Slashdot Mirror


AMD Goes Open Source, Announces GPUOpen Initiative, Linux Compiler, Drivers (hothardware.com)

MojoKid writes: AMD announced today that the company is releasing a slew of open-source software and tools to give game developers, heterogeneous applications, and HPC applications deeper access to the GPU and GPU resources. AMD and their Radeon Technologies Group (RTG) are looking for ways to ease game development, so developers can more easily re-use code and port their games from consoles over to the PC. With GPUOpen, game developers will have direct access to GPU hardware, as well as access to a large collection of open source effects, tools, libraries and SDKs, which are being made available on GitHub under an MIT open-source license. As part of the effort, the company is also releasing a new HCC C++ compiler which will be a tool in enabling developers to more easily leverage the resources of discrete GPU hardware in heterogeneous systems. The HCC complier also allows developers to convert CUDA code to portable C++. According to AMD, internal testing shows that in many cases 90 percent or more of CUDA code can be automatically converted into C++ with the final 10 percent converted manually in the widely popular C++ language. An early access program for the "Boltzmann Initiative" tools is planned for Q1 2016.

89 comments

  1. So... by TheDarkMaster · · Score: 3, Interesting

    Decent Windows/Linux drivers now?

    --
    Religion: The greatest weapon of mass destruction of all time
    1. Re:So... by Anonymous Coward · · Score: 0

      Slightly better Windows drivers but don't get your hopes up about Linux.

    2. Re:So... by Anonymous Coward · · Score: 0

      The AMD drivers have always be decent to me. It only accelerated games, never pr0n.

    3. Re:So... by Anonymous Coward · · Score: 0

      If they can steal the Steam Machine market from Nvidia they will finally be a step ahead of them in something.

    4. Re:So... by Anonymous Coward · · Score: 0

      The AMD drivers have always be decent to me. It only accelerated games, never pr0n.

      you haven't lived until you've experienced your pr0n collection in gliv and impress!ve

    5. Re:So... by malditaenvidia · · Score: 1

      They have a stranglehold on the console market. Steam machines are largely irrelevant in comparison.

    6. Re:So... by Rinikusu · · Score: 3, Insightful

      Isn't this the AMD version of "Year of Linux on the Desktop"?

      --
      If you were me, you'd be good lookin'. - six string samurai
    7. Re:So... by florin · · Score: 1

      They're industry leaders in churning out Powerpoint decks full of bold announcements about new initiatives to tide us over till the next change of direction.

    8. Re:So... by Anonymous Coward · · Score: 0

      LOL! Good one.

    9. Re:So... by hairyfeet · · Score: 4, Insightful

      Been AMD/ATI exclusive since the X1650 Pro and I only had a single driver that was flaky and that was 14.4 and even on it gaming was fine, it merely had issues providing video acceleration through third party programs like VLC..

      So perhaps you could give us a list of what issues you've been having and on what AMD/ATI hardware? Because I know I see a hell of a lot more crap drivers dealing with nvidia hardware at the shop, frustrating enough to me that for the only time in history me and Linus Torvalds agree on something so I look forward to hearing about your issues, perhaps you could provide a couple screencaps?

      BTW I thought /. was all gung ho FOSS, so what happened? Is it not the FOSS community that has been saying for years "just give us the specs and open the code and the kernel devs will handle it"? Is that not EXACTLY what AMD did, in fact hiring guys to work on the FOSS driver to get it up to speed quicker? I have to find it frankly baffling how many here are supposedly all for the four freedoms yet when the rubber meets the road will go with Nvidia, a company that constantly tries to force proprietary solutions like CUDA over FOSS solutions like OpenCL, a company that has made it clear they are not gonna open squat over a company that has been providing docs and specs and opening their code since 2012...wth people? Is this bizzaro world, where support for FOSS equals supporting the proprietary choice?

      --
      ACs don't waste your time replying, your posts are never seen by me.
    10. Re:So... by Anonymous Coward · · Score: 0

      I have to find it frankly baffling how many here are supposedly all for the four freedoms yet when the rubber meets the road will go with Nvidia, a company that constantly tries to force proprietary solutions like CUDA over FOSS solutions like OpenCL, a company that has made it clear they are not gonna open squat over a company that has been providing docs and specs and opening their code since 2012...wth people?

