Announcement

Collapse
No announcement yet.

Linux Kernel Developers Discuss Dropping x32 Support

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

  • Linux Kernel Developers Discuss Dropping x32 Support

    Phoronix: Linux Kernel Developers Discuss Dropping x32 Support

    It was just several years ago that the open-source ecosystem began supporting the x32 ABI, but already kernel developers are talking of potentially deprecating the support and for it to be ultimately removed...

    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
    In the x32 link you should have linked to this article instead of a search for "x32", as it is hard to find what x32 really is, and hence is very prone to be confused with "32-bit applications under 64-bit".

    Comment


    • #3
      Seems reasonable. x32 is only interesting for a few niches and overall doesn't a noticeable benefit in most cases. And if it is beneficial in some specific case, it's more than good enough to restrict x32 usage to the userspace.

      Comment


      • #4
        I'm sure there'd be more users if more applications took advantage of it. I'm sure more applications would take advantage of it if devs weren't confused about it in the way tildearrow pointed out. Most software doesn't need 64 bit pointers.
        To me, dropping x32 isn't really a whole lot different than dropping support of all the features that makes Clear Linux so fast. If it weren't for Clear, I'm sure instructions like AVX or SSE4 would be a lot less used than they are now, but now that people are aware of the huge performance potential, they're suddenly a lot more interesting.

        Comment


        • #5
          Will this affect executing 32-bit binaries with wine? Or is this really non-userspace only, as they're also talking about the binutils and another package in the conversation.

          Comment


          • #6
            it's time, to save resources and time.

            Comment


            • #7
              Rather shortlived. From 3.14 and glibc 2.16 to now. A shame I think.

              Comment


              • #8
                RAM is precious ressource, x32 is important to save memory where you manipulate lot of pointer. Then mostly on optimized software.
                But: most of the software is not optimized (KDE is not), in general you need choose: features or optimisation, nobody work on both.
                Maintenance cost is important, disk/fs layer, GPU, network can be improved.
                Developer of Ultracopier/CatchChallenger and CEO of Confiared

                Comment


                • #9
                  Originally posted by brent View Post
                  Seems reasonable. x32 is only interesting for a few niches and overall doesn't a noticeable benefit in most cases. And if it is beneficial in some specific case, it's more than good enough to restrict x32 usage to the userspace.
                  A lot of things in the Linux kernel are "for a few niches".
                  I think x32 is underused. You could probably do more than 4G RAM with your 32-bit pointers. Since the machine is effectively 64-bit you could probably implement something like PAE and bouncing data between 32-bit windows without the massive PAE overhead. You could have more than 4G ram and benefit from the smaller data shoveling requirements on reasonably small applications.

                  Comment


                  • #10
                    Originally posted by alpha_one_x86 View Post
                    x32 is important to save memory
                    I think that saving memory wasn't even the main motivation for x32, it was saving memory bandwidth.
                    Especially on mobile platforms which have comparable amounts of RAM to low-end desktop systems, but much smaller memory bandwidth.

                    Comment

                    Working...
                    X