A fantastic article that goes into a lot of depth and rigor until I got to the very end:
> GB here means 2^30, as it should when talking about base-2 memory. The only people who think it should be 10^9 are hard drive salesmen and the type of people who like to correct all their friends by saying “It’s centripetal, not centrifugal force!”. Also “gibibyte” sounds like Pokémon invented by a first grader.
This attitude ruined it for me. You're a technologist. You should care about vocabulary and precision. You are on the wrong side of history for both of these things - a gigabyte is 10^9 bytes as per SI, and centrifugal force is fictitious.
I challenge you: If a 1 GHz processor can process 1 byte on every cycle, how long does it take to process a 1 GB file? Hopefully you'll see the error in your ways.
Fragmenting the definition of the giga- prefix based on context leads us down some dark paths that we already traversed with the pound-mass vs. pound-force, pound Avoirdupois vs. pound Troy, fluid ounce vs. weight ounce, US gallon and UK gallon, liquid bushel vs. dry bushel, different varieties of tons, statute mile vs. nautical mile, and the list goes on.
Good luck fitting a 16 "GB" RAM dump on a 16 GB flash drive when you hibernate the computer.
Filesystem overhead notwithstanding, it's worth noting that NAND flash usually had true capacities too, with some additional hidden extra for error correction/wear-leveling. I have a 16MB SLC one that truly has 16,777,216 usable bytes (32M 512-byte sectors), and a slightly newer 2GB one with 2147483648 bytes.
Let's not forget the original IBM PC "10MB" hard drive held a little more than 10MB: 10653696 bytes, a bit more than the 10485760 one would expect.
No one seriously talks about "GiB", and the only ones I hear using it are pedantic assholes. It just sounds absurd and stupid.
As for JEDEC, standards bodies are made of people and people are fallible.
I can say the same goes for SI and their "iB" nonsense.
> you just have to ask for 16 GiB of RAM, as per the proper definitions.
If you walked in an asked for 16 "gibibytes" of RAM they would just look at you funny. Or think you were non-native speaker.
I work in the hardware software boundary, and I have never once heard someone use the term kibibyte, mebibyte, or gibibyte in real life. Occasionally I will see KiB, MiB, and GiB in an academic paper, but that's about it. When I am talking with people and they say kilobyte, they always mean 1024. Trying to pretend otherwise would just cause unneeded confusion. That's what it meant went the term was coined, that what it meant when IEC tried to change the definition 25 years ago, and (unless you are selling hard drives) that's what it means to most people today.
I agree that it is unfortunate that they reused the terms from SI prefixes, since they are not SI prefixes. And if I had a time machine I would tell them to use different prefixes. But they didn't and those are the terms the world has adopted. It's also unfortunate that hard drive manufactures played fast and loose with the definition to pretend like their drives were bigger then they really were. Words are democracy, and they mean what people say they mean, not what the IEC says they should mean.
> As for JEDEC, standards bodies are made of people and people are fallible.
Exactly, so why is the IEC redefinition somehow more valid?
You're needlessly diluting the meaning of words and making them context-dependent. You create problems at the boundaries, like when 1 GHz faces off with 1 "GB". You're helping to undo the unification of definitions that SI strives for. You're doing this in the name of keeping tradition - which is exactly how we end up with messes like US Customary units.
P.S. Don't forget how "1.44 MB" floppy disks are neither MB nor MiB; their definition is MB = 1024000 bytes. This free-for-all benefits no one. Hard drive manufacturers were right, RAM was wrong.
> GB here means 2^30, as it should when talking about base-2 memory. The only people who think it should be 10^9 are hard drive salesmen and the type of people who like to correct all their friends by saying “It’s centripetal, not centrifugal force!”. Also “gibibyte” sounds like Pokémon invented by a first grader.
This attitude ruined it for me. You're a technologist. You should care about vocabulary and precision. You are on the wrong side of history for both of these things - a gigabyte is 10^9 bytes as per SI, and centrifugal force is fictitious.
I challenge you: If a 1 GHz processor can process 1 byte on every cycle, how long does it take to process a 1 GB file? Hopefully you'll see the error in your ways.
Fragmenting the definition of the giga- prefix based on context leads us down some dark paths that we already traversed with the pound-mass vs. pound-force, pound Avoirdupois vs. pound Troy, fluid ounce vs. weight ounce, US gallon and UK gallon, liquid bushel vs. dry bushel, different varieties of tons, statute mile vs. nautical mile, and the list goes on.