      1. NVIDIA drivers work, most of the time.
      2. see #1
      3. see #1

      Since 99.999% of the users don't know their ass from the driver code, including 99% of developers, then GPL driver is only as good as the maintainer. What GPL brings is that these drivers can be off-loaded to community for maintenance. But it doesn't mean community will magically write these complex drivers for you.

      As to CUDA vs. OpenCL, this has NOTHING to do with FOSS. OpenCL is just a standard API, while CUDA is nVidia software. OpenCL specs derive from lots of CUDA, because CUDA was there first. But nothing to do with FOSS.

    11. Re:So... by Anonymous Coward · · Score: 0

      Too many idiotic Homer Simpson's of the world that would sell their soul for a donut.

    12. Re:So... by TheDarkMaster · · Score: 1

      Well, I had a 4870X2 in the main computer and today I have a HD6000 series on the secondary computer. The fact is, to start the driver user interface is really shitty and amateur. 160MB (or more) to make a user interface for a driver, really? Second, I suppose I don't need to mention the thousands of cases of problems caused by the driver (the hardware itself is good), I will not do all the necessary research for you just ignore it.

      But what bothers me more in this case is that AMD known for years that their drivers are shitty, but nevertheless rather than fix the confusion they only makes even worse (see the new "crimson" driver, that seems to have been written by a designer who knows nothing about usability or drivers, even less about K.I.S.S.)

      --
      Religion: The greatest weapon of mass destruction of all time
    13. Re:So... by hairyfeet · · Score: 1

      You are bitching about 160Mb...in 2015? Really? WTF kinda low rent mickey mouse hardware you running that you can even fricking NOTICE 160Mb of RAM? For fucks sake simply reading this used more than 160Mb of RAM! And why in the hell are you even launching the control center enough to know this? BTW have you bothered to see how much the Nvidia Control Panel uses?

      If the most you can find to bitch about is how the control panel uses less MB than a smartphone UI from 2007? I think you are making my case more than you are your own. And sorry but "just take my word for it" isn't good enough, lets see them screencaps from YOUR system. Otherwise I'm sure I can grab random screencaps from Nvidia errors all over the web, that don't prove shit.

      --
      ACs don't waste your time replying, your posts are never seen by me.
    14. Re:So... by TheDarkMaster · · Score: 1

      You are bitching about 160Mb...in 2015? Really?

      Why I should spend 160MB to make my application work if I can make it work just fine with only 10MB? Just because I now have 16GB of RAM to use I should be wasting her with sloppy applications? It seems to me you live in a very wasteful society.

      --
      Religion: The greatest weapon of mass destruction of all time
    15. Re:So... by Anonymous Coward · · Score: 0

      Empty RAM is wasted RAM.
      Sloppy implementation of an application saved developer time, which was used for something more valuable. Not ideal, yes. But you have to prioritize.

    16. Re:So... by Anonymous Coward · · Score: 0

      Same here. Went ATI/AMD years ago, starting with x600 IIRC. Used about 6 different cards in total. Mostly on Windows. Never knew what's all the fuss is about. Everything worked, even AGP x600 and HD2600 cards on the shoddy XP x64.
      The control center UI sucks: it's slow, awkward and more convoluted than it needs to be. But you rarely if ever need to use it, so whatever.

      Probably people mostly keep repeating some outdated horror stories from the 90's or early 00's.

  2. Thank you for the source code by Anonymous Coward · · Score: 1

    Certainly makes AMD to look a bit more ethical in my eyes now.

  3. Coming next... by Anonymous Coward · · Score: 1

    Coming next is Windows Server for GPU an will be licensed by GPU core.

  4. Of course by Anonymous Coward · · Score: 0

    this had to happen just riiiiiiight after they discontinued their pre-GCN video cards.

  5. is this the last gasp for amd? by Anonymous Coward · · Score: 1

    i.e. similar to how Netscape went open source right before "Netscape" the commercial company went away (bought by aol... then becomes mozilla which outputs current firefox).

    1. Re:is this the last gasp for amd? by beschra · · Score: 1

      Maybe if AMD only made software, but I believe they're still considered a hardware company. CPUs and GPUs and stuff like that.

      --
      It is unwise to ascribe motive
    2. Re:is this the last gasp for amd? by mrchaotica · · Score: 1

      AMD owns no fabs; they don't make hardware. They only design it.

      --

      "[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz

    3. Re:is this the last gasp for amd? by Anonymous Coward · · Score: 1

      Apple owns no fabs; they don't make hardware. They only design it.

