Please create an account to participate in the Slashdot moderation system

 


Forgot your password?
Close
typodupeerror
×
Graphics Software Linux

ATI Releases AIGLX Linux Driver 113

Posted by kdawson from the making-compiz-sing-and-dance dept.
Michael writes "A month after AMD released a Radeon HD 2000 'R600' Linux driver based on their new Linux driver codebase, they have now released another driver that provides AIGLX support used for Compiz and Compiz Fusion. In addition to this long-awaited AIGLX support, this driver also addresses issues with previous Radeon product families, performance improvements, AGP fixes, and added features to their graphical control panel. Phoronix has a review of the 8.42 Linux driver with all of the details about this much-anticipated release."
This discussion has been archived. No new comments can be posted.

ATI Releases AIGLX Linux Driver More

ATI Releases AIGLX Linux Driver

Comments Filter:

  • So.. (Score:4, Funny)

    by Anonymous Coward on Tuesday October 23, 2007 @03:16PM (#21090621)
    ..will it run on Linux? Seriously...

    • Re:So.. (Score:5, Informative)

      by Kozar_The_Malignant ( 738483 ) on Tuesday October 23, 2007 @03:41PM (#21091005)

      ..will it run on Linux? Seriously...

      From TFA: "but there are a few words of caution. Be forewarned that there is a bug in Compiz 0.3 affecting the fglrx 8.42.3 driver and there may be a few other situations where Compiz or Compiz Fusion may not work immediately. The bug found in Compiz 0.3 and that's causing havoc with fglrx 8.42.3, has been resolved in Compiz 0.6. Next month in fglrx 8.43, AMD will be introducing a workaround for Compiz 0.3 support. "

      So that's a definite "Maybe".

  • Finally!!!! (Score:2, Funny)

    by slap20 ( 168152 )
    Well it will be nice to get my cheap ATI card out of vga mode at 640x480 at 4bit color.... Maybe one day I can move up to 8 bit color.
    • Well, my cheap ATI card runs at 1024x768 at 32-bit color with full 3d acceleration

      (Note: Mobility Radeon 9100, R200)
      • I have an ATI ES1000 (Dell SC440) that is supposed to have 2D acceleration, including xvideo. The driver reports working xvideo, but actually playing a video (on xine, mplayer, or totem) gives a blue screen where the video should be. (And the same software on older 2D only ATI cards works fine).
    • Well it will be nice to get my 256Mb ATi X1550 out of VGA mode at 1024x768 at 16-bit color.

      Btw I am NOT joking, thats what I am stuck at on Gutsy. Not a good look on a 22" Widescreen monitor

  • Bullshit... (Score:5, Informative)

    by MostAwesomeDude ( 980382 ) on Tuesday October 23, 2007 @03:33PM (#21090889) Homepage
    ...at least in my experience.

    Problem one: Doesn't actually work with Compiz. While AIGLX works, XComposite does not, and loading Compiz results in massive screen corruption. Joy.

    Problem two: Anybody who had XVideo problems before, will probably still have them now. Sad but true. Ditto with font selection and rendering.

    Problem three: While X.org server 1.4 is supported, Linux 2.6.23 is not. Anybody running on the bleeding edge is once again locked out.

    I'm sure more bugs will show up, but I'm pretty disappointed that they haven't improved the heavily broken XComposite support that they claim "works just fine."

    • Re:Bullshit... (Score:5, Insightful)

      by FMZ ( 1178473 ) <kj_sonny@@@hotmail...com> on Tuesday October 23, 2007 @03:37PM (#21090957)

      Problem three: While X.org server 1.4 is supported, Linux 2.6.23 is not. Anybody running on the bleeding edge is once again locked out.
      Umm... there's a reason they call it the "bleeding edge". Sometimes it hurts.

      • Re: (Score:2)

        by Hatta ( 162192 )
        2.6.23 [kernel.org] was released on October 12th. Once it's released it's not really bleeding edge, just leading edge. Working drivers for a released kernel is not too much to ask.

      • Re: (Score:1, Informative)

        by cachimaster ( 127194 )
        There are unofficial patches to make it work with 2.6.23 in the Phoronix forum. Also, i got compiz-fusion working (With a little tweaking) on Gutsy perfectly. So much for the "Ati drivers suck" meme.

