Floating-Point
Fixed point notation is useful for representing small-valued integers and scaled fractional numbers.
Floating-point processors handle the scaling of the mantissa automatically
- Additional hardware for Floating-Point is relatively complex and adds to the cost of the computer
- But the operation of the machine is more efficient and we can put enough transistors on a single piece of silicon that the incremental cost is reasonable.
Consider the dynamic range of values that can be expressed by a 16-bit fixed-point number. When interpreted as an integer, the value range is:
with the high-order bit representing the sign. Maximum negative value is: