Forgot your password?
typodupeerror
AMD Graphics Open Source Linux Hardware

AMD Releases Open Source Fusion Driver 126

Posted by kdawson
from the knock-yourselves-out dept.
An anonymous reader writes "Yesterday AMD released open source Linux driver support for their Fusion APUs, primarily for the first Ontario processor. As detailed on Phoronix, this includes support for kernel mode-setting, 2D acceleration, and 3D acceleration via both Mesa and Gallium3D graphics drivers."
This discussion has been archived. No new comments can be posted.

AMD Releases Open Source Fusion Driver

Comments Filter:
  • SALE ? (Score:1, Interesting)

    by Anonymous Coward
    Aren't too many Phoronix news lattely ? Are they on sale ?
  • Fusion (Score:1, Insightful)

    by Anonymous Coward

    Fusion is going to be important. AMD will finally have a portable product that rivals Intel. Integrated video hardware is now commonplace on the desktop [dell.com]. Embedded AMD hardware is beginning to appear [innocoregaming.com] and Fusion will accelerate this.

    Intel doesn't have a 3D core they can integrate onto the CPU die. Bottom line is AMD has an edge.

    • Re: (Score:3, Insightful)

      by hedwards (940851)
      Intel doesn't have a 3D chipset they can integrate onto the processor die largely because they'd need to have a competent 3D chipset to start with. They haven't gotten that right up until now, so it's not a shocker that they haven't managed to get a competent one on die.
      • Re:Fusion (Score:4, Informative)

        by RightSaidFred99 (874576) on Tuesday November 23, 2010 @02:29PM (#34320858)

        huh? [geek.com]

        Double huh? [anandtech.com]

        It's rare to read someone post something both factually and subjectively wrong at the same time in so few words. Congratulations.

        • by TheEyes (1686556)

          1) Sandy Bridge doesn't exist yet, and won't until next year. It'll be great when it does exist, though.

          2) The GP was probably talking about OpenCL support, which is generally lagging on Intel IGPs. Apple is into OpenCL in a big way, and the lack of OpenCL support was reportedly one of the reasons they stayed with the aging Core2s (paired with NVIDA GPUs) for the current generation of Macbook Air. Maybe we'll see OpenCL support next year with Sandy?

          • Well, he was saying Intel doesn't have a "3D chipset they can integrate onto the processor die" when in fact they have already done so. And if we're talking about released products, where's Fusion?

            And actually their IGP is reasonable these days, even pre-SB. It's not fit for high res gaming, of course, but it's suitable for a majority of non-gaming needs.

            I don't think SB will have OpenCL support, from my limited understanding Intel's GPU solution is less flexible than ATI/NVidia's.

      • by TeknoHog (164938)

        I'm using an Atom D510 right now, with on-die integrated graphics. Of course, these are not GPUs in the modern sense of "general processing unit", but at least it is an Intel 3D chipset on the processor.

        http://en.wikipedia.org/wiki/List_of_Intel_Atom_microprocessors#.22Pineview.22_.2845_nm.29_2 [wikipedia.org]

    • What about a vertex?
  • by erroneus (253617) on Tuesday November 23, 2010 @01:33PM (#34320032) Homepage

    Long ago, I went with ATI video because it had the best support for Linux. Eventually, NVidia caught on to this trend and started supporting Linux too... and better than ATI. So I switched. Now NVidia has screwed the community that had helped it to grow in popularity by putting out "Optimus" hybrid graphics everywhere and then refusing to update their Linux drivers to support it and refusing to release any details about it either. So now, the best anyone had been able to do is disable the nvidia GPU to reduce power consumption in laptops not able to utilize the nvidia hardware.

    As AMD/ATI is doing this, perhaps my next selection will be to the exclusion of NVidia (again).

    When will these jerks ever learn? The future of computing is in embedded devices and those devices will run Linux. Get Linux developers using YOUR hardware and it will have a better shot at a prosperous future as well. So far, Intel and ATI are the only options.

    • by hedwards (940851) on Tuesday November 23, 2010 @01:49PM (#34320236)
      nVidia is the last man standing in a sense. Both Intel and ATI (Obviously now owned by AMD) have released or are releasing pretty much everything necessary to have native drivers for whatever OS one wants to use. At some point they'll likely give up on that as more and more geeks decide that they don't want to recommend something that's limited like that.

