Announcement

Collapse
No announcement yet.

NVIDIA, Mentor Graphics May Harm GCC

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #21
    Originally posted by schmidtbag View Post
    I'm not trying to be a suck-up, but you'd think by now readers here either know when to take something with a grain of salt or know that you have yet damage your credibility. I've been visiting this website just about every day for several years and I got to say, a lot of the community here is pointlessly negative. I'm not sure what it takes to get more credibility, but I think getting the Source Engine benchmarks will help.
    thanks, especially when people think I would put out a pointless article just after receiving a lot of great hardware from nvidia...
    Michael Larabel
    https://www.michaellarabel.com/

    Comment


    • #22
      Obvious troll is obvious.

      Comment


      • #23
        Originally posted by Michael View Post
        thanks, especially when people think I would put out a pointless article just after receiving a lot of great hardware from nvidia...
        Let's cut the crap and be honest with ourselves for a second. You tried to throw nvidia under the bus here and based on what information? Hearsay that, in the end, is of no real consequence anyway. Unless you can better explain this evil that nvidia is potentially inflicting on the world...?

        I believe the word you used was............ "poisonous".

        Poisonous, how? Because NVIDIA is paying a third party to put OpenACC compiler support for their GPUs into GCC?

        Did your contact mention for exactly how much "Mentor Graphics is selling GCC to NVIDIA"?

        ...Maybe I'm confused here but I thought GCC was an entity owned by the FSF and they were free to make their own decisions...

        Comment


        • #24
          Originally posted by XorEaxEax View Post
          Hmm... it's very early morning here so I may be missing something, but can someone explain to me how OpenACC will 'force multiple closed source dependencies' on GCC?

          PTX is not 'closed source', it's an undocumented instruction set, however if OpenACC were to be supported in GCC, and it's done through intermediary PTX code generation, how would that possibly be part of GCC without being documented by it's open source implementation ? (it has to be open source to be included in GCC to begin with)
          Parallel Thread Execution (PTX) is a pseudo-assembly language. The graphics driver contains a compiler which translates the PTX into a binary code which can be run on the processing cores.

          So yes the PTX compiler is a closed source implementation. Sure GCC would have the open source code to generate PTX but then it has a dependency on the closed source driver to compiler it.

          Originally posted by XorEaxEax View Post
          Of course, if this has been about LLVM getting OpenACC support, Michael would have spun this story into something super-positive. Also when LLVM got a PTX backend from NVidia there was no talk from Michael about how 'LLVM is being sold to NVidia', instead it was described as 'new possibilities are opened up'.
          The point is GCC is a free software foundation project LLVM isnt. Whether you believe in the principles of the foundation or not it is a big deal for a FSF project to become reliant on non-free software no matter how small that reliance may be.

          Originally posted by johnc View Post
          Let's cut the crap and be honest with ourselves for a second. You tried to throw nvidia under the bus here and based on what information? Hearsay that, in the end, is of no real consequence anyway. Unless you can better explain this evil that nvidia is potentially inflicting on the world...?

          I believe the word you used was............ "poisonous".

          Poisonous, how? Because NVIDIA is paying a third party to put OpenACC compiler support for their GPUs into GCC?

          Did your contact mention for exactly how much "Mentor Graphics is selling GCC to NVIDIA"?

          ...Maybe I'm confused here but I thought GCC was an entity owned by the FSF and they were free to make their own decisions...
          You are choosing to ignore or belittle the issue because you do not share the same values. Adding support to a FSF project that relies on non-free software may not matter to you but it should matter to the FSF and it should be brought to Nvidias attention. I believe this is what was being referred to as poisonous, once you give in and loosen your values its hard to say no to the next thing that comes along and wants to restrict how free your software is.

          Comment


          • #25
            Originally posted by tarceri View Post
            Parallel Thread Execution (PTX) is a pseudo-assembly language. The graphics driver contains a compiler which translates the PTX into a binary code which can be run on the processing cores.

            So yes the PTX compiler is a closed source implementation. Sure GCC would have the open source code to generate PTX but then it has a dependency on the closed source driver to compiler it.



