Announcement

Collapse
No announcement yet.

FreeBSD 9.1: LLVM/Clang Battling GCC

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

  • FreeBSD 9.1: LLVM/Clang Battling GCC

    Phoronix: FreeBSD 9.1: LLVM/Clang Battling GCC

    With LLVM/Clang having become the default FreeBSD x86 compiler as of last year and the recent FreeBSD 9.1 release shipping not only LLVM/Clang but also the libc++ library, new benchmarks were carried out of FreeBSD 9.1 looking at its two stock compilers.

    http://www.phoronix.com/vr.php?view=18447

  • #2
    Originally posted by phoronix View Post
    Phoronix: FreeBSD 9.1: LLVM/Clang Battling GCC

    With LLVM/Clang having become the default FreeBSD x86 compiler as of last year and the recent FreeBSD 9.1 release shipping not only LLVM/Clang but also the libc++ library, new benchmarks were carried out of FreeBSD 9.1 looking at its two stock compilers.

    http://www.phoronix.com/vr.php?view=18447
    Would be more interesting to see GCC 4.7 compared with Clang 3.2.

    Comment


    • #3
      C'mon Michael, you know that this test is _far_ from fair!

      How about you test GCC 3.2 vs the latest LLVM/Clang. I bet the latter one will win everywhere... But will it be fair? No.

      Comment


      • #4
        The benchmarks are meant to compare the old FreeBSD system compiler to the new one, not to allow the current versions of both compilers to duke it out - that's a totally different article.

        I'm sure GCC 4.7 vs Clang 3.2 will be coming soon :-) (if it hasn't already been done)

        Comment


        • #5
          Guys, the tests you are talking about have already been done (in fact, several times).... The most recent GCC 4.7/4.8 vs. LLVM Clang 3.1/3.2 - http://www.phoronix.com/vr.php?view=18321

          This article is just about the stock FreeBSD compiler performance...
          Michael Larabel
          http://www.michaellarabel.com/

          Comment


          • #6
            I thought they'd been done, I was just feeling too lazy to find the URL...

            Thanks for the article Michael, I've just set up a new FreeBSD box and was wondering about exactly this.

            Comment


            • #7
              Originally posted by Michael View Post
              Guys, the tests you are talking about have already been done (in fact, several times).... The most recent GCC 4.7/4.8 vs. LLVM Clang 3.1/3.2 - http://www.phoronix.com/vr.php?view=18321

              This article is just about the stock FreeBSD compiler performance...
              There's a little difference: they've been done on Linux.

              Comment


              • #8
                Oh, lol: GCC 4.2?

                Originally posted by Pawlerson View Post
                There's a little difference: they've been done on Linux.
                Oh, why they have to compete with grandfather of GCC aka gcc 4.2? It would make more sense to battle gcc 4.7 . Next time it's better to use GCC 2.95 or so. It would be easier to beat, 4.2 still kicks the butt in half of tests, which is not something *bsd/clang guys would like, he-he-he .

                Comment


                • #9
                  How about you drop tests requiring OpenMP for comparison until OpenMP is mainlined into LLVM/Clang 3.3? Retest in June.

                  Comment


                  • #10
                    Originally posted by Marc Driftmeyer View Post
                    How about you drop tests requiring OpenMP for comparison until OpenMP is mainlined into LLVM/Clang 3.3? Retest in June.
                    And why someone should make such discounts? We need operating systems here and now. Not "in June" or whenever. Let's go further and drop all tests where clang loses. Or even better, drop all tests where Linux + recent GCC beats BSDs to a hell. Then BSD guys would be happy for sure . Yet, I doubt this approach would make BSDs anyhow popular.

                    Comment


                    • #11
                      Originally posted by 0xBADCODE View Post
                      And why someone should make such discounts? We need operating systems here and now. Not "in June" or whenever. Let's go further and drop all tests where clang loses. Or even better, drop all tests where Linux + recent GCC beats BSDs to a hell. Then BSD guys would be happy for sure . Yet, I doubt this approach would make BSDs anyhow popular.
                      I don't quite get your logic. Why would dropping the OpenMP benchmarks (of course with explicitly stating in the article that they are dropped because Clang lacks support for it) making a difference in you making a decision? Do you really need to see a fancy graphic to realize that?
                      And how is a comparison of two compilers on the same platform intended to make the platform more popular?
                      This doesn't make much sense.

                      Comment


                      • #12
                        All I see is brand new clang being beaten by an out dated version of gcc.

                        In the areas where clang beats gcc, all margins are narrow. In the areas where gcc beats clang, the margins are very wide.

                        This benchmark is hardly a fair contest which heavily favors clang and yet gcc still beat the shit of of clang which goes to show how slow development of clang is compared to gcc.

                        Comment


                        • #13
                          Originally posted by Vim_User View Post
                          Would be more interesting to see GCC 4.7 compared with Clang 3.2.
                          It will be something like this http://www.phoronix.com/scan.php?pag...c_open64&num=5

                          Where clang beaten the shit out of and raped by gcc. Oh YEAH!!!!

                          Comment


                          • #14
                            Originally posted by systemd rulez View Post
                            All I see is brand new clang being beaten by an out dated version of gcc.

                            In the areas where clang beats gcc, all margins are narrow. In the areas where gcc beats clang, the margins are very wide.

                            This benchmark is hardly a fair contest which heavily favors clang and yet gcc still beat the shit of of clang which goes to show how slow development of clang is compared to gcc.
                            Not even remotely correct. The only time GCC won in these tests are where OpenMP is involved. Clang/LLVM does not support OpenMP in the latest stable, I think support is set to arrive in either the next release or the one after it--both coming later this year. Until OpenMP support lands in a stable release of Clang/LLVM testing it vs GCC is honestly fairly useless because there are some programs that have 2 paths OpenMP (fast) or No OpenMP (slow) or there are ones that have OpenMP (Fastest), No OpenMP but still multithreaded through standard libs (Faster), or no multithreaded at all (Slow). GCC is getting one path, LLVM/CLang is getting another and its pointless.

                            Comment


                            • #15
                              Originally posted by systemd rulez View Post
                              It will be something like this http://www.phoronix.com/scan.php?pag...c_open64&num=5

                              Where clang beaten the shit out of and raped by gcc. Oh YEAH!!!!
                              More something like this: http://www.phoronix.com/scan.php?pag...32_final&num=1

                              Comment

                              Working...
                              X