• pivot_root@lemmy.world
    link
    fedilink
    English
    arrow-up
    9
    arrow-down
    22
    ·
    edit-2
    2 months ago

    Metric is excellent until it gets into data units. There shouldn’t be a difference between 4T and 4TB, but it’s actually a (10244-10004) ≈ 92.6G (99.5GB) difference because of the fuckers who decided to make data units metric and rename the base-2 data units to “kibibyte”/“mibi*”/“gibi*” (KiB/MiB/GiB)

    • x00za@lemmy.dbzer0.com
      link
      fedilink
      English
      arrow-up
      32
      ·
      2 months ago

      People were using them ambiguously so a real standard was made which is the kibibytes. Vendors and even OSes define KB differently, but KiB will always be base 2. It’s stupid yes, but making the original one base 10 was not deliberate.

      • blusterydayve26@midwest.social
        link
        fedilink
        English
        arrow-up
        7
        ·
        edit-2
        2 months ago

        People weren’t using them ambiguously, drive manufactures picked a non-standard unit to lie with on their boxes, and then tricked courts into going along with their shit because it was the old case of money vs truth.

    • megane-kun@lemmy.dbzer0.com
      link
      fedilink
      English
      arrow-up
      15
      ·
      2 months ago

      I think the biggest mistake there is using SI prefixes (such as kilo, mega, giga, tera) with bytes (or bits) to refer to the power of two near a power of ten in the first place. Had computer people had used other names for 1024 bytes and the like, this confusion between kibibytes and kilobytes could have been avoided. Computer people back then could have come up with a set of base·16 prefixes and used that for measuring data.

      Maybe something like 65,536 bytes = 1,0000 (base 16) = 1 myri·byte; ‭4,294,967,296 bytes = 1,0000,0000 (base 16) = dyri·byte; and so on in groups of four hex digits instead of three decimal digits (16¹² = tryri·byte, 16¹⁶ = tesri·byte, etc). That’s just one system I pulled out of my ass (based on the myriad, and using Greek numbers to count groups of digits), and surely one can come up with a better system.

      Anyways, while it’d take me a while to recognize one kilobyte as 1000 bytes and not as 1024 bytes, I think it’s better that ‘kilo’ always means 1000 times something in as many situations as possible.

      • sep@lemmy.world
        link
        fedilink
        English
        arrow-up
        2
        ·
        edit-2
        2 months ago

        Everybody knew exactly what kilo mega and giga ment. when drive vendors deliberatly lied on there pdf’s about their drive sizes. Warnings were issued: this drive will not work in a raid as a replacement for same size!!. And everybody was throwing fumes on mailinglists about the bullshit situation.

        But money won, as usual.

        Source: threw fumes!

        • megane-kun@lemmy.dbzer0.com
          link
          fedilink
          English
          arrow-up
          2
          arrow-down
          1
          ·
          2 months ago

          Not too sure if they outright lied, but I suppose we can say that they used the change to make their drives seem larger!

          That’s why I wished computer people had used a prefix system distinct from the SI ones. If we’re measuring our storage devices in yeetibytes rather than gigabytes, for example, then I suppose there’s less chance that we’ve ended up in this situation.

      • michaelmrose@lemmy.world
        link
        fedilink
        English
        arrow-up
        3
        arrow-down
        2
        ·
        2 months ago

        There is no reason whatsoever to use base 16 for computer storage it is both unconnected to technology and common usage it is worse than either base 2 or 10

        • megane-kun@lemmy.dbzer0.com
          link
          fedilink
          English
          arrow-up
          5
          ·
          2 months ago

          I guess? I just pulled that example out of my ass earlier, thinking well, hexadecimal is used heavily in computing, so maybe something with powers of 16 would do just fine.

          At any rate, my point is that using a prefix system that is different and easily distinguishable from the metric SI prefixes would have been way better.

          • michaelmrose@lemmy.world
            link
            fedilink
            English
            arrow-up
            1
            ·
            2 months ago

            They could have easily used base 2 which is actually connected to how the hardware works and just called it something else

            • megane-kun@lemmy.dbzer0.com
              link
              fedilink
              English
              arrow-up
              2
              ·
              2 months ago

              I realized why I didn’t think of base 2 in my previous reply. For one, hexadecimal (base 16) often used in really low-level programming, as a shorthand for working in base 2 because base 2 is unwieldy. Octal (base 8) was also used, but not so much nowadays. Furthermore, even when working in base 2, they’re often grouped into four bits: a nibble. A nibble corresponds to one hexadecimal digit.

              Now, I suppose that we’re just going to use powers of two, not base-2, so maybe it’d help if we do a comparison. Below is a table that compares some powers of two, the binary prefixes, and the system I described earlier:

              Decimal value Value with corresponding binary prefix Hexadecimal Value Value with prefixes based on powers of 16
              20 1 1 1 1
              24 16 16 10 16
              28 256 256 100 256
              210 1 024 1 Ki 400 1 024
              212 4 096 4 Ki 1000 4 096
              216 65 536 64 Ki 1 0000 1 myri
              220 1 048 576 1 Mi 10 0000 16 myri
              224 16 777 216 16 Mi 100 0000 256 myri
              228 268 435 456 256 Mi 1000 0000 4 096 myri
              230 1 073 741 824 1 Gi 4000 0000 16 384 myri
              232 4 294 967 296 4 Gi 1 0000 0000 1 dyri
              236 68 719 476 736 32 Gi 10 0000 0000 16 dyri
              240 1 099 511 627 776 1 Ti 100 0000 0000 256 dyri
              244 17 592 186 044 416 16 Ti 1000 0000 0000 4 096 dyri
              248 281 474 976 710 656 256 Ti 1 0000 0000 0000 1 tryri
              250 1 125 899 906 842 624 1 Pi 4 0000 0000 0000 4 tryri
              252 4 503 599 627 370 496 4 Pi 10 0000 0000 0000 16 tryri
              256 72 057 594 037 927 936 64 Pi 100 0000 0000 0000 256 tryri
              260 1 152 921 504 606 846 976 1 Ei 1000 0000 0000 0000 4 096 tryri
              264 18 446 744 073 709 551 616 16 Ei 1 0000 0000 0000 0000 1 tesri

              Each row of the table (except for the rows for 210 and 250) would be requiring a new prefix if we’re to be working with powers of 2 (four apart, and more if it’d be three apart instead). Meanwhile, using powers of 16 would require less prefixes, but would require larger numerals before changing over to the next prefix (a maximum of 164 - 1 = 216 - 1 = 65 535)

              One thing that works to your argument’s favor is the fact that 1024 = 210. But I think that’s what caused this entire MiB vs. MB confusion in the first place.

              However, having said all that, I would have been happy with just using an entirely different set of prefixes, and kept the values based on 210.