77

What is the difference between Big Endian and Little Endian Byte order ?

Both of these seem to be related to Unicode and UTF16. Where exactly do we use this?

Kiran
  • 1,067
  • 1
  • 11
  • 25
web dunia
  • 9,381
  • 18
  • 52
  • 64

5 Answers5

129

Big-Endian (BE) / Little-Endian (LE) are two ways to organize multi-byte words. For example, when using two bytes to represent a character in UTF-16, there are two ways to represent the character 0x1234 as a string of bytes (0x00-0xFF):

Byte Index:      0  1
---------------------
Big-Endian:     12 34
Little-Endian:  34 12

In order to decide if a text uses UTF-16BE or UTF-16LE, the specification recommends to prepend a Byte Order Mark (BOM) to the string, representing the character U+FEFF. So, if the first two bytes of a UTF-16 encoded text file are FE, FF, the encoding is UTF-16BE. For FF, FE, it is UTF-16LE.

A visual example: The word "Example" in different encodings (UTF-16 with BOM):

Byte Index:   0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15
------------------------------------------------------------
ASCII:       45 78 61 6d 70 6c 65
UTF-16BE:    FE FF 00 45 00 78 00 61 00 6d 00 70 00 6c 00 65
UTF-16LE:    FF FE 45 00 78 00 61 00 6d 00 70 00 6c 00 65 00

For further information, please read the Wikipedia page of Endianness and/or UTF-16.

Ferdinand Beyer
  • 64,979
  • 15
  • 154
  • 145
  • Endianness is perpetually counter-intuitive in that BE stores the most significant byte in the smallest address, not the last/end address. Whatever. This site makes things clear, [Big and Little Endian](https://www.cs.umd.edu/class/sum2003/cmsc311/Notes/Data/endian.html), _In big endian, you store the most significant byte in the smallest address._ BTW, the _Visual Example_ was helpful. – bvj Sep 06 '16 at 07:16
  • If you wish, you could change around the words to make more sense: [_In big endian, you store the most significant byte in the smallest address._] OR [_In big endian, you store the least significant byte in the largest address._] Same thing – Lakey Dec 28 '16 at 15:28
  • 2
    Link broken @bvj this one works: https://www.cs.umd.edu/~meesh/cmsc311/clin-cmsc311/Lectures/lecture6/endian.htm – eric Dec 04 '21 at 15:15
  • I still don't see how does this answer help decide BE or LE i should choose for my subtitle files. – zhuhang.jasper Mar 04 '23 at 04:35
35

Ferdinand's answer (and others) are correct, but incomplete.

Big Endian (BE) / Little Endian (LE) have nothing to do with UTF-16 or UTF-32. They existed way before Unicode, and affect how the bytes of numbers get stored in the computer's memory. They depend on the processor.

If you have a number with the value 0x12345678 then in memory it will be represented as 12 34 56 78 (BE) or 78 56 34 12 (LE).

UTF-16 and UTF-32 happen to be represented on 2 respectively 4 bytes, so the order of the bytes respects the ordering that any number follows on that platform.

Justin T.
  • 3,643
  • 1
  • 22
  • 43
Mihai Nita
  • 5,547
  • 27
  • 27
8

UTF-16 encodes Unicode into 16-bit values. Most modern filesystems operate on 8-bit bytes. So, to save a UTF-16 encoded file to disk, for example, you have to decide which part of the 16-bit value goes in the first byte, and which goes into the second byte.

Wikipedia has a more complete explanation.

joev
  • 1,384
  • 9
  • 18
  • 1
    this answer is incorrect. endianess is related to the underlying hardware architecture – Mitch Wheat Mar 31 '09 at 15:48
  • 7
    You can store a UTF-16 encoded file in either byte order regardless of the underlying hardware. – joev Mar 31 '09 at 15:51
  • 1
    Given in the context of the question, this answer is perfectly acceptable IMHO – Binary Worrier Mar 31 '09 at 15:56
  • @joev: Exactly. It often *is* related to hardware architecture, but needn't necessarily be. For cross-platform compatibility, Unicode encoders/decoders should therefore be able to use either endianness. – Noldorin Mar 31 '09 at 15:56
5

little-endian: adj.

