Announcement

Collapse
No announcement yet.

GCC 5.2 vs. GCC 6.0 On An Intel Haswell-E Linux System

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

  • GCC 5.2 vs. GCC 6.0 On An Intel Haswell-E Linux System

    Phoronix: GCC 5.2 vs. GCC 6.0 On An Intel Haswell-E Linux System

    With GCC 6 feature development now over I decided to run some benchmarks comparing GCC 5.2.0 against GCC 6.0.0 (the 20151124 snapshot) on an Intel Haswell-E Xeon system running Ubuntu...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Most numbers are well within inaccuracy of measurements, and the only interesting result is probably JTR, which seems to expose some more or less steady boost on GCC 6.

    Comment


    • #3
      I am not expecting any performance boosts to C/C++ codes from new GCC versions because there is no indication GCC developers are working on groundbreaking features.

      I am more concerned about the fact that it takes GCC 5.2 more time to compile my C++ codes than GCC 4.9.3.

      Comment


      • #4
        as far as im concerned it can take all the time it likes as long as it is thinking longer about how to better compile my codes

        Comment


        • #5
          Originally posted by << ⚛ >> View Post
          I am not expecting any performance boosts to C/C++ codes from new GCC versions because there is no indication GCC developers are working on groundbreaking features.

          I am more concerned about the fact that it takes GCC 5.2 more time to compile my C++ codes than GCC 4.9.3.
          Depends on your definition of ground breaking. There is very active development on GCC optimizers: a lot of work is being put into maturing the link time optimization, support for offloading, improving inter-procedural optimizers (and adapting them to modern C++ programs), loop optimizers, alias analysis, hardware enablement and other areas. The news page is not very well maintained though.

          Comment


          • #6
            Looking at raw numbers when comparing performances is not insightful, you may like to add a percentage column.

            Comment


            • #7
              Originally posted by hubicka View Post

              Depends on your definition of ground breaking. There is very active development on GCC optimizers: a lot of work is being put into maturing the link time optimization, support for offloading, improving inter-procedural optimizers (and adapting them to modern C++ programs), loop optimizers, alias analysis, hardware enablement and other areas. The news page is not very well maintained though.
              Definition of "ground breaking": I think anything giving a 10% performance boost for /bin/bash qualifies as a groundbreaking compiler optimization.

              Comment

              Working...
              X