Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
AMD Graphics Software Linux

Future AMD GPUs To Be More 'Open-Source Friendly' 180

Posted by Zonk from the finally-get-some-good-games-going dept.
skaroo writes "Phoronix is reporting that future AMD GPUs will be more open-source friendly. After AMD started releasing their GPG specifications to the open-source community, questions arose whether there would be information covering the Unified Video Decoder (UVD) found on the Radeon HD 2000 graphics cards. The UVD information is needed in order for hardware-accelerated video playback, but it likely cannot be opened due to DRM. However, an AMD representative said that moving to a modular UVD design is a requirement for future GPUs and that they will be more open-source friendly. They will also be opening the video acceleration information for their earlier graphics cards."
This discussion has been archived. No new comments can be posted.

Future AMD GPUs To Be More 'Open-Source Friendly' More

Future AMD GPUs To Be More 'Open-Source Friendly'

Comments Filter:

  • I remember a time... (Score:5, Interesting)

    by mangu ( 126918 ) on Sunday December 30, 2007 @06:39AM (#21855300)
    ...when chip manufacturers gave away the full specifications. I even received by snail-mail thick books, 500 pages or so, with the specs from companies like Texas instruments and Motorola. Some manufacturers even sent free samples of the chips themselves.


    Where have they gone wrong?

    • Re: (Score:3, Insightful)

      by Travoltus ( 110240 )
      The big picture is that runaway Capitalism is where they have gone wrong.

      Consumer freedom is now irrelevant. What good is the free market when you can't buy what you want - namely, to keep this on topic, where can we get a powerful video card with the full specs for making open source drivers? Good luck building your own fab. Apparently only a trillionaire can afford such simple freedoms.

      This is clearly an example where capitalism fails miserably.

      (Uh oh, here come the angry "Capitalism is God, how dare you

      • Re:I remember a time... (Score:5, Insightful)

        by sqrt(2) ( 786011 ) on Sunday December 30, 2007 @07:05AM (#21855390) Journal
        I don't have mod points but I'll agree with you here. With one caveat. The "free market" (can't exist btw, it's just a idealized concept) IS working how it's supposed to. There isn't really a demand for an open platform GPU and thus the market doesn't provide one. If there were enough people wanting one that a company could make money selling them then you could buy it. Capitalism does tend to screw the little guys who have niche or obscure needs, unless you can pay to get things custom designed and produced.

        But it looks like AMD is finally going to start servicing that section of the market, I'm still skeptical but we'll see how things turn out.

        • Re:I remember a time... (Score:5, Insightful)

          by MMC Monster ( 602931 ) on Sunday December 30, 2007 @08:24AM (#21855688)
          Capitalism is working, only slowly. Open APIs and open source drivers weren't a selling point, so they weren't available. Now with the slowly rising popularity of Linux, and the realization that Linux users are generally more influential in the purchase of hardware than the average buyer, the APIs and drivers will open up.

          It's important that we, as a community, reward the good guys (with more purchases) and to let the sales people know why we choose them over their competitors.

          That being said, I'm a little ambivalent about the whole AMD/ATI video mess. They've been talking this up for the last year, but have the 3d specs for the hardware been released? Is there a stable opensource driver for Linux even close to the performance of the WinXP/Vista drivers (I don't know).

          • Re:I remember a time... (Score:5, Informative)

            by AncientPC ( 951874 ) on Sunday December 30, 2007 @09:39AM (#21856154)
            From the Read More section: AMD Release 900+ pages of GPU Specs [slashdot.org].

            There is a fairly stable closed source ATI driver from the AMD website that supports AIGLX (required for Compiz).

            As for ATI open source drivers refer to this list [ubuntu.com]. Copy and pasted for convenience:

            Unsupported
            X1300 / R515 based cards.
            X1600 / R530 based cards.
            X1800 / R520 based cards.
            X1900 / R580 based cards.

            2D acceleration only
            Xpress 200M Northbridge integrated GPUs

            Good 3D acceleration support
            9500 / R300 based cards.
            9600 / rv350 or rv360 based cards.
            9700 / R300 based cards.
            9800 / R350 or R360 based cards.
            X300 / rv370 based cards.
            X600 / rv380 based cards.
            X700 / rv410 based cards.
            X800 / R420 or R423 or R430 or R480 based cards.
            X850 / R480 or R481 based cards.
            X1050 / rv370 based cards.

