Announcement

Collapse
No announcement yet.

RadeonHD CS Driver Gets EXA Improvements

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

  • RadeonHD CS Driver Gets EXA Improvements

    Phoronix: RadeonHD CS Driver Gets EXA Improvements

    For those interested in playing with the latest open-source ATI code but don't want to dive into kernel mode-setting or the new AtomBIOS parser, you may want to test out the CS branch. The CS branch was started last week by Luc Verhaegen as it turns all calls made by the driver into macros...

    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
    If GCC doesn't optimize inline functions... Isn't that a HUGE bug then?

    Comment


    • #3
      Originally posted by phoronix View Post
      ...you may want to test out the CS branch.
      CS = Counter Strike?

      Comment


      • #4
        Command Submission
        Test signature

        Comment


        • #5
          Originally posted by Louise View Post
          If GCC doesn't optimize inline functions... Isn't that a HUGE bug then?
          It doesn't optimise properly in this case, so it is some sort of shortcoming... But then, the gcc version included in 10.3 is an evolution of a 2y old compiler, and the newer versions are supposedly better. GCC is like any other free software project, it's moving constantly, but never perfect

          Still, i'm stuck with 4 macros now, whereas i would've preferred to have gotten rid of them completely. But 10x speedup when turning RegWrite/Write into macros, and a few more percent for Grab/Advance, those are numbers i cannot go around

          The situation isn't that bad, the macros themselves are pretty small still.

          Maybe in a year or two, when even debian stable no longer shows such differences, we can get rid of them.

          Comment


          • #6
            Originally posted by libv View Post
            It doesn't optimise properly in this case, so it is some sort of shortcoming... But then, the gcc version included in 10.3 is an evolution of a 2y old compiler, and the newer versions are supposedly better. GCC is like any other free software project, it's moving constantly, but never perfect

            Still, i'm stuck with 4 macros now, whereas i would've preferred to have gotten rid of them completely. But 10x speedup when turning RegWrite/Write into macros, and a few more percent for Grab/Advance, those are numbers i cannot go around

            The situation isn't that bad, the macros themselves are pretty small still.

            Maybe in a year or two, when even debian stable no longer shows such differences, we can get rid of them.
            Very impressive =)

            Looking forward to when I can simply "./configure && make && make install" =)

            Comment


            • #7
              *wants Clang and LLVM to work for everything*

              Comment

              Working...
              X