you are viewing a single comment's thread
view the rest of the comments
[–] 54 points 2 days ago (16 children)

The OS uses 1024 and the drive manufacturers are using 1000.

Which is why a 1TB drive shows 931GiB of capacity. (1024 GiB =1 TiB, 1024MiB =1 GiB, 1024KiB =1MiB, 1024 Bytes =1KiB) but the drive manufacturers using SI prefixes doing 1000 Bytes =1 KB, 1000 KB = 1MB, 1000MB = 1GB, 1000GB = 1TB.

Thus the discrepancy.

  • source
  • hideshow 16 child comments
  • [–] 33 points 2 days ago (4 children)

    The problem arises when the OS uses GiB but labels them GB, like Windows does.

  • source
  • parent
  • hideshow 4 child comments
  • [+] -13 points 1 day ago (3 children)

    No, the problem is making up a new word (gibibyte) and expecting people to use it when there’s already a word for that (gigabyte).

    It’s the hard drive manufacturers that should change.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 6 points 1 day ago (2 children)

    I want to agree with you, but the war is lost. Giga is billion and Gibi is 2^30

  • source
  • parent
  • hideshow 2 child comments
  • [–] 9 points 13 hours ago* (1 child)

    I want to agree with you two, because I hate competing standards, but you gotta admit gigabyte was a poor choice from the start because it took the SI prefix and made it mean something else. Creating Gibi is just correcting that mistake

  • source
  • parent
  • hideshow 1 child comment
  • [–] 15 points 1 day ago (3 children)

    Yeah:

    1000 GB × 1000 × 1000 × 1000 / 1024 / 1024 / 1024 = 931.32 GiB

    It's a sufficient explanation. I don't know why the meme has the highest IQ person say "I don't know", unless the joke is that it's easier than explaining...

  • source
  • parent
  • hideshow 3 child comments
  • [–] 5 points 12 hours ago

    I don’t know why the meme has the highest IQ person say

    Every time I see the midwit meme format, it reads like OP imagines that they're the 145 IQ guy who is above it all, when all other evidence suggests that they're closer to the 55 IQ guy.

  • source
  • parent
  • [–] 6 points 1 day ago

    Which itself is for marketing reasons, a manufacturer would find it difficult to compete with others using units that make their drives look larger or cheaper at "same capacity".

    There is also the smaller amount w!ndows will secretly waste for its own pleasure (few GiB iirc). And an even smaller amount that gets wasted in partitioning. Usually around 5MiB, you can reduce it to 16kiB with some manual configuration in gdisk, or you can skip partition tables entirely for some end-game cursedness. And then the filesystem will take some. ext4 takes based on the number of inodes, for large drives that can be tens of GiB. Btrfs takes almost nothing (though in the background allocates in steps of 1GiB, so some tools might show 2GiB used as soon as data is written). Then if you try to sum all files on disk, that will land even lower for many more reasons.

  • source
  • parent
  • [–] 4 points 1 day ago (1 child)

    At the beginning of time, K/M/G etc was based on 1024. This was considered a bad move by some people.

    What I don't get is memory packages don't come in thousands, but in volumes dividable by 1024, so why the 1TB=1000 somethings...

  • source
  • parent
  • hideshow 1 child comment
  • [–] 0 points 1 day ago

    Because back in the 1980s RAM chips were individually sold and the number of address pins exactly matched the 2^(# of pins) amount of memory addresses contained in the chip (for cost savings purposes) and only with the later column strobe addressing DRAM did that change but at that point they couldn’t deviate from the 1024 standard without getting sued (a couple did and went out of business).

    And since the RAM chips themselves are stuck using 1024 and the RAM sticks are made of matching RAM chips (for cost and performance reasons); it will remain that way until people start buying systems on a chip with embedded RAM.

  • source
  • parent
  • [–] 2 points 1 day ago (1 child)

    And it gets more complicated when we get to the metadata and filesystem structure. So, you technically never get to store 931GB, I wonder what the (average) ratio of metadata to data is? 3:1?

  • source
  • parent
  • hideshow 1 child comment
  • [–] 2 points 1 day ago

    Depends on the file system. If you are using FAT-8; it is 256:1 if you are on ZFS it is as low as 1:4.

    It also depends on composition of the data stored. If it is Linux ISOs; it is 4GB:64KB. If it is old C source code; it is 512B:4KB on ext4

  • source
  • parent