            Full 3D acceleration support
            7000 / rv100 based cards.
            7200 / R100 based cards.
            7500 / rv200 based cards.
            8X00 / R200 based cards.
            9000 / rv250 based cards.
            9100 / R200 based cards.
            9200 / rv280 based cards.
            • So... of those cards with full 3D support, will *any* of them fit in an x16 slot? The only one I can find is the 7000, and using it for anything practical seems futile, even if they made it to fit my system board.

              What about support for onboard ATI graphics as well?
              • Pretty much no, they haven't released anything relevant to 3d. The 900 pages was perhaps a useful cross indexed reference to 2d display, but I have not seen anything referencing useful things like textures or geometry. The drivers the GP mentioned were done without aid of recent documentation.

          • Re: (Score:2)

            by Curtman ( 556920 )

            It's important that we, as a community, reward the good guys (with more purchases) and to let the sales people know why we choose them over their competitors.

            That's very hard to do when who the good guy is keeps changing every week. Sometimes it's Via, sometimes it's Intel, sometimes it's AMD. All of them making an effort to move in the open direction, but none of them with a truly spectacular offering.

            ATI was once the darling of the open source community, then fell out of favour, and seems to be c

            • Re: (Score:3, Insightful)

              by SanityInAnarchy ( 655584 )
              Well, recently, it seemed pretty obvious to me -- Intel always has good, open video drivers. But, the performance on Intel video cards sucks compared to nVidia or ATI.

              ATI is more open, but their Linux drivers suck, and have pretty much always sucked, for new hardware. You could get a free driver for old hardware, but not everything would be supported.

              nVidia is completely closed, but their Linux drivers generally work. You pretty much never have to worry, when upgrading your kernel, whether you're going to b

        • But it looks like AMD is finally going to start servicing that section of the market, I'm still skeptical but we'll see how things turn out.

          I too will join your skepticism and it was the very reason I stopped buying ATI some years ago. At first, ATI support seemed good, then it faltered miserably. Myself and others asked ATI for some information to fix this and didn't even get the time of day. Now that their is talk of coming around I will sit back and wait to see if AMD/ATI can walk the talk. But for

        • Re: (Score:3, Informative)

          by jbengt ( 874751 )

          The "free market" (can't exist btw, it's just a idealized concept) IS working how it's supposed to.

          Actually, the classic "free market", as an idealized concept, requires commodities that can be produced by multiple people/companies, with little or no barrier to entry, knowledgeable buyers, and the flexibility to switch products at will as supply, demand, and prices dictate.

          The current situation with graphics cards fails those prerequisites on all counts.

          • Re: (Score:2)

            by dh003i ( 203189 )
            Wrong. All that the free market requires to work is that the State not create artificial barriers to entry, or prevent competition in various ways (e.g., mandate cartelization, regulate, allow patents, etc). All that the free market requires is the lack of initiation of aggression. To the extent that there is aggression, the free market is hindered. This is the correct concept of the free market, as elaborated by the Austrian school of economics.

            What you refer to is not the classical concept of the free mar

            • Wrong. All that the free market requires to work is that the State not create artificial barriers to entry, or prevent competition in various ways (e.g., mandate cartelization, regulate, allow patents, etc). All that the free market requires is the lack of initiation of aggression. To the extent that there is aggression, the free market is hindered. This is the correct concept of the free market, as elaborated by the Austrian school of economics.

              How convenient that lying is not a form of aggression. I su

              • Re: (Score:2)

                by dh003i ( 203189 )
                If the lying constitutes fraud, then it's a tort. If I say I'll give you a box of apples for $10, you give me $10, and then I give you a box of sand, I've defrauded you, and you're entitled to your $10 back (or a box of apples from me). Of course, local expectations and reason dictate what "a box of apples" is (e.g., what condition do they have to be in, how big, etc). That's something for courts to figure out. Buyer and seller beware.

                Furthermore, regarding asymmetric information problems, consumers have an

            • Re: (Score:2)

              by jbengt ( 874751 )
              Without multiple buyers and multiple sellers, you cannot have a market in the traditional sense, so you don't have a "free market" either, however you want to construct your definition of freedom.

              I admit I don't know the details of "classical economics" or "neoclassical economics", but I used the word "classic" as an adjective, not "classical" as a name.

