Announcement

Collapse
No announcement yet.

GCC 4.9 vs. 5.3 vs. 6.0 Compiler Benchmarks On Debian 8.4

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

  • GCC 4.9 vs. 5.3 vs. 6.0 Compiler Benchmarks On Debian 8.4

    Phoronix: GCC 4.9 vs. 5.3 vs. 6.0 Compiler Benchmarks On Debian 8.4

    With GCC 6.1 due out soon with its plethora of new features and improvements, I decided to run some fresh benchmarks this week of GCC 4.9.3 vs. GCC 5.3 vs. GCC 6.0.0 on a Debian stable system...

    http://www.phoronix.com/scan.php?pag...5.3-6.0-Debian

  • #2
    Cool!

    BTW, what are the noise margins here?

    Comment


    • #3
      Don't see big difference between them. Then what do newer gcc versions do? Supporting more kinds of cpu's? Generating faster code?

      Comment


      • #4
        Originally posted by dxxvi View Post
        Don't see big difference between them. Then what do newer gcc versions do? Supporting more kinds of cpu's? Generating faster code?
        https://gcc.gnu.org/gcc-5/changes.html
        https://gcc.gnu.org/gcc-6/changes.html

        Comment


        • #5
          Originally posted by vladpetric View Post
          Cool!

          BTW, what are the noise margins here?
          It's shown on each of the graphs when relevant.
          Michael Larabel
          http://www.michaellarabel.com/

          Comment


          • #6
            1. Such simple tests are pointless. Main improvements lie in -flto part and C++14 features etc, if we ignore new architectures here ( Zen etc). So main test example would be to try it on some big library and/or executable, compiled with aggressive optimisations ( -O3 at least) and -flto. Simple tests as those shown, are adequate just for quick check for regressions, nothing more.

            2. gcc-6.0 is AFAIK in the works, so shouldn't this be the first gcc-6 release ?



            Comment


            • #7
              Originally posted by Brane215 View Post

              2. gcc-6.0 is AFAIK in the works, so shouldn't this be the first gcc-6 release ?
              gcc version numbering is the oddest I have ever seen

              https://gcc.gnu.org/develop.html

              Comment


              • #8
                Originally posted by Brane215 View Post
                1. Such simple tests are pointless. Main improvements lie in -flto part and C++14 features etc, if we ignore new architectures here ( Zen etc). So main test example would be to try it on some big library and/or executable, compiled with aggressive optimisations ( -O3 at least) and -flto. Simple tests as those shown, are adequate just for quick check for regressions, nothing more.

                2. gcc-6.0 is AFAIK in the works, so shouldn't this be the first gcc-6 release ?
                I think -O3 is a bit harsh since it is generally advised NOT to use O3 since it can actually result in incorrect calculations, etc. -O2 is a more reasonable choice, don't you think?

                Comment


                • #9
                  No O3 will not produce incorrect results. Unsafe optimizations are only done for Ofast. O3 will just take longer time and produce larger binaries than O2.

                  Comment


                  • #10
                    -O3 -march=native is usually the default for Phoronix tests, isn't it? I mean, no point in comparing the speed of unoptimized code.

                    The only reason people recommend not to use -O3 is that it often breaks incorrect code, i.e. code that relies on undefined behaviour to behave in a specific way, which it may do with -O2. Can't blame the compiler for that.

                    Comment

                    Working...
                    X