            The point is GCC is a free software foundation project LLVM isnt. Whether you believe in the principles of the foundation or not it is a big deal for a FSF project to become reliant on non-free software no matter how small that reliance may be.



            You are choosing to ignore or belittle the issue because you do not share the same values. Adding support to a FSF project that relies on non-free software may not matter to you but it should matter to the FSF and it should be brought to Nvidias attention. I believe this is what was being referred to as poisonous, once you give in and loosen your values its hard to say no to the next thing that comes along and wants to restrict how free your software is.
            So you're Michael's "source"?

            Comment


            • #26
              Nvidia, what the fuck do you want? Decide or begone!

              Phoronix is great. Steam is great. DRM is great, give me my digital handcuffs please! I'll at least wear them until linux becomes mainstream...am I contradictory now?

              Btw, I think Nvidia has lots of trolls in the company itself, it sent them here to troll as they really are angry that Nvidia refuses to stay 'closed source only'(hey it said to Nouveau coders, "hey ask us anything") and as it opens up it may reveal their great incompetence in their coding skills as Nvidia will publish that source code. They went here to vent their frustration that soon it will be proven that they didn't even have a computer science diploma.

              Nvidia of course partway wants to stay as a closed source company, and another little part of it wants to embrace open source. Nvidia, what the fuck do you want anyway?? Make up your mind!! Intel will bust both AMD and Nvidia's asses, we will rule your asses one day with Broadwell! We are ruling your asses now with Haswell so your days are numbered haha!

              Comment


              • #27
                Originally posted by johnc View Post
                So you're Michael's "source"?
                Haha, no I'm just someone who is interested in making sure people reading/writing these posts understand what's going on. It only takes a little bit of research and common sense to come to these conclusions.

                You have every right not to think this matters, but those concerned with free software should. You might have noticed this news was posted under Free Software news while the LLVM news was under Compiler News.

                Comment


                • #28
                  While AMD has documented their various generations of GPU ISA, NVIDIA does not provide any documentation on their instruction set architecture. Intel, ARM, and other processor companies also document their ISA publicly. This source had written in a private email, "NVIDIA refuses to publicly disclose their instruction set. They force any company targeting their hardware to use PTX. They claim it's for portability, but that's a lie. AMD's GPU instruction set changes, but they and *EVERYONE* else in the industry publishes their details no problem. The arguments about protecting their IP is just bs.. How can *everyone* else publish the details and open source drivers, but stay in business?"
                  As far as I know ARM doesn't provide documentation for their GPUs. Imagination don't either. And Qualcomm don't.

                  In fact, I guess only Intel and AMD have provided doc for their instruction set, so Michael's contact seems wrong about everyone publishing documentation...

                  Comment


                  • #29
                    Hope that Intel would release discrete graphic cards some day. Intel is the only big guy that sincerely supports open source and open standard. Nvidia? If I were to consider nividia, sadly that's because there are too few to choose from....

                    Comment


                    • #30
                      Originally posted by tarceri View Post
                      Sure GCC would have the open source code to generate PTX but then it has a dependency on the closed source driver to compiler it.
                      And? I also need a Nvidia card to use it in the first place. Oh my god, they are taking my freedom away to use any card I want!!1!11

                      Originally posted by tarceri View Post
                      You are choosing to ignore or belittle the issue because you do not share the same values. Adding support to a FSF project that relies on non-free software may not matter to you but it should matter to the FSF and it should be brought to Nvidias attention. I believe this is what was being referred to as poisonous, once you give in and loosen your values its hard to say no to the next thing that comes along and wants to restrict how free your software is.
                      How does this take away any of your freedom? Before they had an OpenACC to OpenCL backend. Now they will have an additional OpenACC to PTX backend. This does not take away any freedom, but adds more freedom!

                      I'm with the people who call this article pure FUD. Articles like this are the reason why Michael no longer gets AMD cards. Improved interoperability between open source software and closed source software is never poisonous, as long as it is optional. People who do not want to use this new backend can still decide not to compile it into their gcc and their gcc version will just work fine as before. The gcc compiler as a whole does not depend on the Nvidia driver, only the openacc->ptx backend does.

                      Comment

                      Working...
                      X