              • Re: (Score:2)

                by dh003i ( 203189 )
                That is the neoclassical definition of a market, or of "perfect competition", but it is a nirvana fallacy (compare the market as it exists to some unattainable, impossible "ideal", then note how short it comes up). In many cases, especially with entirely new products, there is only one seller. And of every product in its particularity, there is only one seller. It is thus an arbitrary definition. The free market need not refer to a state of the complexity of the economic system. There can be a primitive fre

      • Re:I remember a time... (Score:5, Insightful)

        by Marcion ( 876801 ) on Sunday December 30, 2007 @07:16AM (#21855428) Homepage Journal
        I think the problem is not too much capitalism, it is too little. Adam Smith's free markets have been replaced by an international neo-conservative monarchy and nobility.

        Among certain industries, free markets have been replaced by cartels. These cartels then send out waves of lobbyists and campaign contributions to get governments to further weight the system against the consumer.

        If you look at how the airplane developed, the market was hampered by cartels, patents and so on. However, in the two world wars, the war effort was considered more important than entrenched interests within the early aviation industry. All these cartels and patents were swept aside in favour of truly free markers, and they could finally build decent planes, and build them in quantity.

        Society is slowly but surely going to realise that computers are more important for the development of the economy and society as a whole than for the narrow interests of the technology industry, and then radically free markets will be introduced once again.

        Look at the Microsoft vs EU decision and the OLPC project, both of these in their different ways are interesting early signs.

        • Re: (Score:2)

          by Marcion ( 876801 )
          BTW, the last post is a reply to the post above it, I have no idea how it relates to graphics cards.

        • Re: (Score:3, Insightful)

          by eggnoglatte ( 1047660 )
          Ah, but in the world wars, the bill was footed by the taxpayers. Only the influx from government money was able to push forward the necessary research under these conditions.

          If you change the rules to treat all products as commodities, then only commodities will get built. If you abandon patents, companies will try to protect their knowledge keeping more trade secrets. Products will become less open. If you legislate that possiblity away(*), then research and advanced development will simply stall, becau

          • Re: (Score:3, Insightful)

            by Marcion ( 876801 )
            Ah, but in the world wars, the bill was footed by the taxpayers. Only the influx from government money was able to push forward the necessary research under these conditions.

            I think it is important to separate the supply side from the demand side. Ditching all the patents and agreements and so on was a supply-side change. The government needing lots of planes for the war is on the demand side.

            So yes the money is part of it, but I think the longer lasting effect was on liberating the supply, it did need the
            • Well, I am at a university, and I have been through the whole startup thing (we sold out earlier this year). Let me assure you that of course you patent when you do commercially relevant research at a university. It is a bare necessity for raising investment capital. And you do need capital to take university research and turn it into a product. That process takes years and is not cheap.

              Patents and trade secrets are key pillars of any technology startup. Take those away and there is NOTHING that prevents th

              • Re: (Score:2)

                by Marcion ( 876801 )
                > It is a bare necessity for raising investment capital.

                Again, that is just a symptom of the current system. Which can only measure innovation by counting patents.
                • Actually, it is deeper than that. Before investors give you money, they expect a certain level of security. Startups are risky investments under the best conditions, of course, but at least you want to make sure that if everything goes as well as it could, you are not going to fall prey to a rip-off at the end.

                  I'll give you a non-startup example of how things can go wrong if you don't patent. A colleague of mine at the university invented a really cool technology that would blend in nicely with existing fir

          • Re: (Score:2)

            by ajs318 ( 655362 )

            (*) As an aside, it is beyond me how some (note I say some) supposed libertarians can advocate mandatory opening of product specs. Apart from the irony of a "libertarian" requesting more regulation, this would reduce freedom, not improve it. One of the key aspects of freedom of speach is the freedom to shut the fuck up and not tell you what I don't want to.

            I think you are confusing freedom with power [gnu.org]. The freedom to make use of your own hardware is more important than the power to withhold information from

            • The freedom to make use of your own hardware is more important than the power to withhold information from the people who paid your wages, and that freedom is what government should be protecting.

              Absolutely, definitely, positively NOT.