    4. Re:is this the last gasp for amd? by Anonymous Coward · · Score: 0

      They do make the chips IIRC, just not the rest of the video card.

    5. Re:is this the last gasp for amd? by Anonymous Coward · · Score: 0

      AMD owns no fabs; they don't make hardware. They only design it.

      No shit, Sherlock. A lot of hardware companies (especially chipmakers) do not have their own fabs.

    6. Re:is this the last gasp for amd? by Anonymous Coward · · Score: 0

      Does anyone consider Apple a "hardware" company?

    7. Re:is this the last gasp for amd? by florin · · Score: 1
    8. Re:is this the last gasp for amd? by Anonymous Coward · · Score: 0

      As long as they design their own stuff, yes.

      Problem is, "hardware company" can mean several different things.

    9. Re:is this the last gasp for amd? by Anonymous Coward · · Score: 0

      Apple does both, and state so on their packages.

    10. Re:is this the last gasp for amd? by Billly+Gates · · Score: 2

      Maybe if AMD only made software, but I believe they're still considered a hardware company. CPUs and GPUs and stuff like that.

      Well go google Nvida Gameworks? All the crappy buggy ports of games that are optimized for Nvidia hardware are making AMD GPUs look bad and use secrets and hacks to prevent debugging.

      This is an alternative to Nvidia gameworks so developers stop using hairworks, physics works, and other things that make AMD appear slower

  6. Yeah but... by Anonymous Coward · · Score: 0

    What about OpenBSD?

    1. Re:Yeah but... by nvm_my_comment · · Score: 1

      then having a MIT license might give them a edge of nvidia. Or not.

    2. Re:Yeah but... by Anonymous Coward · · Score: 0

      Using an MIT licence with a Berkeley OS? What is the world coming to?

  7. AMD business model by Anonymous Coward · · Score: 0

    1) Dump failed, low-quality projects to open source developers

    2) Hope someone else will do you job for free

    3) ???

    4) PROFIT!!!

    1. Re:AMD business model by Anonymous Coward · · Score: 0

      Too bad no one writes GPU drivers for free. They are millions of lines of code and require massive amounts of experience from the field when it comes to the programmer.

  8. Again? by Anonymous Coward · · Score: 0

    Don't AMD announce some new open source initiative every year?

    Nothing ever happens.....

    1. Re:Again? by Tough+Love · · Score: 2

      Don't AMD announce some new open source initiative every year?

      Nothing ever happens.....

      Then please inform me, wise coward, what this "radeon" driver is, that has satisfied my needs for years?

      --
      When all you have is a hammer, every problem starts to look like a thumb.
    2. Re:Again? by requerdanos · · Score: 1

      >> Then please inform me, wise coward, what this "radeon" driver is

      A free driver wrapper around a nonfree firmware blob. To use it, you first have to install firmware-linux-nonfree. See https://wiki.debian.org/AtiHow...

    3. Re:Again? by Anonymous Coward · · Score: 0

      Wow, that sounds horrible. Good thing it's not actually true. The "firmware blob" is (are actually, there are several) HW microcode images which we happen to load into on-chip microcode RAM rather than burning them into the chip like other vendors.

    4. Re:Again? by Anonymous Coward · · Score: 0

      That HW microcode is still needed for the radeon driver to work at all. There are some fixes done from the linux-libre kernel but you will not get any 3D acceleration without the blobs. Thus the radeon driver will not work.
      If going all "open source" is the goal, then releasing the code for the blobs would be the first thing to to. Not even older cards will work without it.

    5. Re:Again? by Anonymous Coward · · Score: 0

      No, that HW microcode is needed for the HARDWARE to work at all.

      Agree that if going open source HARDWARE was the goal (which it is not) then releasing the code for the HW microcode would be an easy first step.

      If you go old enough (Mach64) the microcode is not needed. We started soft-loading microcode with the Rage 128; prior to that the microcode was built into the chip.

    6. Re:Again? by requerdanos · · Score: 1

      The firmware blobs are released under a nonfree license. I am not saying this is bad or good, just observing that it is. Though I guess it does sound horrible now that you mention it--and I wish there were a reasonable solution.

    7. Re:Again? by Anonymous Coward · · Score: 0

      There is a reasonable solution - recognize that the microcode images in this case are part of the hardware, not part of the driver, and stop rejecting what are otherwise completely free drivers over what can at best be called a technicality.

