# Kilo as 1 024, Mega as 1 048 576

Under SI, the kilo prefix refers to 1,000 and mega refers to 1,000,000. In strict computer terms, kilo refers to 1,024 (210), mega refers to 1,048,576 (220) and giga is 1,073,741,824 (230).

The original question was how widely known outside this domain is this distinction?

Going by the answers provided, I should be using binary prefixes in order to remove the ambiguity in my documentation.

-
The question is: what is your answerable question? – Jürgen A. Erhard Jan 10 '11 at 17:17
Agree with @jae. You need to reformulate this question. See en.wikipedia.org/wiki/Binary_prefix – ghoppe Jan 10 '11 at 17:20
Last time that I counted, 1M people (1,000,000). – Ivo Rossi Jan 10 '11 at 17:20
I'll second @jae. If your question is really "how many", I'm afraid we can't reliably answer that. And even if it's more like "are there many", that's quite subjective still. – RegDwigнt Jan 10 '11 at 17:22
All the answers — all possible answers — are subjective; this question is really a near-canonical example of "subjective and argumentative". I vote to close. – ShreevatsaR Jan 10 '11 at 18:51

Not many. At least that’s presumed — by the people who then invented Kibi and Mebi and Gibi. Yup, there’s sort-of (I don’t deign to honor them as) SI prefixes: kB is kilobytes, that is, 1000 bytes (some used “KB” to signify 210 bytes, but that’s really nonstandard). While kiB, kibibytes, is 1024 bytes.

Likewise for MB -> MiB (Mebibytes) and GB -> Gibibytes.

The bi here stands for binary, so the long longform could be Kilobinarybytes. It’s all weird anyway.

I guess one could presume that even fewer people “outside the domain” are aware of those new binary prefixes than were aware of the old power-of-two one, if you consider that even people inside said domain are unaware of those new prefixes.

-
Not to mention that few people in general know about the SI prefixes. :D – Jürgen A. Erhard Jan 10 '11 at 17:24
It's all a scam by hard drive manufacturers anyway! – Eric Jan 10 '11 at 17:25
@Eric sort-of. They did use the precise SI definition of the prefixes when it was generally (in the field) understood that K, M etc were power-of-two prefixes, which the (desired) effect to make HDs looks bigger than they "actually" were. The "binary" prefixes are an attempt to work around that. An apparently failed one. – Jürgen A. Erhard Jan 10 '11 at 17:30
False. Even in the computer industry how much 1 kilohertz means (1000 hertz), or how much 1 megabit-per-second means (1000000 b/s). If you look at history, the hard-drive manufacturers were using kilo/mega correctly (powers of 1000) even before the RAM people came up with their powers-of-1024 abuse of terminology. The new prefixes, increasing in popularity, are hopefully a way of both resolving confusion, and (because they are so awkward :p) discouraging the powers-of-1024 units that users don't need to be bothered with. (It's happening slowly: already Mac OS X uses decimal units.) – ShreevatsaR Jan 10 '11 at 18:03

Not so many. Most people within the computer domain aren't even aware of all the details.

The slightly larger kilo unit actually have gotten it's own unit, or actually a set of units. 1024 bytes is one kibibyte, 1024 kibibyte is one mebibyte, and so on.

Even within the computer domain, both types of units are used. Disk storage manufacturers for example uses 1000 as kilo to measure disk sizes, while operating systems uses 1024 as kilo to measure disk usage. That means that a 2.0 terabyte disk only has room for 1.82 terabyte data. (Also, as some of the space is used to keep track of the data on the disk, the actual amount that you can store on the disk is even less.)

Even disk manufacturers aren't consistent, and both units has even been mixed to measure the size of a single type of disk. The (once) so well known 1.44 megabyte 3.25" floppy disk is actually neither 1.44 megabyte nor 1.44 mebibyte. It's actually 1.44 * 1000 * 1024 bytes, or 1.44 kilo-kibi-byte...

-
+1 for at least the kilo-kibi-byte :D – Jürgen A. Erhard Jan 10 '11 at 17:31
In my opinion, if people wanted these binary-based units to actually catch on, they should have — at the very least — consulted their own internal linguistic intuition and realized that "kibibyte" and "mebibyte" are awkward (or even silly) sounding and quite difficult to articulate properly in rapid speech. They probably thought that the -bi- part would relate to the fact that it is binary, but come on: people are supposed to actually speak these words! – Kosmonaut Jan 10 '11 at 21:05
"while operating systems uses 1024 as kilo to measure disk usage" Microsoft Windows does, but Mac OS X and Ubuntu have both converted to follow the standards. – endolith Mar 28 '12 at 19:54
@Kosmonaut: gibibyte sounds more silly than gigabyte? No, you're just used to it. – endolith Mar 28 '12 at 19:55

The answer is not many. That's why there is an increasing move to refer to use "kibi", "mebi" and so on as prefixes for these powers of two. Wikipedia has an article about it: http://en.wikipedia.org/wiki/Binary_prefix

-

Umm, not very many? The computer definitions aren't much use outside computer stuff, so I'd say that the vast majority of non-computer people are ignorant of the distinction (or don't care enough to remember it).

