previous | start | next

Normalized and Denormalized Floating Point Values

Except for the special values, the bit fields: s, e, and f are used to determine the sign, the exponent E, the fraction frac and the significand M, respectively, of any other floating point value. The value of a non-special value is calculated as follows:

The frac is needed to determine M.

The fraction frac and significand M are calculated:

      frac = f / 223  

      (this is for 32 bit floats; for 64 bit doubles, divide by 252)
   

Then the float value V corresponding to these bits is:

      V = (-1)s 2E M
   


previous | start | next