      Can you really say that providing source code for the parts of the GDDR5 memory controller HW which are implemented in microcode (while not caring a whit about the other 95% of the hardware source code) somehow makes the drivers more free ? Can you really say that GDDR5 link training would otherwise have been done in the driver ? If you can't make a solid argument for both of those (and for the other microcode images) then this is really all about an over-enthusiastic interpretation of a rule that was correct for a single specific case (moving the entire network driver onto on-chip memory with a separate general purpose processor) but which leads to bad decisions when applied too generally.

    8. Re:Again? by Anonymous Coward · · Score: 0

      All hardware need software for it to run. Weather its firmware, drivers or just simple microcodes loaded into the hardware. Without the software, the hardware is useless. its just blocks of nice looking material.
      I see what you are getting at, but you also seem to not see the part where you need to install non-free blobs (As in software...) on your computer is not "open source". Also the license for the code stops anyone from reverse engineering it.
      It would also not mean open source hardware. That would be if you released cad-files for the PCB and let anyone make, develop and/or sell it. That microcode is for the hardware just as needed as the drivers.
      Would you say that open source bios and bootloaders would also mean open source hardware for motherboards? That makes no sense to me.

    9. Re:Again? by Anonymous Coward · · Score: 0

      I see the part where you need to install non-free blobs, but the point I'm making is that if those same non-free blobs were built into the chip (as most of our competitors do) then there would be no issue. So it's not the existence of the non-free blobs that's the problem, it's just the fact that the blobs are distributed through the package manager rather than being built into hardware or burned into the VBIOS ROM.

      >>Would you say that open source bios and bootloaders would also mean open source hardware for motherboards? That makes no sense to me.

      Makes no sense to me either. That's why you never hear me say it. Distinction is that the BIOS and bootloaders execute on the CPU, so I understand why they get lumped in with the OS / driver / application software. What we are talking about here is part of the HW design which we happen to execute from RAM rather than ROM in order to give us more flexibility during the R&D cycle, which leads us to the totally broken conclusion that :

      - if we distribute the images as part of the GPU then the microcode is hardware and considered OK
      - if we distribute the images as part of the VBIOS ROM then the microcode is firmware and considered OK
      - if we distribute the images via package manager then the microcode is software and considered evil

      What's wrong with this picture ? Same driver, same microcode, totally different outcome.

    10. Re:Again? by Anonymous Coward · · Score: 0

      I am a bit tired so my answer might not be all well thought through. Forgive me for that.

      I would not consider any type of code to be hardware. It might be a part of the hardware thou, as in your example with the microcode either as part of the GPU or (maybe, not to sure about this one) the VBIOS ROM.

      I would perhaps consider the border of when its OK and not OK to be if you can change the code or not. That is also the reason Im not sure about the VBIOS part.

      I don't want you to the get wrong picture. Its great that the microcode can get updated. But its more of an ethical question. By forcing me to add the non-free repositories I have to take a stance I am not willing to do. I either run all graphics with no acceleration (as I do now) or I contradict my own principles and add the repo, making the OS a non-free OS. Freedom protecting distros like Trisquel or gNewSense simply wont have it in the repos because they dont have any non-free repos.
      The outcome is not the same for the user even if the code itself is the same. The user have to lower its own freedom because of the distribution through package managers.

      Hope you understand what Im trying to say. Im sure I sound like a negative and pessimistic jerk but Im really not.

    11. Re:Again? by bridgmanAMD · · Score: 1

      >>Hope you understand what Im trying to say. Im sure I sound like a negative and pessimistic jerk but Im really not. Nope, I understand completely and you said it very well (better than I was doing). The problem is arguably as simple as "guilt by association"... ie in order to let the drivers load microcode that is distributed in hardware or flash on competing products you need to use an OS mechanism which also allows the use of other, obviously more problematic "firmware" blobs which clearly do perform functions that would otherwise be implemented in the driver. I don't know if there is a solution in there but if nothing else it might help to explain to board vendors why people care about not having to ship the microcode images as part of the "software". Thanks.

    12. Re:Again? by bridgmanAMD · · Score: 1

      I would perhaps consider the border of when its OK and not OK to be if you can change the code or not. That is also the reason Im not sure about the VBIOS part.

      I don't want you to the get wrong picture. Its great that the microcode can get updated. But its more of an ethical question.