              Freedom of speech, including the right to say nothing is a basic human right. The right to use some crappy piece of hardware is a privilege you have won through contractual negotiations, and is not even close in importance to freedom of speech. One of the other pillars of free societies is the right to negotiate any contract (so long as it doe not violate basic human rights (*)), and have the contract enforced by the court system. Well, the vendor of your hardware

              • Re: (Score:2)

                by ajs318 ( 655362 )

                And yes, I do realize that in practice it would be very hard and expensive to strike a contract that provides you with specs. So what? The right to negotiate a contract does not guarantee that you will find a partner who accepts your conditions.

                So, it is the legitimate purpose of Government to ensure that the people get all the rights they deserve, all the time. Including the right to be privy to any secret embodied in any article they rightfully own. In other words: If you won't tell me of your own fre

                • You did not pay for the specs, so you don't have the "right" to get them. Just because you want something does not make it a right.

                  There is no constitution on this planet that defines the right to obtain product specs as a basic human right. Since it isn't a basic human right, it is subject to contractual negotiations. You loose. Get over it.

                  • Re: (Score:2)

                    by ajs318 ( 655362 )
                    It's my property, so nothing about it is a secret from me. That is a fundamental right -- it's part of the definition of ownership. Any constitution which permits the ownership of property should grant this right, if governments were doing the job we paid them to do properly.
                    • That is simply not the case. Just because you buy a Coke does NOT entitle you to the recipy. For food and drugs you are at least entitled to basic nutritional information since it can be important for mainatining your health, but even there the specifics are trade secrets. For anything else, you can try to reverse engineer, but the manufacturer has no obligation whatsoever to provide you with additional info.

                      Trade secrets have centuries if not meillenia of tradition. Any attempt to get rid of them like you

        • Re: (Score:3, Interesting)

          by symbolic ( 11752 )
          I think the problem is not too much capitalism, it is too little. Adam Smith's free markets have been replaced by an international neo-conservative monarchy and nobility.

          I'm not sure I agree. I think there's plenty of capitalism, but I also think that capitalism is a lot like freedom - if you take it forgranted, you'll eventually get screwed. Participation in an capitalistic society is an active process, not a passive one. The passivity is born from laziness. Change only really seems to happen when somethin

          • Re: (Score:2)

            by Marcion ( 876801 )
            I'm not sure I agree.

            I think we probably do agree. Taking capitalism for granted is what allows a neo-nobility to control us. I agree with your point entirely.

            As an example, it is conceivable that the citizens of the US could have put the RIAA out of business a long time ago. But people still insist on giving them money (for the mediocre crap they produce, no less), which the RIAA then uses to continue to tightening the noose around copyright law and fair use.

            I think the RIAA is a bit like AIDS. If you hav

      • Re: (Score:2, Troll)

        by Bloater ( 12932 )
        We don't have free market capitalism in the western world. Most markets are tightly controlled and made mostly worthless - hence all our technological development hasn't made most of us rich. What happens to us these days doesn't happen in free markets thus the markets are not free.

        • Re: (Score:2, Interesting)

          by rdoger6424 ( 879843 )
          because free markets can't exist in real life.

          • Re: (Score:2)

            by h4rm0ny ( 722443 )

            Why not?

          • Re:I remember a time... (Score:4, Insightful)

            by Marcion ( 876801 ) on Sunday December 30, 2007 @07:58AM (#21855588) Homepage Journal
            Markets start free, they just become non-free through layers of government intervention and large firms colluding. Certain agricultural products are more or less free markets, as well as light bulbs, screws, etc.

            The resistances in the system are not physical properties of nature but man made structures. The problems are never that the speed of light is too slow or gravity is too strong. The problem is that those who think they are against government intervention, often are the first to argue for patents, trademarks, trade barriers, special protections, and so on.

            Free markets are the optimal solution for the majority of the population, both as consumers and employees. Cartels only benefit the minority.

            If governments became truly accountable to voters, such that the voters could clearly get actual representatives, rather than a choice of two identical people who will ignore the voters for the next four-five years, then I don't see why the interference cannot be removed in most industries if the will was there.
            • ``If governments became truly accountable to voters, ...'' ...and voters actually had a clue, ...

              I mean, seriously. You can't know everything about everything. And when it comes to economics, I'm willing to bet most voters believe things that are the absolute opposite of what experts believe. I think many of the barriers we have in our markets are there exactly because voters want them to be.

              • Re: (Score:2)

                by Marcion ( 876801 )
                And when it comes to economics, I'm willing to bet most voters believe things that are the absolute opposite of what experts believe. I think many of the barriers we have in our markets are there exactly because voters want them to be.

