floating point multiplier

[shan14]

Hi

In floating point multiplication,
my significant is (27 downto 0) -----(it is aligned and padded with zeros)

SIGNIFICANT: out std_logic_vector (27 downto 0);
B_sig : in std_logic_vector (31 downto 0); 
A_sig : in std_logic_vector (31 downto 0); 

signal prod:  std_logic_vector (63 downto 0);

dtemp := (unsigned(A_SIG) * unsigned(B_SIG));
 
SIGNIFICANT<= prod(47 downto 20);

why prod is taken (47 downto 20) and not any other combination like (57 downto 30) etc

 

[std_match]

The input numbers are probably padded so only (23 downto 0) contains information, but I can't explain why someone would design it like that.