is this a bug?

Number on left is a double precision real.  Number on right is what is returned by NINT

   3210000000.00000      -2147483648
   20000000.0000000         20000000
   1027500000.00000       1027500000

I'm questioning this:

   3210000000.00000      -2147483648

And, it gives a floating invalid when assigned to what should be a 64 bit integer. (or double precision integer if you prefer)