      Yep, the ability to update the microcode is often used as a criteria for whether something is "software" or not, but following that line of thought quickly leads to what I think everyone agrees are counter-productive conclusions. For example:

      1. Burning microcode into the chip is OK, and we promote use of those chips when accompanied with free drivers even if the microcode is not free

      2. Making the microcode upgradable (which helps us) is even better, but we explicitly prohibit use of those chips unless the microcode is also free, even if the drivers are completely free except for loading the microcode

      3. Even though microcode changes are normally made only during pre-launch development or in order to SUPPORT the development of free drivers we penalize users of that hardware because the microcode COULD be changed

      4. Even if we put controls in the driver code to prevent changes to the microcode (eg validating a hash) so that the microcode can never change we still penalize users of that hardware, even though we think changeable microcode is otherwise better

      5. If the microcode were distributed in the VBIOS ROM the libre OS builds would allow full, accelerated use of that hardware because the drivers could then be considered totally free, despite the microcode still being changeable and the drivers being otherwise identical, but we don't want to make exceptions for that exact same microcode/driver combination in the distros themselves

      (in fairness the decision to lump all microcode and firmware into a single repo was made upstream in Debian not downstream in the libre OSes)

  9. AMD desperate by Anonymous Coward · · Score: 1

    AMD seems desperate to garner any headlines it can these days. How much open source gaming is really around, and how many will think this is going to sway any gaming developer over to AMD solely? I get that many games for PC's are designed for certain hardware, but that's always been the case and it has never really amounted to any significant dedication from users for certain hardware. I keep reading how great Linux games are, but its nothing compared to the rest of the retail gaming ecosystem and never will be. Good luck AMD catering to a very small group as if this will save yourselves from complete collapse. It won't.

    1. Re:AMD desperate by Teancum · · Score: 2

      AMD seems desperate to garner any headlines it can these days. How much open source gaming is really around, and how many will think this is going to sway any gaming developer over to AMD solely?

      While open source gaming might be a point here, the real market is toward even full commercial software development companies with decent software developers who can really dig into the source code to tweak their games to the full potential on this hardware. Since they don't need to pay licensing fees, the traditional retail game developers will be able to have some of their smaller projects or slightly risky games that may have slightly smaller budgets to use this hardware instead of projects where they pay hundreds of thousands of dollars for various kind of hardware development tools.

      That makes a huge difference here, and it means there is a real possibility you will see some real game innovation happening rather than the same old boring copies of copies of "safe" games that now are being made by the major studios.

      It isn't nearly the small group of developers writing games that go onto Source Forge or Git you should be talking about here. It is everybody else that could be interested in at least supporting some sort of alternative to Nvida, and there are plenty of really solid commercial reasons to be doing just that.

  10. Actual Status by bulled · · Score: 4, Interesting

    I see lots of articles about how AMD plans to do this, that, and the other using open source components. What I want to know, is can I run 3D games using the in-tree kernel module with the proprietary user modules yet? This was promised a while back and I haven't seen any more about it. I want to support the effort, but I am not buying another AMD card until I see it actually work.

    1. Re:Actual Status by Anonymous Coward · · Score: 0

      Don't know do some cards have issues, but I've been playing

      Cities: skylines,
      XCom
      Civ5
      Metro Redux
      Kerbal Space Program

      etc etc without issues.

    2. Re:Actual Status by Anonymous Coward · · Score: 0

      I played Command & Conquer Generals, and the follow up, Zero Hour, with wine, using the stock ATI opengl support, no proprietary driver loaded.

    3. Re:Actual Status by Anonymous Coward · · Score: 0

      why insist on the proprietatary modules? the open source drivers are pretty much just as fast and still getting better.

    4. Re:Actual Status by Anonymous Coward · · Score: 0

      Those games are also quite old, anon. That said, I remember having trouble running the original Unreal Tournament through WINE on a laptop with a mobile Radeon HD 5870 about 3 years ago. I know there's a native Linux port, but being that I was using the GoG version (in other words, no disc or iso), and that it hasn't been updated for modern distros, that option was out of the picture for me.

    5. Re:Actual Status by bulled · · Score: 1

      I should have been more clear, I don't care about proprietary in user space, as long as I can build and boot the mainline kernel of the moment and still play modern games (something I cannot do with nVidia).

    6. Re:Actual Status by Anonymous Coward · · Score: 0

      That's fine but this is about AMD. AMD is ATi.