      Not so much with cutting edge gaming rigs, but with older computers especially it's fairly common for video cards to outlive their manufacturer support and still contain a few bugs or optimization problems.
      • Re: (Score:3, Insightful)

        by Kjella (173770)

        That said, nVidia has given very long support on old graphics cards. The primary reason that they support Linux is OpenGL workstations, which demand long support cycles and regular users get the benefit. And as even AMD admit, you get bigger benefits from cross-platform code (Win/Mac/*nix) than you do from the open source collaboration, as long as it's not possible to open up the closed source drivers due to DRM licensing, licensed code and so on. The open source team is very small compared to the Catalyst/

      • Re: (Score:3, Insightful)

        Arguably, it might actually be getting less likely that Nvidia will ever provide decent OSS support...

        Intel has all-but-formally-announced their intention to lock Nvidia out of everything they can, as fast as the feds will let them. On die-video, no QPI licence, trimming PCIe lanes off lower end products, etc. AMD has not been as frankly rude about it; but their on-die video will be even more competent than Intel's, and they control a smaller slice of the market, in any case. Pretty much across the board
      • When do you think we'll get VDPAU support?
    • by RotateLeftByte (797477) on Tuesday November 23, 2010 @01:50PM (#34320252)

      NVidia had their opportunity but since AMD got their ATI dept's act together their GPU performance and importantly their Linux support has come on in leaps & bounds.
      With NVidia being squeezed out of the chipset market by AMD & Intel and even the consumer Graphics card able to play most FPS games at more than adequate frame rates, I see (sadly) NVidia slowly but surely going the way of Novel's Netware. i.e. to an inevitable death.
      They really need to buy an ARM user and get their GPU's into mobile devices, provided they can make them sip power instead of gulp it like a 6ltr Dodge Charger

      • by ArcherB (796902) on Tuesday November 23, 2010 @03:39PM (#34321856) Journal

        They really need to buy an ARM user and get their GPU's into mobile devices, provided they can make them sip power instead of gulp it like a 6ltr Dodge Charger

        Doesn't NVidia make the Tegra/Tegra2 processors for mobile devices?

        • Yeah, I'd forgotten that. But like AMD & Intel, their market share is miniscule. Most other CPU/GPU builders are far better placed in the market.

          The Register has a bit about the ARM A15 and a 1U server rack with the current A9 CPU inside. Now this is interesting. Low power Powerful servers. I use a couple of EEEBoz boxes but the A15 has certainly made me think again about dumping X86 CPU's for this use.

      • by antdude (79039)

        So for an ATI Radeon 4870 video card (512 MB) in Debian/Linux, will I have easy, good support, and speeds with ATI's closed binary drivers like NVIDIA's? I am currently using my old NVIDIA GeFVorce 8800 GT and GeForce 5200 FX video cards in my old Debian/Linux boxes. They work well even with games, Google Earth, 3D screen savers, etc. I do want speed. I am not crazy about the open drivers since they tend to be slow and I do 3D stuff. :(

    • Normally I'd shake my head and walk away. However, with the Boxee Box and other embedded devices running on Atom and other low end x86 CPUs and Linux, having the community bash out bugs in your drivers means that nVidia can get to the top of the embedded heap.

    • by diegocg (1680514) on Tuesday November 23, 2010 @02:05PM (#34320476)

      Time? You are late. ATI has been releasing specs and employing engineers to write opensource drivers for some time already. I haven't bought a Nvidia GPU for years, and I have no plans to do it in the future.

      • by Compaqt (1758360)

        I had considered ATI but heard bad things about the driver situation at the time (open source problems hadn't fully been worked through at the time).

        How is $YOUR_LINUX_DISTRO these days with open-source ATI drivers? Any problems with Linux games, other programs requiring 3D (like KDE Stars), and Windows games running on Wine?

        • Re: (Score:2, Troll)

          by ArsonSmith (13997)

          Any problems with Linux games

          Nope, they both work.

          I know, "both", we all know the obvious one, but can you name the other one?

        • I regret getting an ATI for my desktop at work. (yes Linux)
          - Using 2 monitors is sketchy at best... the performance dropped with the 2nd monitor. 3D support is OK on one monitor, but not so good on dual. I miss the cube. Xinerama or separate desktops?! that's crap ATI! Twinview is better from NVidia.
          - I have frequent system crashes because of the proprietary video driver (once a week is frequent), and have to hand configure the xorg.conf to remove crap from the display, etc. Waste of time.

