next up previous
Next: Decimal to Floating Point Up: Data Representation Previous: Example: Character to Numeric

Floating Point

Similar to scientific notation, floating point representation consists of a mantissa, an exponent and a sign bit.

The value of a floating point number, N, with mantissa, m, exponent, e, and sign bit, s, is given by:

displaymath1409

The mantissa is normalized with the binary point to the right of the most significant bit so that

displaymath1410

An n bit mantissa has the general form:

displaymath1411

In a normalized mantissa, the bit to the left of the binary point, tex2html_wrap_inline1415 , is always 1. Thus, tex2html_wrap_inline1415 is omitted from the floating point representation and is called the hidden bit.

Note: Using the hidden bit convention, there is no floating point representation for zero.





Mitch Roth
Wed Oct 9 13:38:30 ADT 1996