    7. Re:Actual Status by Anonymous Coward · · Score: 0

      that is because AMD issues a press release for every little thing they do, or plan to do, or thought about doing, which might result in positive press. "AMD announces its landlord is repainting the lines on the parking lot AMD used to own..." "AMD announces that it's most recent round of layoffs adds hundreds of available engineers to the industry talent pool" "AMD announces that CEO Lisa Su found half a twix in her desk drawer..."

    8. Re:Actual Status by antdude · · Score: 1

      What about the (lat/new)est games?

      --
      Ant(Dude) @ Quality Foraged Links (AQFL.net) & The Ant Farm (antfarm.ma.cx / antfarm.home.dhs.org).
    9. Re:Actual Status by requerdanos · · Score: 2

      The "open source" drivers depend on nonfree binary blobs https://wiki.debian.org/AtiHow... and hence are not "free software" despite their code being free.

  11. Crying Wolf by Anonymous Coward · · Score: 1

    I think the appropriate response is "I'll believe it when I see it". AMD has cried wolf WAY too many times on this matter. I can remember getting excited about AMD "open sourcing" their stuff 10 years ago, but it turned out to be little more than a Microsoft "open sourcing", i.e. half-assed with strings attached.

    1. Re:Crying Wolf by Anonymous Coward · · Score: 0

      How do you mean? Open source ati drivers have been pretty good already for long time. Granted OpenGL 4 support has been long time coming, but newer card have had it since summer and it's finally in mesa git master for r600 also, so coming with mesa 11.2.

      Seems ok at for Radeon 6870 card I have at least. Been testing it for couple of days.

    2. Re:Crying Wolf by Anonymous Coward · · Score: 1

      This is the most dishonest post I've ever read on slashdot, congrats. They're "pretty good" if all you do is troll facebook and twitter or play 10 year-old and non-demanding Valve games. Try to play any recent non-Valve game and you are stuck with low fps and asking yourself why you would spend several hundred dollars on hardware only to gimp yourself with hardware from a company that simply can't compete anymore.

    3. Re:Crying Wolf by Anonymous Coward · · Score: 0

      They've been good except for that wake from sleep problem.

    4. Re:Crying Wolf by requerdanos · · Score: 2

      A little known fact: the "free/libre/open source" ATI/AMD drivers ("radeon" for example) are, in Debian parlance, more "contrib" (depends on non-free) than "free." The only completely FLOSS driver for the ATI/AMD cards is "vesa" (see "poor performance").

      Have a look, for example, at what the Trisquel folks say about ATI/AMD here: https://trisquel.info/en/wiki/...

      I think it would be *great* if this or something like it led to an actually free software driver. Most of my machines here have Radeons, and I am one of those nuts that values software freedom.

    5. Re:Crying Wolf by Anonymous Coward · · Score: 0

      Might be worth digging a bit deeper into the "non-free" claims. The issue is not "firmware" in the traditional sense (code running on the CPU) but GPU HW microcode images which we happen to load via the driver rather than burning them into the chip or loading them in the VBIOS. If we were to load the HW microcode images in VBIOS then the same driver would magically be "totally free" and the only change from a user perspective would be that they would have to wheedle VBIOS updates from their board/laptop vendor instead of picking up microcode fixes via a driver/package update.

      bridgman @ AMD

    6. Re:Crying Wolf by requerdanos · · Score: 2

      >> Might be worth digging a bit deeper into the "non-free" claims...

      This sounds right, but I don't think it is. Here's why:

      The README.radeon firmware/microcode license, for example, says...

      - no reverse engineering allowed as a specific condition of installation, reproduction, copying, or redistribution.
      - no decompilation allowed as a specific condition of installation, reproduction, copying, or redistribution.
      - no disassembly allowed as a specific condition of installation, reproduction, copying, or redistribution.
      - extraneous specific export law clause binding even for those not under the jurisdiction of the US law as a specific condition of redistribution.

      I see what you are saying, it's the same microcode whether it comes on the chip (fine, no problem) or is loaded by the driver, but the thing is, if it comes as software loaded by the driver, *with a nonfree license*, it's nonfree. That's not magic, it is specific words written by AMD. What's free about it is, quoting the license itself, "free of any license fees". That's also true, for example, of the binary-only drivers which are also nonfree.

      There's a copy of the license here: https://github.com/cernekee/li...

      I wish I knew enough about microcode and/or licensing to suggest a reasonable, practical way to release the microcode under a free license.

      But a nonfree license makes something nonfree. Make sense?