                I agree. But what are these symptoms of? They are symptoms of people believing what the advertising, the lobbyists, and the spin-doctors want them to believe. Yes the majority of people are programmable by the elite.

                If power is unequally distributed, then the minority can make
              • Democracy and free-market capitalism have exactly the same flaw. Free-market capitalism relies on informed customers, democracy relies on informed voters. Since neither of these prerequisites are ever found in the real world, we typically create compromises, which have some aspects of these `ideal' systems and some of others to make up for the weaknesses in them.

      • Not capitalism (Score:3, Insightful)

        by atlep ( 36041 )
        >The big picture is that runaway Capitalism is where they have gone wrong.

        Capitalism is about free competition. CRM, the cited problem here, is about regulation. Regulation is not the same as free competition.

        Some people have a tendency to think that when exploitation and capitalizing on other people is going on, then automatically capitalism is to blame. It is of course not that simple. Exploitation and capitalizing on others happen under capitalism, but also under a lot of other systems.

        This time it is

      • Not capitalism (Score:3, Insightful)

        by atlep ( 36041 )
        >The big picture is that runaway Capitalism is where they have gone wrong.

        Capitalism is about free competition. DRM, the cited problem here, is about regulation. Regulation is not the same as free competition.

        Some people have a tendency to think that when exploitation and capitalizing on other people is going on, then automatically capitalism is to blame. It is of course not that simple. Exploitation and capitalizing on others happen under capitalism, but also under a lot of other systems.

        This time it is
      • You have freedom, you don't have to buy products if you don't want to. GPUs are a luxury, not a necessity. Stop being a spoiled brat. If someone wants to manufacture something, sell it to people, but not give people the detailed specifications for it, that is their right. I wonder why it is that producer freedom is never mentioned. If I make something, and sell it to you, I have absolutely 0 obligation to tell you how I made it.

        Btw, if capitalism really couldn't service those "niche" markets, then AMD would
      • Kudos for you.
    • You still get full specs [microchip.com] from many companies. Free samples [microchip.com] are still around. Though it sadly doesn't seem to be popular with PC components.

    • ...when chip manufacturers gave away the full specifications. I even received by snail-mail thick books, 500 pages or so, with the specs from companies like Texas instruments and Motorola. Some manufacturers even sent free samples of the chips themselves.

      That still happens. All chip manufacturers still give away full specs for chips that are traded at the open market.

      The thing is: your mind is also fooling you. While simple stuff like small CPUs and other cheap integrated circuits always had documentat

    • ...when chip manufacturers gave away the full specifications. I even received by snail-mail thick books, 500 pages or so, with the specs from companies like Texas instruments and Motorola. Some manufacturers even sent free samples of the chips themselves.

      Where have they gone wrong?

      Monoculture, and convenient drivers.

      Back in the day, most anyone who developed a software program that used a given hardware device had to create their own driver for that device, and needed the hardware specifications to do it.

  • Hopefully they will also be more commercial-source friendly as well. I've "resorted" to buying XiG [xig.com]'s product in the past, because Xorg wasn't working quite right for me... but I (and many others with driver problems [phoronix.com]) simply don't have that option unless AMD passes the info to commercial vendors such as these as well. Yes, I know FOSS is all that, but when you need things to Just Work, sometimes it's easier to pay the money.

  • I just upgraded my system. I'm not a big graphics user but I bought an ATI HD 2600 Pro over an Nvidia card because AMD seem to be really supportive of Open Source at the moment. Driver support will catch up with me soon enough and I expect ATI cards to end up the best supported cards under Linux, until Nvidia starts following.

    • Re: (Score:2, Insightful)

      by Eddi3 ( 1046882 )
      I hate to be the bearer of bad news, but Nvidia cards are supported just fine under Linux. Just because they aren't supported how you like, doesn't mean they aren't supported. Not that I wouldn't like it if they opened up their specifications, though.

      • Re: (Score:2)

        by h4rm0ny ( 722443 )

        Oh, I know that, but thank you. :) I'm not dismayed. I'm actually using an Nvidia card at the moment. I just felt that with my new computer (updating the OS now), that I should show my appreciation of ATI opening up to the open source world. My new card works okay with the proprietary drivers (so far). I've been following the forums on Phoronix and I think it wont be that long before I see an equivalent open source driver. And from there on, it's all improvement.