      • Re: (Score:1, Informative)

        by Anonymous Coward
        [blockquote][blockquote]Problem three: While X.org server 1.4 is supported, Linux 2.6.23 is not. Anybody running on the bleeding edge is once again locked out.[/blockquote]
        Umm... there's a reason they call it the "bleeding edge". Sometimes it hurts.[/blockquote]

        Except the 2.6.23 is the previous *release* (stable version), 2.6.23.1 being the newest release, and 2.6.24-rc1 being the bleeding edge.

    • Re:Bullshit... (Score:5, Insightful)

      by DeadManCoding ( 961283 ) on Tuesday October 23, 2007 @03:38PM (#21090977)
      While the system may still be broken, it doesn't matter. The Linux community (myself included) has wanted better open source drivers for video cards, and AMD is finally making good on it. I may not be able to play bleeding edge games on my Linux system, but the software is getting there, and that's the important part.
      • Let me know as soon as you find those open source drivers you're lauding over on ATI's site. Neither source nor the binary driver appear to be available from where I'm sitting.

        • Separate stuff. (Score:5, Insightful)

          by DrYak ( 748999 ) on Tuesday October 23, 2007 @06:00PM (#21092857) Homepage
          TFA is about the current closed source ATI drivers, (the one downloadable from AMD's website).

          The GP was speaking about the opensource drivers, which is a different project [livejournal.com]. Anyway, AMD/ATI has promised to help them too, and is currently in the process of releasing specs, step by step. Currently they have provided enough information for the mode setting :
          - it's now possible to switch to a 2D mode using opensource drivers. Before that, VESA was the only working solution because of important change between the Radeon 2D architecture (up to R4x0 / Radeon X850) and the Avivo 2D architecture (from R5x0 Radeon X1x00 onward).

          Other specs will follow step by step. Anyway, you'll still have to wait at least 1 year befor good and stable opensource drivers for Radeon HD 2900 start to popup in your favorite Linux/BSD distro. The good news from today's article is that until then the current closed source drivers are ratter good.
          And AMD is promising to keep releasing specs for the opensource drivers project.

          • Re: (Score:2)

            by Inhibit ( 105449 )
            And apparently neither you nor the original responder use said drivers. The latest showing up on their site as of 9:31 EST is 8.40.1, not the driver this article purports to cover. Which is what I was why I was asking where the driver was.

            This makes four pages on how it almost works not worth much. Also, performance gains through increasing the artifact count don't do much for me.. although I'd be able to say for sure if I could get the driver.

            At any rate, you can check for yourself right here: http://at [amd.com]

            • Re: (Score:1)

              by Taleron ( 875810 )

              Ummm...not to rain on your fiery speech, but you might wanna check that you're looking at the right driver page. That link's for older Radeons like the R500/Radeon X1k, version 8.40.4. Not the R600/RadeonHD 2000 series, which is using 8.41.7. :)

              It's still not 8.42 like the Phoronix article's talking about, but if you want the drivers that bad it might help to bookmark the corresponding chipset page. Here's the right link (it's got a little "hd" in the URL) which is where it would show up; 8.41's still up

            • The different versions and URLs (Score:4, Informative)

              by DrYak ( 748999 ) on Wednesday October 24, 2007 @05:19AM (#21097427) Homepage
              • 8.42 : is currently *being* released, links are not updated everywhere. But a few google request may bring you to forums where it is already available [fuckinggoogleit.com]. For exemple, Phoroinix [phoronix.com] have published a link to the driver they did test. I think the release is not official yet because of the reported problems with 2.6.23 kernel. The same google search can also bring out patches to circumvent those problems [phoronix.com] and even howtos about using the new AIGLX [ubuntuforums.org] for desktop compositing.
              • 8.41 : Is the previous release. It was mainly centered around bringing RadeonHD support on linux. Thus some bugs may have managed to slip by with older chipsets. IT IS available on the ATI [amd.com] website. But it comes with a caveat [amd.com] explaining the situation, that this driver is mainly targeting Radeon HD and that it's "use at your own risk" with previous chipset generations. You're still free to try it on X800XL if you want (Phoroinix [phoronix.com] did it in their).
              • 8.40 : is the latest release using the older code base. Currently it is what has been the most widely tested and debugged for older chipset, so that's why it's the first thing you land on.
              • There's a nice wiki [cchtml.com] about ATI on Linux, with distro specific pages, links to the latest bleeding edge versions and such.


