How to determine the range of maximum floating point numbers?

Question

Electronics

How to determine the range of maximum floating point numbers, and determine how many times the maximum numbers that can be manipulated by processors with 82 bit registers and 64 bits differ?
I compare the architecture data of IA-32 and IA-64
The first uses the standard maximum of 64 bits (double or double precision).
In the second, the FR register format includes a 64-bit mantissa, a 17-bit exponent and 1 bit under the number sign. That is a total of 82 bits.
From here I conclude:
(2^81)/(2^63)=262144
That is, the maximum number of 82 bits is greater than the maximum number of 64 bits by 262144 times, but this is not a correct conclusion.
Where is the mistake?

Answer 1

64 bits - 2 ^{(2 10 -1)} ×(1 + (1 - 2 ⁻⁵² )) = 1.797693135 × 10 ³⁰⁸
82 bits - 2 ^{(2 16 -1)} ×(1 + (1 - 2 ⁻⁶⁴ ) ) = 2.00352993×10 ¹⁹⁷²⁸
2.00352993×10 ¹⁹⁷²⁸ / 1.797693135×10 ³⁰⁸ = 1.114500517×10 ¹⁹⁴²⁰