      • Re: (Score:2, Informative)

        by chromatic ( 9471 )

        Nvidia cards are supported just fine under Linux.

        Perhaps that subset of Linux known as x86 has fine support, but the last time I looked, PPC Linux (for example) had no support from NVidia. Synecdoche does not make for accurate engineering.

    • $500 say you won't see quality open source drivers for your card within the next 3 years. By quality I mean something that supports all features, is robust, and within 10% or so of the performance of commercial drivers.

      No, I am not trolling. This is a serious offer to you h4rm0ny. If you agree, just reply to this posting within the next 2 days.

    • I just upgraded my system. I'm not a big graphics user but I bought an ATI HD 2600 Pro over an Nvidia card because AMD seem to be really supportive of Open Source at the moment.

      So far, this is lip service. ATI first has to release the information or reference driver source and then give it 6 months for testing before you see see it in the sources.

      If you are buying a GPU, I would first look at the hardware list for your favorite Linux distro, and purchase based on that as support is proven, the vendors h

    • Re: (Score:2)

      by Aardpig ( 622459 )

      ...and in upgrading my laptop (from an abysmal Asus to an outstanding Lenovo), I took the decision to ditch ATI and go with NVidia. ATI drivers are slow, buggy, and change in unpredictable ways from one release to the next. The final straw was a driver upgrade that made the GPU run at 70 degrees *centigrade* when idling.

      Whereas, with my new NVidia-based system, the drivers have been as solid as a rock.

      I wish you good luck with ATI, but I fear you're in for a whole world of pain...

  • Don't buy AMD. (Score:3, Insightful)

    by ni1s ( 1065810 ) on Sunday December 30, 2007 @06:58AM (#21855364)
    DRM "functionality" in hardware? No thanks.

    • Re: (Score:2)

      by baadger ( 764884 )
      You might want to rethink that. [wikipedia.org].

    • Correction: Don't buy Vista. (Score:5, Interesting)

      by n dot l ( 1099033 ) on Sunday December 30, 2007 @07:42AM (#21855518)

      DRM "functionality" in hardware? No thanks.
      You know, I remember an NVIDIA engineer complaining to me about how they'd had to do a bunch of really fucked up stuff to get the G80 GPUs to support HD playback on Vista. I'm pretty sure Intel's latest stuff has to deal with the same bullshit too. So really, the title of your post should read "Don't buy post-Vista GPUs". That kinda puts a damper on the whole 3D graphics thing, doesn't it?

      Better advice would be, "Don't run your new GPU on an OS that forces it to enable the stupid DRM logic that the engineers really didn't want to build into it in the first place." Yeah, that's much better.

      • Re: (Score:2, Insightful)

        by mooothecow ( 1210044 )
        Hardwired HDCP compliance which is required to get a high definition stream that is standard? Without it, it's illegal to play Blu-ray and HD-DVD? Let's place blame where it's due.
  • Other than the specs released back in September [x.org], have AMD even released full specifications to their full range of GPU's yet?

    I was speaking to a X driver developer on IRC a few weeks back, and in the course of discussion, he claimed AMD hadn't yet released specs for the 3D engine of *any* of their GPU's yet. Is this true?

    • Re: (Score:2, Informative)

      by Anonymous Coward
      Yes, it's true. All the documents have to filter through their legal department before publication, and that takes time.
  • I don't believe AMD/ATI. their video cards have always had really bad unix support.

    nvidia is no better, really. their 8-series still has zero accel linux video (HD) support. even their XP (!) drivers don't fully support their own chips and this is almost a year after the 8's came out.

    when I can use full mpeg and avc playback in linux, I'll believe AMD/ATI. but until that day, its all lies and false promises.

    • I don't think you're being fair on AMD. It wasn't too long ago that they announced that they would be releasing documentation for some of their GPUs, and people (understandably) didn't believe them. But then they actually did it a couple of weeks later! [phoronix.com]

      As they came good on their last promise, I'm willing to believe them on this one.

      In the end it doesn't really matter whether we believe them or not, though. What matters is whether they end up doing it, and no one should rush to buy their GPUs until they have

    • I don't believe AMD/ATI. their video cards have always had really bad unix support.