              GPL drivers are currently standard on most distribution for cards up to R4#0 (Radeon X8#0). If you want bleeding edge you can get them from freedesktop's git repository.
              GPL drivers for R500 and up are currently being created. You can get the currently couple of working pieces from its corresponding irregular devel companion [livejournal.com].

              You either have to wait more time until it's trivially offered as the first choice on the ATI selector (for the binary drivers) out of the box with major distros (for the GPL driver).
              Or you have to accept "bleeding edge" mean, understand that all those drivers are fresh from the oven, not thoroughly tested thus maybe not ready for the public at large, and that you need a little bit of google before assembling the necessary pieces, or use specialised resources like the afore mentioned wiki.

    • Re: (Score:2, Insightful)

      by AmaDaden ( 794446 )
      Keep in mind how new this is and how complex a GPU is these days. It's gonna be a mess for a few sometime. What I would like to know is who is more dedicated to making Linux drivers that are (eventually) rock solid? I'm planing to build a new rig soon. Should I go nVidiea because I know they have decent drivers now or ATI because once they get the kinks out it'll be better? I'm sure people are gonna say that ATI is better because they went Open Source. But nVidiea could open up any day now in response to al
      • I you still buy nvidia when ati is opening the source code of the drivers, why would nvidia ever open it's source code ?
        • Because nVidia provided adequate Linux support for years, while ATI (now =AMD for a year) fu***** up Linux users? Look at fglrx development, they didn't add a single important feature for years, the drier development looks like one-man-show.
        • last I checked all that had been released were 2D specs which aren't worth much given that we already have an opensource 2D driver.

          If ati releases usable opensource 3D drivers or even closed source ones that are better than nvidia maybe I will consider switching but not until then.

      • I've just gone with Nvidia. My reasoning was along the lines of I need to use the system now. If/when ATI/AMD get enough specs out and everything is coded correctly, I can ebay/craigslist the nVidia and switch to ATI. I'm not holding my breath, as someone else said GPU's are complex, it'll take time for it to stabilize (and yes I'd prefer the open source version as well but I'm not willing to work with an unstable/minimally functionally system either)

      • Re: (Score:2, Insightful)

        by Spy Hunter ( 317220 )
        ATI committed to providing complete documentation, but did not provide the source code to their current driver. This is (partially) because their current closed driver contains proprietary IP that ATI does not own, so they can't open-source it. NVIDIA is in the same situation and will have to take the same slow route to open-source drivers when they eventually come to their senses. They cannot "open up any day now" and "have a working code base" instantly, any more than ATI could.
    • Did you install xserver-xgl?

      That worked wonders for my Radeon XPRESS 200M on my 7.04->7.10 Ubuntu install
    • Maybe the solution for the bleeding edge problem is to have STABLE Linux kernel APIs for graphics cards. You know, just like how the rest of the world works.

      • Re: (Score:1, Flamebait)

        by Slashcrap ( 869349 )
        Maybe the solution for the bleeding edge problem is to have STABLE Linux kernel APIs for graphics cards. You know, just like how the rest of the world works.

        Fuck you. Move to a different OS and try to ruin that with your retarded suggestions instead. Alternatively fork the kernel and implement what you're asking for. The best bit is that if you're unable to do that, you are automatically unqualified to tell the kernel devs that they should be doing it.
  • "The Linux 2.6.23 support isn't found in fglrx 8.42.3 due to issues with x86_64 support. However, we do expect that fglrx 8.43 will contain the Linux 2.6.23 kernel support."

    Skilled folks, those ATI engineers..

    • Re: (Score:2)

      by Hatta ( 162192 )
      So, because there were issues with x64 on 2.6.23, everyone with 2.6.23 has to suffer? How hard would it have been to leave in 2.6.23 support with the caveat that it won't work on x64?

