“Big Endian” and “Little Endian” Formats for Integers
In base 10, integers are represented with the most significant digit first;
e.g., 9,263. A 16-bit (2’s complement) representation of this number in
base 2 or base 16 (hexadecimal, or hex) is given by
926310 = 0010 0100 0010 11112 = 24 2F16
and its inverse by
-926310 = 1101 1011 1101 00012 = DB D116.
If the representation is extended to 32-bit 2’s complement (represented
in hex) the numbers become
926310 = 00 00 24 2F16
and -926310 = FF FF DB D116.
This representation has the advantage of lining up the bytes in the way
people are used to writing numbers, which is not necessarily the best
way at the level of hardware.
A commonly used practice (e.g., Intel microprocessors employ this) is to
change the byte order of the number to have the lowest byte first (the
bits within each byte still have highest order bit first); e.g.,
926310 is stored in the order 2F 24 00 00
and -926310 is stored in the order D1 DB FF FF.
This has been dubbed the “little endian” format. The advantage is that
the number starts at the same address in memory, regardless of the
precision (16-bit, 32-bit, 64-bit, etc), facilitating construction of
extended precision routines for arithmetic regardless of data path width.
A byte order that conforms to the “natural” order has been dubbed “big
endian”, since the most significant byte comes first. It is not without
advantages. The sign of the number is on the first byte. For little
endian, you have to know how long the representation is and index to
the last byte to find the sign. Also since the order conforms to print
