Announcement

Collapse
No announcement yet.

Hangover Project Restarted To Run Windows 32-bit/64-bit Apps On ARM64/PPC64 & More

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

  • Hangover Project Restarted To Run Windows 32-bit/64-bit Apps On ARM64/PPC64 & More

    Phoronix: Hangover Project Restarted To Run Windows 32-bit/64-bit Apps On ARM64/PPC64 & More

    Several years ago the open-source "Hangover" project started as allowing Windows x86_64 and x86 programs to run on 64-bit ARM Linux by leveraging the upstream Wine software as well as QEMU. Hangover also saw work for allowing Windows programs to run on POWER9 Linux hardware and other architectures. The Hangover project has been on hiatus but is now back to being revived...

    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
    Note that it's less about WOW64 itself than separating unix and PE parts, which is what WOW64 also uses, but is not the same thing.

    Comment


    • #3
      If you're running qemu (assumig there is a systarget for the host arch), why not run plain windows?
      Doesn't that work?

      Comment


      • #4
        But can it run on RISC-V?

        Comment


        • #5
          Oh hey, I used this before. I had to run some Windows GUI while travelling, and it actually worked on my Android phone.

          Originally posted by milkylainen View Post
          If you're running qemu (assumig there is a systarget for the host arch), why not run plain windows?
          Doesn't that work?
          Systems that dont support windows, for one ^. Otherwise, maybe native ARM linux system packages (other than the app you are trying to emulate) run better than Windows?

          Comment


          • #6
            Originally posted by milkylainen View Post
            If you're running qemu (assumig there is a systarget for the host arch), why not run plain windows?
            Doesn't that work?
            For one, things like GPU acceleration work with hangover, whereas GPU acceleration to a VM, especially without a second GPU, is rather difficult.
            For another, this looks like it can go in any direction, such as ARM PE running on POWER, which would be quite the task if one wanted to run ARM Windows.

            Comment


            • #7
              Originally posted by milkylainen View Post
              If you're running qemu (assumig there is a systarget for the host arch), why not run plain windows?
              Doesn't that work?
              Overhead. Massive, massive overhead.

              You could run a full-blown copy of windows, plus all the services and other crap it has going on in the background, plus the GUI stuff, plus the game you actually want to run, all in ultra-slow emulation. But that sucks ass.

              Instead, the whole point of the hangover work is to emulate ONLY the application in QEMU, with syscalls and such going through translation to the host. Run an x86 game binary, but every time it tries to access a windows API or the GPU, translate the call to ARM and do it on the native ARM linux host.

              QEMU has had this user-mode emulation mode for years now, but it only supports running linux applications on a linux host. ie. running an x86 linux application on an ARM linux host. Hangover brings wine into the picture so that the emulated application can be a windows one instead of a linux one.
              Last edited by Developer12; 11 February 2023, 05:30 PM.

              Comment


              • #8
                Originally posted by uid313 View Post
                But can it run on RISC-V?
                Patches have only been written/proposed for ARM and POWER9. When this work was last underway it was basically impossible to get any RISC-V hardware, and that's still mostly true. There's basically no RISC-V hardware out there that could run any of this stuff with enough performance to make it worthwhile. Hypothetically though, it could be done.

                Comment


                • #9
                  I hope that with ARM support being worked on again we eventually see POWER9 work too following it. Those systems are definitely not lacking in processing power.

                  Comment


                  • #10
                    Originally posted by Developer12 View Post
                    I hope that with ARM support being worked on again we eventually see POWER9 work too following it. Those systems are definitely not lacking in processing power.
                    POWER's unpopularity is kind of sad TBH.

                    Not only is POWER free (unlike ARM), its also well fleshed out (while RISC-V is still working out stuff you'd probably want in large cores) and has some very performant shipping silicon as proof that it works.

                    Comment

                    Working...
                    X