24 September 2013 (UTC)