Why do we store the exponent in a biased form in a floating point representation?
Storing the exponent in bias form means that we add a constant number to the actual value of the exponent, before storing it. For example, for an 8-bit exponent, we can use a bias of 128, which means that instead of storing the numbers from -128 to 127, in two’s complement, we store them as 0 to 255 respectively. By doing so, we make sure that the stored exponent is non-negative. By avoiding negative numbers for the exponent, it becomes easier to check for negative overflow — checking for overflow with two’s complement numbers is slightly more expensive.
