Announcement

Collapse
No announcement yet.

Mold 1.3 High Speed Linker Released With LTO Improvements

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

  • Mold 1.3 High Speed Linker Released With LTO Improvements

    Phoronix: Mold 1.3 High Speed Linker Released With LTO Improvements

    Mold 1.3 has been released today as the newest version of this high-speed linker that serves as an alternative to GNU Gold and LLVM's LLD...

    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
    Mold is one of the tool which I didn't realize I needed, Rui work is always exceptional, it literally saved me a lot of hours.

    Comment


    • #3
      there is also a lot of work on the Mac OS side: https://github.com/rui314/mold/searc...o&type=commits
      Once Mac support is done, Windows will be work on : https://github.com/rui314/mold/issues/190

      Comment


      • #4
        Originally posted by vtorri View Post
        there is also a lot of work on the Mac OS side: https://github.com/rui314/mold/searc...o&type=commits
        Once Mac support is done, Windows will be work on : https://github.com/rui314/mold/issues/190
        Thanks for that links.

        Where can Mold make more performance?

        On Linux for non static compiled programs already.
        So *.so files created with Mold are faster then with ld ?
        Why then isn't the GNU ld replaced with Mold?
        Why llvm/clang don't doing it, is clear: Because of the current license. But why don't doing GNU it?

        As you pointed out, it would also help to improve the performance of the linked libaries on macOS.

        And for PE/COFF it would also help. Would it also help if anyone used Visual C++ instead of GCC or LLVM ?
        So it would help to improve the handling with *.DLL files ?

        And what is with Virtual Machines like Java and .NET ? Could it also improve the performance of programs with a lot of class files or do it only help for native binaries?

        Greetings
        theuserbl

        Comment


        • #5
          If I understand you correctly, you think that mold makes the resulting shared lib faster. But mold does not speed up the code in shared libs : it makes the linking step faster. So the creation of the shared lib is faster.

          Comment


          • #6
            Lots of good progress, but still a long way to go.

            Comment


            • #7
              Originally posted by theuserbl View Post

              Thanks for that links.

              Where can Mold make more performance?

              On Linux for non static compiled programs already.
              So *.so files created with Mold are faster then with ld ?
              Why then isn't the GNU ld replaced with Mold?
              Why llvm/clang don't doing it, is clear: Because of the current license. But why don't doing GNU it?

              As you pointed out, it would also help to improve the performance of the linked libaries on macOS.

              And for PE/COFF it would also help. Would it also help if anyone used Visual C++ instead of GCC or LLVM ?
              So it would help to improve the handling with *.DLL files ?

              And what is with Virtual Machines like Java and .NET ? Could it also improve the performance of programs with a lot of class files or do it only help for native binaries?

              Greetings
              theuserbl
              Unless I am mistaken, Mold merely speeds up overall compilation time by being a more efficient linker. It doesn’t affect binary performance at all.

              As a gentoo user who does LTO, using one linker between compilers simplifies my environments a bit so that’s good. The time saving on compilation is a nice bonus.

              Comment

              Working...
              X