      • Re: (Score:1)

        by pairo ( 519657 )
        So because there are issues with x64 on 2.6.23, everyone with x64 has to suffer? How hard would it have been to leave in x64 support with the caveat that it won't work on 2.6.23?
        The difference between a kernel version and the architecture of your CPU is... well, it's easier to downgrade your kernel than install a new system.
    • Keep in mind that ATI:s development cycle for a given driver spans over 2-3 months, with the last month for testing (i.e. no new features). 2.6.23 was released less than two weeks ago. I think it is better they say that they don't support 2.6.23, wait to the 8.43 driver and make things correct right away, instead of rushing support and testing which would remove resources from other stuff.
      Just because someone releases a new version of a software it doesn't mean that all those that use it must support it ri

  • Am i only one? (Score:2, Interesting)

    by Fri13 ( 963421 )
    I'm having problems with Nvidia and ATI drivers on windows and on GNU/Linux side. With different motherboards and different 3D cards and with different drivers. Many users bashes ATI because they old driver style. My friend dont have any issues with ATI on windows but linux side he has. I didnt have 9700 on Linux side but with windows i did. Now on 8600GT i get so much problems that it's just enought just to mention that i have them. Best drivers what i have seen has come from Intel and #2 position is ATI
    • I have ZERO problems with my nVidia 8800GTS on Ubuntu Gutsy (and previously Feisty) x86-64 and Windows XP x64 SP2.

      I used to have an assortment of ATI cards, but none of them have been without issues.

      My X800GTO had ram trouble, and had to have the ram underclocked to run stable. I had a dead, followed by a working X1900XTX. With that card, in Windows x64, suspend (mode 3) didn't work more than once between restarts, among other issues. In Ubuntu Feisty, I had to alter xorg.conf to enable DRI (hardware accele

  • Suspending? (Score:5, Interesting)

    by The Analog Kid ( 565327 ) on Tuesday October 23, 2007 @04:07PM (#21091387)
    Can you suspend your laptop using these drivers. There was a problem using the 8.40.3 drivers and any distro that uses the SLUB allocator that causes the system to hang while trying to suspend. Since most distros (if not all) are moving to SLUB this is a pretty big issue, ATI is usually behind the 8-ball though so I'm not getting my hopes up.
    • From my (limited) testing, the same problem exists. This was pretty much the only thing I was waiting for, so it looks like I'll cross my fingers and wait for 8.43.
      • The SLUB problem was that laptop would never complete suspend. The manifistation of it was that the half-moon would just keep blinking and never go solid. Now with 8.42 (mind you, hacked for a FireGL PCIID), it completes suspend. Resuming however, has yet to return me a working screen. However, if I kill X with alt-sysrq-k, and I can set capslock and change VTs, indicating it almost works. I'll play with the resume scripts, but it appears that the SLUB-blocking-suspend was addressed, but for some stran
        • Looks like I have a bit more experimenting to do then. I'll try a few more things and see if I can get it to work.

          • Let me know... (Score:3, Informative)

            by Junta ( 36770 )
            So far, I've gotten it to suspend, and resume successfully in Gutsy, *but* the catch that makes it useless, is I essentially make /etc/acpi/resumed/65-console.sh *not* attempt a single chvt, and so I can never get back to X. So the kernel seems fine, but X won't come back and chvt will hang in the process. Have tried saving vbestate and post_video, and neither, but of those two there is no success. The graphics did work with the vbestate saving and post_video in text consoles though....

        • Re: (Score:3, Interesting)

          by Wizarth ( 785742 )
          Interesting, I've noticed since upgrading to Gutsy my laptop (with ATI video chipset) wouldn't suspend, would shut down the PCMCIA WiFi card but then leave me at a blank screen (blinking cursor, I think, not in front of it right now). I hadn't played with it yet though. I thought it might be because I modified the sleep/resume scripts slightly, when I was having some issues with said wifi card, ndiswrapper, WPA and sleeping.
    • The new driver seems to fix my suspend *and* resume problems. Gutsy, IBM T42 with Radeon Mobility 9600.
    • Unfortunately, no.

