Floating-Point
So how can we represent a calculation like:
34,200,000,000 ? 0.0000000762
with 16 bit numbers?
Rewrite in scientific notation: (342 ? 10^8) ? (7.62 ? 10^-8)
Floating point is nothing more than scientific notationNote that both magnitude and precision are important aspects of scaling.
Where f and e are both signed, fixed-point numbers and r is the radix (base). The fractional and exponent parts of floating-point numbers are sometimes called the mantissa and characteristic, respectively.