Vector data types

The Vector/SIMD Multimedia Extension model adds a set of fundamental data types, called vector types.

Vector types are shown in Table 1. The represented values are in decimal (base-10) notation. The vector registers are 128 bits and can contain:

Sixteen 8-bit values, signed or unsigned
Eight 16-bit values, signed or unsigned
Four 32-bit values, signed or unsigned
Four single-precision IEEE-754 floating-point values

The vector types use the prefix vector in front of one of standard C data types—for example vector signed int and vector unsigned short. A vector type represents a vector of as many of the specified C data type as will fit in a 128-bit register. Hence, the vector signed int is a 128-bit operand containing four 32-bit signed ints. The vector unsigned short is a 128-bit operand containing eight unsigned values.

Note: Since the token, vector, is a keyword in the Vector/SIMD Multimedia Extension data types, you are recommended not to use the term elsewhere in the program (for example, as a variable name).

Table 1. Vector/SIMD Multimedia Extension data types
Vector Data Type	Meaning	Values
vector unsigned char	Sixteen 8-bit unsigned values	0 ... 255
vector signed char	Sixteen 8-bit signed values	-128 ... 127
vector bool char	Sixteen 8-bit unsigned boolean	0 (false), 255 (true)
vector unsigned short	Eight 16-bit unsigned values	0 ... 65535
vector unsigned short int	Eight 16-bit unsigned values	0 ... 65535
vector signed short	Eight 16-bit signed values	-32768 ... 32767
vector signed short int	Eight 16-bit signed values	-32768 ... 32767
vector bool short	Eight 16-bit unsigned boolean	0 (false), 65535 (true)
vector bool short int	Eight 16-bit unsigned boolean	0 (false), 65535 (true)
vector unsigned int	Four 32-bit unsigned values	0 ... 2³² - 1
vector signed int	Four 32-bit signed values	-2³¹ ... 2³¹ - 1
vector bool int	Four 32-bit unsigned values	0 (false), 2³¹ - 1 (true)
vector float	Four 32-bit single precision	IEEE-754 values
vector pixel	Eight 16-bit unsigned values	1/5/5/5 pixel

Introducing fundamental vector data types permits the compiler to provide stronger type-checking and supports overloaded operations on vector types.