Announcement

Collapse
No announcement yet.

It Looks Like Systemd 231 Will Soon Be Released, Adds MemoryDenyWriteExecute

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

  • #61
    Originally posted by starshipeleven View Post
    Because it makes no sense to. Each program in 64bits uses X more ram, period.
    As I showed in the previous post, the first N applications used X amount of RAM on 32-bits and 1.3X on 64-bits. Now if N grows, X grows and thus 1.3X also grows. 1.3X grows a lot faster than X. Basic mathematics.

    Good luck with that, as you need to do it manually.
    If you think that's hard, well.. ok. You can treat it as a 32-bit system with one extra package maintained outside the package manager.

    Comment


    • #62
      Originally posted by caligula View Post
      As I showed in the previous post, the first N applications used X amount of RAM on 32-bits and 1.3X on 64-bits. Now if N grows, X grows and thus 1.3X also grows. 1.3X grows a lot faster than X. Basic mathematics.
      Ah ok, I thought that by "grows" you meant that 1.3 becomes 1.4 or 1.5. The % should remain similar, and given the usage for most old hardware it's not like a huge issue anyway.

      If you think that's hard, well.. ok. You can treat it as a 32-bit system with one extra package maintained outside the package manager.
      I'm not saying it is hard, the package manager can install 64bit kernels (or whatever) too in 32bit debian for example. Then on reboot you choose that kernel.
      It's that it's not a very used combination, so unknown bugs may lie ahead, and I'd rather waste a bit more RAM than doing shenanigans and get smacked by bugs or othe rissues

      Comment


      • #63
        Originally posted by caligula View Post

        Guess what, that's purely theoretical. In practice, the apps use more RAM. I recommend downloading some live distro like Ubuntu and trying it out. The numbers might surprise you. I did some calculation in the previous post above ^

        Also, I'm not really arguing about the perf benefits. It's clear that x86-64 provides more registers, wider registers, more instructions (especially SIMD), different memory model etc. They're not relevant if you are constrained by RAM like you usually are on a low end machine -- these days, even when just doing simple web browsing.
        Well it's of course theoretical in the sense that the programmers have to take this into consideration when programming as I wrote. If you however take it into consideration then it's no longer theoretical but practical which also of course is why I wrote what I did because I have to program in this way since we at my work have to cram out every little byte of RAM and ounce of performance of the CPU (I write server software that handles millions of transactions per second). So I know that it can be done.

        Then of course just because it can be done does not mean that it gets done, that you are correct in.

        Comment

        Working...
        X