      I'm crossing my fingers hoping that an astute developer who knows the power management ins and outs of the kernel, X, and the suspend/hibernate routines figures out what the problem is in the next couple of days.

      Hopefully it's a simple fix, although those always seem to be the hardest fixes, eh?
  • I've been waiting for updated ATI drivers for non-x86 platforms (e.g. PPC) for a while now due to problems with the open-source existing DRI drivers being pretty crap on my specific card (iMac G3 with Rage 128).

    Does anybody know if it's likely that i'll be able to get these working, or is all the GL stuff wrapped up in proprietary x86 only code?

  • I don't normally reply twice, but... (Score:5, Informative)

    by MostAwesomeDude ( 980382 ) on Tuesday October 23, 2007 @04:39PM (#21091815) Homepage
    ...this is not an open-source driver.

    There are three ATI drivers. There is fglrx, which is this driver that was just released. There is radeon, which is the open-source driver that controls Rages, R200s, R300s, and R400s. And there is radeonhd, which controls R500s and R600s.

    fglrx has many issues. It now has AIGLX, but it still has broken XComposite. Xvideo doesn't work for many people. Direct 3D rendering is slower than on Windows. The entire driver is closed-source and shims a binary blob into the kernel. But, it still offers 3D for R400, R500, and R600 chipsets.

    radeon is the dependable open-source driver for older Radeon-based and Rage-based cards. It works excellently, with direct rendering for all chipsets up to the R200 series. People are working on R300/R400 direct rendering right now; see http://tirdc.livejournal.com/ [livejournal.com] .

    radeonhd is a brand-new open-source driver that controls new R500 and R600 cards. It has no direct rendering yet, but there is a promise from ATI/AMD that documents pertaining to direct rendering will be released sometime soon without NDA. This driver is still being worked on, but it offers satisfactory 2D for many people.

    • Direct 3D rendering is slower than on Windows [with fglrx].

      If it's still slower, it's only by a tiny margin now. fglrx 8.42 is *hugely* faster than 8.40 and below. My Radeon X1400 now runs Doom 3 at 37.1 fps (1024x768, medium quality), compared to the 21 fps it got before with the same settings.

      fglrx still has issues, but its 3D performance is no longer one of them.

    • Hi there,

      Good post. s/D3D/OGL. Probably was a typo/brainFart.
       
      BBH
      • XD, didn't even catch that. "Direct 3D rendering" should be "direct rendering, e.g. 3D," not "Direct3D rendering." Thanks.
  • Compiz doesn't work. That's what we waited for for over a year?
    • Actually, it worked for some of us.

      My card is a Mobility Radeon 9000, and after the upgrade to Gutsy, compiz wouldn't work, neither with ati/aiglx, nor with xgl/fglrx. However, using this driver made compiz load!

      My problem is I used the install method, and did not create a deb package, because using this option creates a really tall window, in which I can only see redhat and suse. Maybe ubuntu was somewhere at the bottom of the window, but it didn't fit inside my screen... :(

  • Does AMD have a developer/beta channel? (Score:3, Interesting)

    by r_jensen11 ( 598210 ) on Tuesday October 23, 2007 @04:52PM (#21092009)
    Because I think then many of these issues could get resolved more quickly. By "these issues," I mean things like better Compiz support and more modern kernel support. Fortunately*, I don't even upgrade my kernel more than two or three times a year (unless there's some major feature or fix that I happen to read), so not supporting the latest & greatest kernel isn't a major issue for me.

    *I know that this leaves me vulnerable to security exploits, but I'd rather take my chances with that than not being able to use my programs at all because of an incompatibility. Plus, I've gone through upgrading after every kernel revision and it just gets tiring. There are many systems that are up for a longer period of time than whenever each kernel upgrade is released, so I have a feeling I'm not alone here, either.

    • Re: (Score:2)

      by Harik ( 4023 )
      How many KERNEL security exploits were released in the last year?

      Some occasional obscure driver bullshit, but I can't think of any in the parts of the kernel everyone will have.
      • Looks like a troll, smells like a troll, probably *is* a troll.

