In computers, information is stored electronically as the binary digits 0 and 1, also referred to as 'off' and 'on' respectively.
bit (1 b). One binary digit. The word "bit" is a contraction of 'binary' and 'digit'. The smallest unit of binary info.
nibble (4 b = 0.5 B)
byte (B = 2^3 b = 8 b). A byte is composed of 8 bits. A byte is equivalent to one ASCII character.
word (16 b = 2 B). Size of a Unicode character in UTF-16.
double word (32 b = 4 B). Size of Unicode character in UTF-32.
quad word (64 b = 8 B).
tenbyte (80 b = 10 B).
paragraph (128 b = 16 B).
page (32,768 b = 4,096 B). Assumes an ASCII text page at 8 b/character by 64 characters/line by 64 lines/page.
Beware that people get their bs and Bs mixed up! Except for shysters, most people should try to be consistent with using bits (b) when discussing rates and bytes (B) when discussing storage. Sometimes what is expressed in bits, actually refers to bpp (bits per pixel).
Since bits and bytes are binary, their multiples are also expressed in binary powers: 2^10, 2^30, 2^40, 2^50, 2^60. While almost everyone is consistent about meaning binary powers for bits, there is a lot of inconsistency when it comes to bytes.
The IEEE has suggested that small prefixes refer to decimal powers (EG: kb = 1000 b) and capital prefixes refer to binary powers (EG: Kb = 1024 b, while kb = 1000 b) but practically no one is consistent with this. EG: A "100 MB" disk by IEEE standards would be 100 MB = 100*2^20 B = 1.05E8 B = 1.05 mB, but the author of it may actually mean the smaller value of 100 mB = 100*10^6 B = 1.00E8 B = 95.4 MB. Also this sort of dilutes the purity of the SI prefixes.
The IEC (International Electrotechnical Commission) with standard 60027 has suggested using binary power prefixes that are different from metric's decimal power prefixes. EG: A kilobinary has a prefix of "kibi", a symbol of "Ki", and means 2^10 or 1024. This is removes ambiguity for binary powers but when people use the metric prefixes we can't tell if they mean decimal or binary powers.
My solution would remove all ambiguity by using the IEC system but also add new prefixes for decimal powers. EG: "MiB" for binary and "MeB" for decimal! I think it's a brilliant idea that I came up with today [2004-07-31].
280 ~ 1.21e24
270 ~ 1.18e21
260 ~ 1.15e18
250 ~ 1.13e15
240 ~ 1.10e12
230 ~ 1.07e09
220 ~ 1.05e06
k --> K
210 ~ 1.02e03 = 1024
Since this is my site, it shall be the law on this site henceforth. --> 2005-04-20: No way! My system was too annoying. Henceforth, my site will use SI prefixes (K, M, G, T, P, etc.) for binary units but the multiple will be binary powers instead of decimal. EG: When I say "4 MB", I mean "4 * 2^20 bytes" instead of "4e6 bytes".