Announcement

Collapse
No announcement yet.

Why LLVM/Clang Was Ported To A Super Computer

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

  • Why LLVM/Clang Was Ported To A Super Computer

    Phoronix: Why LLVM/Clang Was Ported To A Super Computer

    Most often whenever writing about LLVM and its Clang C/C++ compiler front-end on Phoronix, within the forums is a flurry of comments from those in support of and against this modular compiler infrastructure. Some are against LLVM/Clang simply because its BSD-licensed and sponsored by Apple rather than the GPLv3-licensed GCC backed by the FSF. Others, meanwhile, see LLVM as presenting unique advantages and benefits. What reasons would a leading US national laboratory have for deploying LLVM/Clang to their leading super-computer? Here's an explanation from them...

    http://www.phoronix.com/vr.php?view=MTIzMDQ

  • #2
    I was under the impression LLVM's auto-vectorization was still much inferior to GCC's, which loses to ICC (by a much smaller margin though).

    An up-to-date look at this would be nice, but I suppose it goes out of Phoronix's targets.

    Comment


    • #3
      I guess you forgot to mention in the article that some people also prefer LLVM over GCC purely because of its BSD license rather than GCC's GPLv3 license. It goes both ways.

      The license issue itself is so polarizing that I wonder if {the people who consider LLVM and GCC purely for their ability to get stuff done for them} are in the minority.

      Comment


      • #4
        Contrib?

        It's great they're doing all that work, but are they contributing it back or just bragging? It is BSD after-all and that's not implicit.

        Comment


        • #5
          Originally posted by snadrus View Post
          It's great they're doing all that work, but are they contributing it back or just bragging? It is BSD after-all and that's not implicit.
          Of course they were invited to the LLVM Developers' Conference just to go over a few vague slides about technology they wrote just so they could go "but haha fuckers it's all ours!"

          Naturally, the GPL would have solved this problem by forcing them to proactively upstream all their work, including copyright assignments to the FSF, thereby allowing the code to actually be usefully used by anyone outside of the laboratory working on their own personal Blue Gene/Q supercomputers at home.

          Comment


          • #6
            Originally posted by allquixotic View Post
            The license issue itself is so polarizing that I wonder if {the people who consider LLVM and GCC purely for their ability to get stuff done for them} are in the minority.
            I don't think that's true at all, except among the population of people who post on message boards.

            Most people just use the most convenient compiler for their circumstances and don't give a lot of thought to anything else.

            Comment


            • #7
              Originally posted by smitty3268 View Post
              I don't think that's true at all, except among the population of people who post on message boards.

              Most people just use the most convenient compiler for their circumstances and don't give a lot of thought to anything else.
              I tend to agree. Another high-quality Free compiler can only be a good thing.

              What worries me is not LLVM's license, but the attempt to obsolete GCC, a bandwagon some journalists are happily jumping on.

              GCC and Clang coexisting is a great thing! Best thing, really. But if Clang ends up killing GCC, that would be a catastrophe for free software everywhere. And anyone wishing for such an outcome is dangerous for our community.

              Comment


              • #8
                Originally posted by elanthis View Post
                Of course they were invited to the LLVM Developers' Conference just to go over a few vague slides about technology they wrote just so they could go "but haha fuckers it's all ours!"

                Naturally, the GPL would have solved this problem by forcing them to proactively upstream all their work, including copyright assignments to the FSF, thereby allowing the code to actually be usefully used by anyone outside of the laboratory working on their own personal Blue Gene/Q supercomputers at home.
                With GPL you're allowed to make private modified versions, without any obligation to divulge the modifications as long as the modified software is not distributed to anyone else ... with GPL would be the same thing, so pleae stop trolling
                I don't ask you to read the license, but at least read the faq you know ... http://www.gnu.org/licenses/gpl-faq....cePostedPublic
                Last edited by vertexSymphony; 11-17-2012, 08:26 PM.

                Comment


                • #9
                  Originally posted by snadrus View Post
                  It's great they're doing all that work, but are they contributing it back or just bragging? It is BSD after-all and that's not implicit.
                  Yes, Hal is owner of several aspects of the code in LLVM and Clang. Spend some time on the Mailing lists and discover the collaboration levels on it. The lack of useless politics that ran rampant on GCC for years is non-existent on LLVM/Clang mailing lists.

                  A lot of improvements by Cray and Argonne Lab are rolling in and will continue to roll in with LLVM/Clang 3.3. AMD GPGPU R600 will be one of the newest sections under Targets.

                  Comment


                  • #10
                    Originally posted by pingufunkybeat View Post
                    I tend to agree. Another high-quality Free compiler can only be a good thing.

                    What worries me is not LLVM's license, but the attempt to obsolete GCC, a bandwagon some journalists are happily jumping on.

                    GCC and Clang coexisting is a great thing! Best thing, really. But if Clang ends up killing GCC, that would be a catastrophe for free software everywhere. And anyone wishing for such an outcome is dangerous for our community.
                    You clearly don't understand the scope and legal situation within the LLVM/Clang community. Ask Debian. They're quite thrilled with the progress of having 2 completely open source Compiler solutions.

                    Comment


                    • #11
                      Interesting list of features for for llvm/clang, but it is unclear to me how this provides any advantage over gcc 4.8, or for that matter llvm+gcc 4.8. Have to admit that I haven't really had a close look at llvm/clang yet (due to rather fundamental shortcomings such as openmp), so clarifications are highly appreciated.

                      On the licensing issue, I prefer gpl simply because it makes eee-tactics much harder, and because it has proven more effective for community driven projects. Seeing how Apple misbehaved when they forked khtml into webkit, the fact that they are behind llvm/clang makes me very wary. It is not a matter of ideology, it is actually very pragmatic.

                      Comment


                      • #12
                        Originally posted by Marc Driftmeyer View Post
                        You clearly don't understand the scope and legal situation within the LLVM/Clang community. Ask Debian. They're quite thrilled with the progress of having 2 completely open source Compiler solutions.
                        No offense, but have you actually read my comment?

                        I'm thrilled with having two completely open source compiler solutions too. What's legal situation within the LLVM community got to do with anything I said?

                        Comment

                        Working...
                        X