          The OSS for the vi

        • by hfranz (101040)

          On my laptop that has an ATI Radeon Mobility X-1800 chipset (R300 series) support by the open source driver is basically OK. Desktop Effects in Gnome/Compiz work well, although not all effects are available. The subjective performance is good. With recent linux kernels there also is support for power management of the graphics chipset.

          BUT:

          1.) VGA output does not work. The video signal is distorted.
          2.) With KDE 3.5 this driver/chipset configuration is blacklisted for Desktop Effects.
          3.) The machine freezes

    • Re: (Score:1, Interesting)

      by Anonymous Coward

      It's always been time to 'move away' from nvidia.
      I asked years ago which card to use. Linux gaming fanboys said go nvidia, their closed blobs have great 3d support.
      WTF do I care about that?!?! When they decide to release new models and want to force everyone to upgrade, they
      quit maintaining the old blobs. Therefore, I'm left in simple open source 2D land with both vendors. Plus some marginal 3D here and there with both, but whatever. And further, when that blob starts crashing as I update
      around it, expose b

      • Re: (Score:3, Insightful)

        by FreonTrip (694097)

        I'll grant that the situation's always sucked for non-x86 platforms, but Nvidia's done a remarkable job of supporting their older hardware in Linux.

        Drivers for GeForce FX Cards, Updated 10/18/2010 [nvidia.com]

        Drivers for GeForce2-4 Cards, Updated 11/16/2010 [nvidia.com]

        Drivers for the Riva 128 (?!) through GeForce256, Updated 08/04/2010 [nvidia.com]

        There are also supported drivers for all of these products for AMD64 Linux. It's no substitute for an open source driver - I support nouveau - but declaring that they leave their old cards un

        • Notice that the release notes for the 96.43.19 version, released a week ago, includes "support for X.org xserver 1.8 & 1.9". Yes, xserver 1.8 wasn't supported until a week ago. It was released in April, seven months ago. In other words, if you wanted to run a current xorg for over half of this year with a Geforce 4 and the binary blob, you were out of luck; and their official position (which they fortunately reverted) was "we still support those cards, but only on xserver1.7.999". This is not "remarkabl
          • by tyrione (134248)

            Notice that the release notes for the 96.43.19 version, released a week ago, includes "support for X.org xserver 1.8 & 1.9". Yes, xserver 1.8 wasn't supported until a week ago. It was released in April, seven months ago. In other words, if you wanted to run a current xorg for over half of this year with a Geforce 4 and the binary blob, you were out of luck; and their official position (which they fortunately reverted) was "we still support those cards, but only on xserver1.7.999". This is not "remarkable support" at all.

            Note that Debian has nothing but experimental support for 1.9 and just recently only stable support for 1.8, while being 4 versions of the kernel behind in Sid [still stuck on 2.6.32]. Yes, they just released 2.6.36 in experimental but basically view it as ``SOL if something goes wrong because we're busy getting another stable out whenever the hell we figure out the time for it to be ready to come out...'' I'm stuck waiting for even Nvidia's 260.19.21-1 [only got a build because a few of us bugged them to

            • by FreonTrip (694097)

              Quite right. As Steve Max noted in the grandparent it can take time for new X server support to manifest in Nvidia's drivers for older products; however, adding that support is a non-trivial undertaking, and I daresay that most systems running pre-GeForce 6 hardware probably enjoy relatively stable and conservative Linux configurations. While that's still a nettling inconvenience for some, I'll take it over the relatively uneven and sometimes slipshod support available for other manufacturer's cards in Li

      • by AvitarX (172628)

        From what I've seen in forums nouveau is not so bad actually.

        I like that the team is working hard on using Gallium effectively (for example they used it to make VA-API work using VDPAU (decode only)).

        As far as I can tell the nouveau driver is more feature complete than the open source ATI driver, though maybe not as performant (I'm waiting for Open Source VA-API support for h.264 with encoding and decoding, I am not too concerned about the overall performance if it can run compiz).

        Intel appears to be the cl

    • Re: (Score:3, Insightful)

      by blair1q (305137)

      I don't know about "time to", but in any case where the software is open vs. closed, the open-source community will not make the effort with the closed system. This will absolutely make linux hackers choose AMD graphics now, which will almost certainly result in improved reliability of AMD cards in linux systems overall, and eventually almost total domination of the consumer linux segment by AMD graphics.

