Announcement

Collapse
No announcement yet.

Benchmarking GCC 4.2 Through GCC 4.8 On AMD & Intel Hardware

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

  • Benchmarking GCC 4.2 Through GCC 4.8 On AMD & Intel Hardware

    Phoronix: Benchmarking GCC 4.2 Through GCC 4.8 On AMD & Intel Hardware

    Here are benchmarks of all major GNU Compiler Collection releases from GCC 4.2.4 through the latest GCC 4.8 development build. Benchmarking was of the seven GCC compiler releases from an Intel Core i7 "Clarksfield" system and an AMD Opteron "Shanghai" workstation.

    http://www.phoronix.com/vr.php?view=17642

  • #2
    I wonder if GCC hackers actually use PTS for regression testing? While its easy to benchmark and compare side-by-side, personally I have quite a trouble navigating via webinterface.. :/
    And huge thanks for the article!!

    Comment


    • #3
      outdated hardware running at vastly different clock speeds and core counts being compared. This is the most uninformative article from Phoronix in a long time.. can we please compare a Bulldozer Opteron versus Intel's latest and have them run at similar clock speeds with the same number of hardware threads ("cores" for AMD and "threads" for Intel via hyperthreading)? or a Bulldozer Opteron versus an equally priced Intel? either way.

      I know the AMD would probably get stomped, but that would be useful information, and it excited me to see that information so I came here to read it and got disappointed.

      Comment


      • #4
        coder543, you seem to have missed the point of the article. It is comparing versions of GCC, not Intel vs. AMD.

        Comment


        • #5
          A couple of interesting things I see:
          1. In C-Ray, intel was faster with gcc 4.2, amd ends up faster with gcc 4.7/4.8.
          2. In FLAC audio encoding, the stiuation is actually reversed (though the differences are smaller this time): amd start out on top end ends up at the bottom.

          Comment


          • #6
            Originally posted by kiwi_kid_aka_bod View Post
            coder543, you seem to have missed the point of the article. It is comparing versions of GCC, not Intel vs. AMD.
            the newer architectures have hugely different personalities and would respond to the different versions of GCC in completely unique ways that these don't.

            Comment


            • #7
              That doesn't invalidate these results in any way?

              Comment


              • #8
                Originally posted by coder543 View Post
                outdated hardware running at vastly different clock speeds and core counts being compared.
                They are not being compared. Only GCC versions are being compared. Perhaps you missed that each image has two groups of comparisons, one for the Intel system, one for the AMD system.

                It's done in the correct way, so that you can compare the performance of the generated code relative to the machine. You are not supposed to compare the Intel results to the AMD results.

                Comment


                • #9
                  Michael,

                  What GCC flags did you use during testing?

                  Comment


                  • #10
                    Originally posted by birdie View Post
                    Michael,

                    What GCC flags did you use during testing?
                    The compiler flags are underneath each graph. If any specific flags were used (e.g. -O3), they are listed beneath the graph for the individual test. The flags that each compiler was built with are at the bottom of the system specs table.

                    I'm guessing that for each of these tests, the flags are the default that are used in the individual PTS test profiles.

                    Comment


                    • #11
                      I'd love to see ffmpeg compared with ASM switched off - so we can actually see which is the better compiler of C code and produces the faster binaries - yes I know no one in their right mind would use these binaries on a real system but the results could be normalised against the ASM builds

                      Comment


                      • #12
                        Originally posted by bug77 View Post
                        A couple of interesting things I see:
                        1. In C-Ray, intel was faster with gcc 4.2, amd ends up faster with gcc 4.7/4.8.
                        2. In FLAC audio encoding, the stiuation is actually reversed (though the differences are smaller this time): amd start out on top end ends up at the bottom.
                        Now the question is why? I wish Phoronix had a big editorial staff...

                        Comment


                        • #13
                          Originally posted by devius View Post
                          Now the question is why? I wish Phoronix had a big editorial staff...
                          (1) In 4.7 Gcc recieved SMP improvements, for example bulldozer rendering went up around 50%. There was an article on it.

                          Comment


                          • #14
                            Originally posted by devius View Post
                            Now the question is why?
                            Basically, because there's no free lunch. Few optimizations will improve performance across the board. Uusually you win some, you gain some.

                            Comment


                            • #15
                              Originally posted by crazycheese View Post
                              (1) In 4.7 Gcc recieved SMP improvements, for example bulldozer rendering went up around 50%. There was an article on it.
                              But why didn't those changes also affect Intel systems the same way?

                              Comment

                              Working...
                              X