        At the risk of feeding you:

        Two examples of exploit lists from different times:
        http://www.novell.com/linux/security/advisories/2007_51_kernel.html [novell.com]
        http://lwn.net/Articles/118952/ [lwn.net]

        To say that Linux has no security exploits is just plain stupid and naive.

        • Re: (Score:2)

          by Splab ( 574204 )
          Speaking of trolls, second link was to a 2005 exploit, so that doesn't really count as being within last year does it now? And the other link looked like a SUSE problem, so who is trolling again?
          • The point wasn't to point out flaws from within the last year, but that the kernel *does* have flaws and isn't perfect. I'm sure there will be a security flaw that pops up within the next year. Am I concerned? No, not terribly, because I have lots of layers of security and am confident that I'll be able to upgrade to a kernel that solves said flaw before I am likely to get hit by it.

            • Re: (Score:2)

              by Splab ( 574204 )
              But the GP statement was for stuff within last year. Just because something used to suck/be buggy/explode doesn't mean they haven't fixed it and gotten on the right track, so when someone gives a specific time frame it makes no sense to counter it with some data from outside that frame.

  • So what's the estimate... (Score:4, Interesting)

    by Amphetam1ne ( 1042020 ) on Tuesday October 23, 2007 @05:59PM (#21092831)
    ..until I can get hardware H.264/AVC1 decoding in LinuxMCE using my Radeon HD card then? Months? Years? the day after the next big encoding standard is released?

  • Goati? (Score:4, Funny)

    by Elyscape ( 882517 ) <elyscape@gCHICAGOmail.com minus city> on Tuesday October 23, 2007 @06:08PM (#21092957) Homepage
    Am I the only one who saw the tag "goati", became confused, and read "goatse"?
  • This driver looks nice with a few improvements over the old one.

    However I was hoping to get their 3D API so we could write our own. When's that going to happen?

  • URL (Score:2, Informative)

    by detain ( 687995 )
    ati.com was still linking to the old driver for me, im not sure if it is for everyone else, however, this URL
    should let you download the new driver:

    https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/linux/ati-driver-installer-8.42.3-x86.x86_64.run [akamai.net]
  • When they release an Open Source driver for a newer card, let me know. There occasional announcement of vague future plans just don't cut it. In the meantime I'm going to back to sleep.

    p.s. Oooh! I just dissed ATI! Expect this post to be modded into oblivion within ten minutes.
    • When they release an Open Source driver for a newer card, let me know. There occasional announcement of vague future plans just don't cut it. In the meantime I'm going to back to sleep.

      What, like the RadeonHD driver for R500/600 being developed by FreeDesktop.org based on the specs that ATI released?

      p.s. Oooh! I just dissed ATI! Expect this post to be modded into oblivion within ten minutes.

      Personally I've always been modded up for dissing ATI. I'd like to think that you'd get modded down for not actually k

      • What, like the RadeonHD driver for R500/600 being developed by FreeDesktop.org based on the specs that ATI released?

        Oh, you mean the RadeonHD driver does 3D now? 'Cause otherwise, it's still entirely fucking useless.

  • At least the new overlords of ATI (AMD, for those not in the know) are actually making a halfway decent effort to support nix. While granted, they have a LONG way to go to meet even the quality of nVidia's drivers, at least its a step in the right direction, instead of the old "We'll make one, but we're really not going to put any effort into it" attitude. Hey, sometimes we have to be thankful for even small miracles.
  • Can't find the link to the download in the Phoronix review or anywhere on ATI's web site. Anyone have it ?
  • Glad to hear linux AGP support is getting attention. The AGP support for Direct 3D in Catalyst 7.8, 7.9, and 7.10 is broke.

    Seth
  • Congratulations, ATI has just wasted another two months, trying to catch up with the competition, and ended up releasing a driver that does not even support AMD's x86_64 architecture.

    Obviously the ATI developers are totally incompetent and I can only wonder how long AMD's management is going to tolerate this. At least I have learned my lesson and will make very sure not to buy any hardware from ATI again. Even if this means that I also have to buy a Intel processor.

    Tom

Slashdot Top Deals

Keep up the good work! But please don't ask me to help.

Close