    • Re: (Score:3, Interesting)

      by drinkypoo (153816)

      I am currently typing on a Gateway LT3103u, which has also been sold by Everex among others. It has a 1.2 GHz "Athlon L110" and AMD RS690M chipset, which in turn contains a Radeon R2xx chipset. The "fglrx" driver does not support it as it is too old, and the open source driver causes massive display trashing and lockups. Whatever they did to it, it's not a faithful R2xx, and so the existing driver (which works on genuine, old R2xx stuff) does not work on it. But that's not all; AMD also didn't bother to con

      • Re: (Score:1, Interesting)

        by Anonymous Coward

        I am currently typing on a Gateway LT3103u, which has also been sold by Everex among others. It has a 1.2 GHz "Athlon L110" and AMD RS690M chipset, which in turn contains a Radeon R2xx chipset.

        The GPU in rs690 is actually a 4xx variant, not 2xx. Are you using the power saving features described at the bottom of http://www.x.org/wiki/RadeonFeature ?

        • Re: (Score:3, Interesting)

          by drinkypoo (153816)

          The GPU in rs690 is actually a 4xx variant, not 2xx.

          Not according to the X driver.

          Are you using the power saving features described at the bottom of http://www.x.org/wiki/RadeonFeature [x.org] ?

          It's the CPU power-saving that AMD did not contribute to Linux, not the GPU. I can't USE the GPU long enough under linux to get to the point where I need power saving.

  • by Anonymous Coward

    Drivers that finally enable full capability of the hardware are a must, be that OSS or closed source.

    nVidia has a long term support in their Linux drivers - they are the same as their Windows or OS X drivers, just added a GPL glue layer. Are the AMD drivers going to be long term supported too? Stable??

    To me,

        stable > long term support >>> OSS > closed-source

    only because I'm not planning on debugging video drivers!

    • Re: (Score:3, Informative)

      by hedwards (940851)
      Assuming that you're on a blessed platform. I remember waiting for nVidia to release their drivers for 64-bit Linux. It took a really long time from when I started waiting, and I waited a few years before getting an AMD64 system.

      But now there is support for Windows, OSX and Linux. If your OS isn't on that list then they aren't providing you with anything, or even the ability to do it yourself without doing some real funky stuff with wrappers.
      • by AvitarX (172628)

        There was GPL glue to get them working on AMD64 pretty quick in my memory, It worked fine, and was pretty easy.

        I actually can't verify it as faster than 1 year, but it was available when i purchased my computer in sept '04, with the first clawhammer released sept '03.

        Not an official driver, just a little fragment to make the official one work (it was a patched NVIDIA installer, so easy to do).

    • The AMD drivers AFAIK are supported almost entirely by the community. There are some devs at AMD that contribute to the drivers, and AMD has been releasing specs and documentation in increments. 3D is still not fully supported on most chipsets, so you need the FGLRX drivers for that, but for normal 2D desktop operations the radeon drivers are actually faster and more stable (fewer artifacts, etc.). At least, that's been my experience. The radeon driver also supports KMS (FGLRX does not) meaning it's the

    • Re: (Score:3, Insightful)

      by koolfy (1213316)
      Long term support is something that only exists in OSS ecosystems. No matter how long a company is going to try to support a hardware, the community will support it longer.
      It all comes down to "supporting old stuff does not bring new sales, therefore is really bad in the long term" vs "I still use/want to use old stuff, therefore I want it to work, and as long as it fits me, I'll support it."

      In the OSS community, the only hardware that's not supported is really the hardware that's not used. Hell they ev
      • The situation is arguably a little more nuanced than that...

        You can get long term support in closed ecosystems, even very closed ones(just ask the nice chap from IBM Mainframe sales...); but it'll cost you. Often a great deal and you had better be sure that the vendor is contractually on board; because they can, otherwise, pull the rug out from under you at their option(If they stop selling product X licenses/support contracts, you pretty much have to stop using product X. Game over. Copyright law. Have
  • by Anonymous Coward on Tuesday November 23, 2010 @01:56PM (#34320340)

    I went to buy an Ontario processor, but cheaped out -- I ended up with a Quebec processor. Now, I can't understand a thing it says, it never seems to do anything, and I keep having to give it money!

  • Wait, I thought fusion was still (perpetually) a few decades away from viability.
    • by Noughmad (1044096)

      We were all waiting for the years of "Fusion" and "Linux on the desktop", but instead we got "Linux on the Fusion".

    • by ArsonSmith (13997)

      No, net power gain from fusion is still a few decades away.

      3d Video cards have been sucking in power to create a small fusion reaction in their GPU for at least a few years now.

