Half
Represents a floating point value of 16 bits. Also known as Half-precision floating-point format (IEEE 754-2008). This is an inline class backed by a UShort. No operations defined for this class, you should convert from and into other values. To get its internal representation, call the toRawBits function.
Mantissa/fraction: 10 bits Exponent: 5 bits Sign: 1 bit
Formula: (2**(exp)) * (1 + mantissa/1024)
Significance: stored 10 bits (but can achieve 11 bits of integral precision)