      Quite true. Many years ago, almost last century ATI had a series of well supported cards that worked well with Linux, but also BSD and Solaris. Quite nice cards too in their time. But then ATI changed hard and fast to being closed and getting drivers became near impossible. A lot of times the VGA basics would work, but you were under utilizing the card. It is about where I stopped buying ATI and moved to nVidia because

  • nVidia got
    a) open source 2D driver
    b) closed source 3D driver

    The hardware is good, the driver is good and the installation of proprietary drivers is extremely easy in my favorite distro. If you want to get back on my computer (I bought one ATI card, flaky POS) then I want open source 3D specs. No secret registries or features that makes it half-assed compared to closed source. That is, *after* you've released the specs for the most recent cards and OSS developers have had some time to work on it. There's no

  • The End Times are Near (Score:5, Funny)

    by sykopomp ( 1133507 ) on Sunday December 30, 2007 @08:06AM (#21855622)
    ..it's the only conclusion I can reach. ATi drivers are going to be Linux-friendly, the courts and colleges are actively pushing back against the RIAA and MPAA, both of which are starting to change their business model, and Duke Nukem Forever is actually going to be done at some point.

    Hold me, I'm scared.
  • This sounds like kids complaining about dessert when they haven't even started the main course. While it would be nice to have hardware-accelerated video playback, video plays just fine on most Linux/BSD systems today. Linux/BSD needs accelerated 3D graphics much more. It's not easy to write a really good OpenGL driver for an advanced video card like that (actually, it's several video cards, which makes it even harder to write and test). Maybe AMD will change their minds in the couple of years it takes the

    • Re: (Score:3, Interesting)

      by Fry-kun ( 619632 )
      I see you haven't tried playing HD content without hardware acceleration.
      I've got some video clips that can't be played on a reasonable-spec laptop (1.8G Core Duo, 2G RAM) unless I'm using the proprietary ATI driver - and even then, the only way to get nice-looking picture is to render to opengl interface.
      • You got me there. I don't have an HD-DVD or BluRay drive on any of my systems, and I don't have enough free disk space to be downloading HD content. Although it might be better to write shaders to help with the decoding and just use OpenGL than to use something specialized (because then the same will work on nVidia, and hopefully on any DX10-level integrated chipsets that come out.

        In one of the weirdest coincidences I've been involved in recently, I happen to be in the middle of writing GLSL shaders to help

  • Specifically... (Score:3, Insightful)

    by Junta ( 36770 ) on Sunday December 30, 2007 @10:10AM (#21856410)
    They are talking about ensuring that in 1 to 2 years time, video acceleration will be architected to be open-source accessible without compromising DRM (if they continue to implement that).

    Contrast this to nVidia which for the GeForce 8 series they've decided to stop supporting XvMC in their closed-source driver.

    From my perspective, currently if you are buying a new video card, your system is probably already able to keep up with 1920x1080 playback using only the CPU. In a year to two years time, I can't imagine the generation of systems not being able to cope. XvMC only helps for MPEG-2, wasn't updated to be usable for more advanced codecs. I've seen at least discussion toward changing that, but I think the community is in largely a 'what's the point?' sort of mentality.

    As much as I'm all for this strategy, if it costs them a significant amount in terms of production cost someway, it may not be worth the benefit, which is relegated mostly to a token gesture now. The 3D acceleration and, by association, the proccesing capabilities of the GPU are far more interesting. It sounds like they face no insurmountable obstacles in releasing those specs (though they have taken their sweet time about it since their announcement a few months ago).
    • In about two years, they will have chips on the GPUs capable of decoding H.264 in hardware, and XvMC will hopefully support MPEG-4 hardware decoding. That's what this is about.
      • They already do h264 in hardware I believe, but without a framework in Linux to take advantage.it currently. Hence my discussion about a replacement being discussed for XvMC (extending XvMC I believe they decided was not feasible, as it wasn't flexibly designed). And I believe that in 2 years time, hardware accelerated 1920x1080 h264 playback will be moot in the face of the processors that wouldn't be the least bit troubled to play that back.
  • Until I see results from ATI in the form of documents that tell you how to render 3D on ATI GPUs, I wont hold my breath.

    Let me know when I can play Quake III (or use Second Life or any other open source OpenGL app) on a GPU still being manufactured by ATI using 100% open source drivers (i.e. no binary drivers at all) then I will care.

Slashdot Top Deals

news: gotcha

Close