Describes a computer architecture in which, within a given 16- or 32-bit word, bytes at lower addresses have lower significance (the word is stored ‘little-end-first’). The PDP-11 and VAX families of computers and Intel microprocessors and a lot of communications and networking hardware are little-endian. The term is sometimes used to describe the ordering of units other than bytes; most often, bits within a byte.

big-endian: adj.

[common; From Swift's Gulliver's Travels via the famous paper On Holy Wars and a Plea for Peace by Danny Cohen, USC/ISI IEN 137, dated April 1, 1980]

Describes a computer architecture in which, within a given multi-byte numeric representation, the most significant byte has the lowest address (the word is stored ‘big-end-first’). Most processors, including the IBM 370 family, the PDP-10, the Motorola microprocessor families, and most of the various RISC designs are big-endian. Big-endian byte order is also sometimes called network order.

---from the Jargon File: http://catb.org/~esr/jargon/html/index.html

Paul Reiners
  • 8,576
  • 33
  • 117
  • 202
1

Byte endianness (big or little) needs to be specified for Unicode/UTF-16 encoding because for character codes that use more than a single byte, there is a choice of whether to read/write the most significant byte first or last. Unicode/UTF-16, since they are variable-length encodings (i.e. each char can be represented by one or several bytes) require this to be specified. (Note however that UTF-8 "words" are always 8-bits/one byte in length [though characters can be multiple points], therefore there is no problem with endianness.) If the encoder of a stream of bytes representing Unicode text and the decoder aren't agreed on which convention is being used, the wrong character code can be interpreted. For this reason, either the convention of endianness is known beforehand or more commonly a byte order mark is usually specified at the beginning of any Unicode text file/stream to indicate whethere big or little endian order is being used.

Noldorin
  • 144,213
  • 56
  • 264
  • 302
  • this answer is incorrect. endianess is related to the underlying hardware architecture – Mitch Wheat Mar 31 '09 at 15:48
  • 1
    UTF-8 is a variable-length encoding, using 1-6 bytes per character and is thus not fixed to a single byte as stated here! – Ferdinand Beyer Mar 31 '09 at 15:50
  • Right, so I haven't stated that endianness depends on hardware architecture, but I don't see how my answer is explicitly incorrect. Consider that text files written/read on different architectures must have their endianness known. – Noldorin Mar 31 '09 at 15:51
  • @Ferdinand: You are correct - I should mention that *some* variants of UTF-8 do not require it... – Noldorin Mar 31 '09 at 15:52
  • Sorry, you still don't got it right. There are no variants of UTF-8 that don't require multiple bytes. If you only use ASCII characters, UTF-8 will represent them using single bytes. All characters with character code >127 will be encoded using multiple bytes! – Ferdinand Beyer Mar 31 '09 at 16:11
  • 1
    Just for completeness - UTF-8 requires between 1 and 4 bytes. Valid UTF-8 cannot contain more than 4 bytes. – Nemanja Trifunovic Mar 31 '09 at 16:12
  • @Noldorin: As I said, ASCII characters use a single byte. This is a *property* of UTF-8, not a variant! Using single-bytes, you cannot encode Non-ASCII unicode values. – Ferdinand Beyer Mar 31 '09 at 16:58
  • 1
    @Ferdinand: Yes, I've realised that since your original correction. Post has been clarified again, as I see your point... though I think I somewhat confused myself in the process of correcting myself. :P – Noldorin Mar 31 '09 at 18:17
  • (contd.) I think I'm right in saying that because "words" in UTF-8 are 8-bits/one byte long (invariably, despite the variable length of char codes), then there is no problem with endianness at least. – Noldorin Mar 31 '09 at 18:18
  • 2
    http://unicode.org/faq/utf_bom.html seems to agree, though again correct me if I'm wrong... – Noldorin Mar 31 '09 at 18:19