In some situations, you might want to treat a vector as having a different type, without changing its value. A set of intrinsics is provided to perform this type of conversion.
Syntax
vreinterpret{q}_dsttype_srctype
Where:
qSpecifies that the conversion operates on 128-bit vectors. If it is not present, the conversion operates on 64-bit vectors.
dsttypeRepresents the type to convert to.
srctypeRepresents the type being converted.
The following intrinsic reinterprets a vector of four signed 16-bit integers as a vector of four unsigned integers:
uint16x4_t vreinterpret_u16_s16(int16x4_t a);
The following intrinsic reinterprets a vector of four 32-bit floating point values integers as a vector of four signed integers.
int8x16_t vreinterpretq_s8_f32(float32x4_t a);
These conversions do not change the bit pattern represented by the vector.
PDF version