    7. Re:Crying Wolf by Anonymous Coward · · Score: 0

      If the microcode were truly "part of the driver" then I would agree completely that its non-free license makes the driver non-free. The question is whether the microcode is really part of the driver in this particular scenario.

      The FSF's current interpretation is that the use of any form of downloadable code that runs on the hardware makes the driver non-free, and I understand the rationale for that. It relates to network cards where a big chunk of what would have been driver logic gets replaced by code running on an on-chip general-purpose processor, partially defeating the purpose of having a free driver in the first place. Since that on-chip code tended to be downloaded rather than built into the chip (partly to deal with evolving network standards, and partly because the code was really big and not a good candidate for building into the chip) people started to associate any kind of downloadable code with attempts to get around free drivers by moving big chunks of functionality out of the driver, and that's where I think things went off the rails.

      Can anyone really say with a straight face that:
      - the drivers would be totally free if the microcode were built into the chip,
      - the drivers would be totally free if we stored the microcode in flash and loaded them in VBIOS code
      - the drivers would probably be considered free if we stored the microcode in flash and loaded them in driver code (o^O)
      - the drivers are unfree and not worthy of consideration if we distribute the microcode via distro package manager... ... when it's EXACTLY the same driver in all cases other than enabling/disabling the microcode loader ?

      I worked with the FSF on including microcode images and loader code in the VBIOS image for a while, but it was really hard to sell HW vendors on the idea of paying a few extra cents for a larger flash and the market shift towards mobile HW (where SBIOS and VBIOS share a single ROM). I would still like to see at least a couple of dGPU cards shipping with microcode images in the VBIOS flash if only to make sure everyone understands the situation, but haven't been able to sell that with HW vendors yet.

  12. It's different for hardware vendors. by Anonymous Coward · · Score: 0

    Where Netscape released their actual product, AMD sells hardware. This actually makes AMD's hardware significantly more valuable.

  13. proof is in the pudding by Gravis+Zero · · Score: 4, Insightful

    while i think this is great, i'll wait for AMD's GPU driver to actually show up in the kernel upstream before putting any stock in what they say.

    --
    Anons need not reply. Questions end with a question mark.
    1. Re:proof is in the pudding by chuckymonkey · · Score: 4, Informative

      It's been in the kernel for a few months and it's constantly being updated and fixed. They're actually doing a pretty good job of it honestly. The catalyst driver still doesn't use it, but they will soonish. Mesa uses it and it works really well actually. This is the earliest article I could find, more recent ones have mesa 11 benchmarks where the Radeon driver has almost caught up to the catalyst in some areas. http://www.phoronix.com/scan.p...

      --
      "Some books contain the machinery required to create and sustain universes."-Tycho
    2. Re:proof is in the pudding by Anonymous Coward · · Score: 0

      So the new AMDGPU driver was already included in Linux 4.2.

    3. Re:proof is in the pudding by Anonymous Coward · · Score: 1

      LOL. Thanks for the laugh, seriously. AMD graphics runs pretty much every console out there, and plenty of systems whose owner didn't care to go out and splurge for the hottest NVIDIA card out there. So sunk. LOL

    4. Re:proof is in the pudding by RavenLrD20k · · Score: 0

      Consoles are now AMD's niche market that they have nailed down. If I need GPU performance, the fiasco I've gone through with AMD's Catalyst drivers glitching out my AMD cards (ie HDMI audio crapping out for 3 or 4 driver iterations before it was finally working again in the update from September (I think)) dictates that for my Discrete needs, nVidia is my only choice. If I don't care to splurge on a system for running games or CUDA processes, then I'm going with Intel Integrated through their i7 line...or even more likely, completely headless through Xeon and set up WYSE or LIN terminals. Fuck AMD until they prove over at least 6 months to a year they've finally got themselves straightened out.

    5. Re:proof is in the pudding by Teancum · · Score: 1

      That is one of the things where an open source driver can really make a difference. Driver development will be more continuous rather than coming out in little spurts, and some 3rd party testing of anything changed will happen well before it becomes part of the stable release cycle.

      I understand your frustration here, and I would agree with your sentiment of waiting perhaps a year or more to see if this experiment is going to actually work out so far as some responsible driver development.

  14. More like an unholy alliance against nVidia by Kjella · · Score: 3, Interesting

    is this the last gasp for amd?
    i.e. similar to how Netscape went open source right before "Netscape" the commercial company went away (bought by aol... then becomes mozilla which outputs current firefox).