But really, I could guess at numbers all day. How on Earth would we know the answer to this?

-

Regardless of how many people know or don't know about it, if you're writing a specification or technical publication, my suggestion is to use the binary prefixes (kibi, mebi), that way there is no room for ambiguity.

-
That, or at the beginning of the document state which kind of kB/MB/GB you are talking about – Mr. Shiny and New 安宇 Jan 10 '11 at 19:52
@Mr. Shiny, that's also a good idea. – Stephen Furlani Jan 10 '11 at 19:53

Actually, kilo-, mega-, etc. are ambiguous, because some think they ought to be based on powers of 2 in a computing context and powers of 10 otherwise, some disagree, and both sides just go ahead and use what they prefer. To solve this, the prefixes kibi- (210), mebi- (220), gibi- (230), and tebi- (240) have been introduced; these I find are rarely used in whole form, but there's increasingly common use of their abbreviations KiB, MiB, GiB, and TiB. I for one always say "K", "meg", "gig", and "terabyte".

The Wikipedia entry on binary prefixes sums up the issues nicely.

-
Well, even "computing context" isn't quite it… within the computer context, people consistently use 1 kilohertz to mean 1000 hertz, 1 megabit-per-second to mean 1000000 b/s. It's only when it comes to measuring bytes, somehow, that some people are attached to the abuse of terminology of kilo meaning 1024.:-) Apart from kibi-/mebi-/gibi- etc. being used, there's also (thankfully) some user-friendly software that use the standard prefixes in the same senses that people are familiar with from all other aspects of their life (i.e., kilo meaning 1000 etc.): Mac OS X 10.6, apt-get, fdisk, gparted… – ShreevatsaR Jan 10 '11 at 18:08
@Jon: "Binary is the basis of our modern computing systems" is, again, the same sort of thing that's true only in restricted contexts. (After all, don't clocks and data transfer have something to do with the "our modern computing systems" as well?) The fact is, unlike RAM which was built on binary trees (having to do with addressing: that's why 32-bit, 64-bit), hard disks are not built on binary trees. So there is no reason to measure hard disk space in powers of two. Besides, as you said, normal users don't have to be tortured with binary even some things internally happen in binary. :-) – ShreevatsaR Jan 10 '11 at 20:06
@Orbling: As I said, it is only with memory addressing (having to do with memory, not disk) that binary of any relevance, and besides, whatever is used internally, there is no reason to confuse users with made-up abuse of terminology. There were already some wise voices among the engineers who warned of the mess when they started to abuse "kilo" to mean 1024; unfortunately they were ignored, engineers continued to be hacky, and we have the mess today. :-) – ShreevatsaR Jan 10 '11 at 20:19
@Orbling: BTW, your earlier comment illustrates the lengths to which engineers, supposedly rational creatures valuing clarity, will go to defend sth they're merely used to.:-) Although 8=2^3, powers of 2^8 would be groups of 8 in binary, not of 3. Further 2^8=256, not 1024, which is 2^10—groups of 10 in binary. But it was chosen rather because it's close to 1000—groups of 3 in *decimal*—because numbers are written in decimal. And history: en.wikipedia.org/wiki/Timeline_of_binary_prefixes (notice e.g. "65K" for 65536) – ShreevatsaR Jan 11 '11 at 5:54
@Orbling: To repeat, memory (RAM) sizes are usually powers of 2, so it makes sense to use powers of 2 there. But disk and file sizes may be any number of bytes; it's just a number. There's no reason to use so-called "binary" (meaning decimal, except that some large power of 2 has been divided out) for them. The claim of "convenience" does not actually stand up to scrutiny. (How many megabytes is 1.76 gigabytes? With standard prefixes it's clear—1760 megabytes—but with the nonstandard prefixes you have to multiply by 1024; conversely (e.g) 1503 MiB=1.4873046875 GiB: too many digits! Rounding? – ShreevatsaR Jan 11 '11 at 6:07