Announcement

Collapse
No announcement yet.

AMD AOCC Compiler 1.1 Released For Zen CPUs

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

  • AMD AOCC Compiler 1.1 Released For Zen CPUs

    Phoronix: AMD AOCC Compiler 1.1 Released For Zen CPUs

    AOCC 1.1 is the second public release of the AMD Optimizing C/C++ Compiler designed for Ryzen/Threadripper/EPYC processors...

    http://www.phoronix.com/scan.php?pag...C-1.1-Released

  • #2
    It would certainly be interesting to see if there are any benefits over llvm - but not sure if there will be anything truly spectacular

    Comment


    • #3
      Pick llvm trunk of 30th Nov and compare, older or newer might be something else

      Extends LLVM 6.0 trunk (llvm.org, 30th Nov) with enhancements and optimizations

      Comment


      • #4
        Originally posted by dungeon View Post
        Pick llvm trunk of 30th Nov and compare, older or newer might be something else
        did a quick comparison using scimark2: clang-7 and aocc-1.1 vectorize better than gcc-7.3 (clang-7 is king), but both are inferior to gcc in the other tests (average over 5 runs).

        Comment


        • #5
          That is interesting. The older compiler was based on a different base compiler so I always assumed they did not want to go into the trouble of mainlineing their (alleged)optimizations. Now that they switched to llvm, why not as well submit the improvements to llvm?

          Comment


          • #6
            Originally posted by pattakosn View Post
            That is interesting. The older compiler was based on a different base compiler so I always assumed they did not want to go into the trouble of mainlineing their (alleged)optimizations. Now that they switched to llvm, why not as well submit the improvements to llvm?
            It'd be nice if they would, I write Rust which uses LLVM and wouldn't mind the improvements they made for optimizing to their hardware.

            Comment


            • #7
              Originally posted by mlau View Post
              did a quick comparison using scimark2: clang-7 and aocc-1.1 vectorize better than gcc-7.3
              why compare preprerelease of clang with mature release of gcc?

              Comment


              • #8
                Originally posted by pattakosn View Post
                That is interesting. The older compiler was based on a different base compiler so I always assumed they did not want to go into the trouble of mainlineing their (alleged)optimizations. Now that they switched to llvm, why not as well submit the improvements to llvm?
                Given that optimizations for Zen are always good for AMD, I'd guess they can't mainline it, rather than that they don't want to. My guess would be that they don't own all the intellectual property.

                Comment


                • #9
                  They'll mainline this stuff once it's mature

                  Comment

                  Working...
                  X