    Probably not, but it is an act of desperation. It's no secret that in the past AMD and nVidia has been very anxious to keep Intel out of the high performance graphics market. That gap is closing fast, even though they don't do discrete cards they're quite efficient and well-supported with almost 20% market share on Steam. Meanwhile nVidia has been very successful pushing their GameWorks middleware, G-Sync, CUDA and other proprietary nVidia-only technologies. So I think this is AMD realizing they can't win a war on two fronts and trying to make common cause with Intel to share AMD's middleware to get game support, while still hopefully being able to find a niche for their hardware.

    Of course the risk is that Intel just gobbles up AMD's graphics market share the same way Intel's almost completely gobbled up the x86_64 market but the way the gaming market is heading right now I don't think they have a choice. If letting Intel use their middleware can lead to better game support (probably) and Intel stays out of discrete cards (probably) and AMD can come up with discrete GPUs that match nVidia (maybe...) it might work. At least if this flops some good technology got open sourced, I don't like the implication that open sourcing is a last ditch attempt though. Intel is working hard on open source drivers in Mesa and that's hardly a failure.

    --
    Live today, because you never know what tomorrow brings
  15. Re:Wake me when linux-libre supports it... by Anonymous Coward · · Score: 0

    This should not have been down-modded. When I saw this article one of my first thoughts was was this going to help get a free vbios for use with libreboot. I am looking at getting the opteron based workstation, but I'd like more powerfull video support than what is provided by the motherboard.

  16. AMD on SoC by Anonymous Coward · · Score: 0

    If AMD had just opened the source for an ARM SoC-able GPU, I'd be on the edge of my chair. Although - they ARE making some noise about doing one in the future. I guess they have to wrestle with power consumption issues first though ...

  17. Oops... by bridgmanAMD · · Score: 1

    Sorry, realized I hadn't been logging in or identifying myself as an AMD-er. So now I have an account. It's funny, for such a technical site the account setup questions list a lot of management positions but nothing for the technical side. AFAIK most tech companies these days have parallel management and technical tracks, don't they ?

  18. Wish I could edit... by bridgmanAMD · · Score: 1

    (removing quote for clarity) Nope, I understand completely and you said it very well (better than I was doing). The problem is arguably as simple as "guilt by association"... ie in order to let the drivers load microcode that is distributed in hardware or flash on competing products you need to use an OS mechanism which also allows the use of other, obviously more problematic "firmware" blobs which clearly do perform functions that would otherwise be implemented in the driver. I don't know if there is a solution in there but if nothing else it might help to explain to board vendors why people care about not having to ship the microcode images as part of the "software". Thanks.

    1. Re:Wish I could edit... by Anonymous Coward · · Score: 0

      Sorry for a late answer. Also sorry for not having an account here.
      Im happy to see you understand the dilemma and that you also think it is a "problematic" situation. The only "quick solution" I can think of would be if AMD hosted your own repository that one could add to the sources. It would still be non-free but one would not have to add the whole non-free repo from the distro, and still know the blobs are from AMD.

      Secondly I would like to praise you for going this far for the community. I really think AMD is a good role model that is proving that cooperation is the way to go. :)
      I know a lot of people are suggesting to buy from other makers because of the blobs. That is a bad suggestion in my opinion because of the mentality of the competitors. AMD on the other hand have always been on the customers side. Thank you for that!

      Right now I am using a HD5850 but since its getting old I was thinking of buying a 380X and try out the AMDGPU driver. Ofcourse I would have to install the blobs to do that. After speaking to you, I might just manually install the blobs needed for the time being. Not sure how I am going to do.

      PS. I will see if I cant make an account so it will be abit easier to follow up on things.

    2. Re:Wish I could edit... by bridgmanAMD · · Score: 1

      The only "quick solution" I can think of would be if AMD hosted your own repository that one could add to the sources. It would still be non-free but one would not have to add the whole non-free repo from the distro, and still know the blobs are from AMD.

      That's an interesting idea, will look into it. Thanks !

  19. microcode by bridgmanAMD · · Score: 1

    http://people.freedesktop.org/~agd5f/radeon_ucode/tonga/

    AFAIK the microcode images can be downloaded directly from Alex's radeon_ucode folder on freedesktop.org (folder above is for 285/380/380X), but once folks come back from vacations I'll ask about packages that can take care of the install details.