What's the funda about de-normalization?
Like when to apply, when not to? What do we exactly infer from 'number in the denormal form'?
Ex: +6.25 stored as 0 1001 10010 (S BE M)
While reporting value, and retrieving data part, we denormalize as -
.10010 x 2*2 -> .11001 x 2*3 (discarding the last bit of M)
Consider calculation of range -
Mantissa max - (consider IEEE STD)
All 1's -> 111...1 (23 times)
It is read as 1.111...1 (23 times) -> 11...1 (24 times) x 2*-23
This way we claim max value Mantissa as (2 - 2*-23) & claim range as 1 to (2 - 2*-23).
Why don't we apply de-normalization to this case while reporting the range?