    • by blair1q (305137)

      That's just what AMD marketing wants you to think. They chose the "Fusion" trademark so they could perpetually delay their products (they're already pushing 2 years late compared to the date they first announced after buying ATI).

  • Darn... for a second there I thought I was going to finally be able to retrofit my car with a Mr. Fusion unit.
  • Originally computers were huge proprietary things.
    Now they are small and commonplace.

    In the past software was written to a specific hardware, now it's not ( C is cross platform compared to assembly, folks ).
    Games no longer draw graphics by directly reading and writing raster data directly into hardcoded "video memory" regions.

    Abstraction layers (such as a graphics API + Drivers) are a must in todays software environment. Why? To support cross platform software development. Many of todays games sit on top

    • by blair1q (305137)

      Uh, software is still written to specific hardware. You may write it in C, but C doesn't determine the register mappings and semantics. *addr=value is still just mov [addr],$value

      • Re: (Score:3, Insightful)

        by VortexCortex (1117377)

        Uh, software is still written to specific hardware. You may write it in C, but C doesn't determine the register mappings and semantics. *addr=value is still just mov [addr],$value

        Yes, but largely NO!

        I write my software in C. The same code compiles and runs on x64 and on x86. The COMPILER translates my cross platform "*addr=value;" into the apropriate machine level instructions. My C software programs do not concern themselves with the specific register mappings and processor semantics; this has been abstracted away by the C Compiler.

        I agree that driver software may be written to the specific hardware, but the purpose of a driver is to abstract said "register mappings and semantic

        • by blair1q (305137)

          My C software programs do not concern themselves with the specific register mappings and processor semantics; this has been abstracted away by the C Compiler.

          Your compiler knows jack about the registers on a video card.

          If you're writing a video device driver, as here, you're going to be poking values into board-register addresses, streaming data in and out of specific port addresses, and that code isn't portable. It may seem very familiar from board to board, but unless the features you implement are completely trivial it's not. Even at the library or application level you'll have to deal with the different features of the cards, which means a whole new set of

    • As an interesting contrast, software used to be all what we'd consider open source -- BSD-style open. The relatively few programmers around back then were mostly academics and openly shared code until folks like Paul Allen and Bill Gates stepped in and decided they should be able to charge for Altair BASIC (see Bill Gates' Open Letter to Hobbyists). You could also see the stirrings of the open source rebellion, too, with the freely published Tiny BASIC design descriptions with full source code, which many
  • There's still a niche that isn't very well served, where these low-power Fusion CPUs appear they could kick some major ass: the always-on-24/7 lightly-loaded server. I'm currently using Athlon II xxxe for this, but I'd happily downgrade processing power in exchange for lower wattage.

    Shit, Atom would be good enough, if the motherboards had enough SATA ports or slots for me to add SATA cards, but I never found any that did. Gimme a 9W or 18W processor on a board that I can somehow hook up 8 drives to, and

    • What you describe is basically a modern NAS appliance. A low-power chip, often an Atom, running an x86 OS, and with lots of SATA interfaces for drives. The only real difference between that and a low-power server is in the interface: Server runs a full OS and software stack, while the NAS runs a minimalist stack with a web interface for all configuration. I've got one of those NASs up im my loft, a QNAP brand. The thing runs linux and samba. It does have an option to SSH in, disabled by default.
    • by amorsen (7485)

      You get quite far with a single 2.5" disk these days, and that's easy to fit in every tiny server. 8 spinning disks is niche.

      • by Sloppy (14984)

        One drive is too few. All mechanical drives must be used in RAID1 pairs. (Or RAID5 if you're super-stingy and have extreme disk space requirements but IMHO that's almost always the wrong answer.)

        Then, for PVRs, I figure the ideal number of pairs (mythtv storage groups) is tuners+1, so you can record and play without any seeking or fragmentation of the recordings, thus 6 drives is the ideal for an hdhomerun-fed system. (The +1 is a simplification; it can theoretically rise to approach the number of fronte

  • Last big announcement about an AMD code drop, there was still something missing, though I don't remember what. Features, performance, whatever, there was still something not there that was either present in proprietary AMD drivers or nVidia drivers.

    Are we past that yet? Is it finally time to dump nVidia for AMD?

  • Let me know when I can buy a GPU where every single feature of the card (INCLUDING the on-board dedicated circuitry for decoding video) can be used in the open source drivers and then maybe I will care...

"Consistency requires you to be as ignorant today as you were a year ago." -- Bernard Berenson

Working...