Initialize 64-bit integer variables with 64-bit real values

Help
5

What @fortran4r wrote works for 3d9, but I wonder how to recover the correct integer when the floating point value is near huge(i). I was going to suggest using the nint function, but the output of the program

program convert
use iso_fortran_env, only: int64
implicit none
integer, parameter :: dp = kind(1.0d0)
real(kind=dp) :: x,x0
integer(kind=int64) :: i,j,k
do j=0,512
   k = huge(i) - j
   x = real(k,kind=dp)
   if (j == 0) x0 = x
   i = x
   if (modulo(j,100_int64) == 0 .or. i > 0) print*,j,k,x0==x,x,i,nint(x,kind=int64)
end do
end program convert

is

                    0  9223372036854775807 T   9.2233720368547758E+018 -9223372036854775808 -9223372036854775808
                  100  9223372036854775707 T   9.2233720368547758E+018 -9223372036854775808 -9223372036854775808
                  200  9223372036854775607 T   9.2233720368547758E+018 -9223372036854775808 -9223372036854775808
                  300  9223372036854775507 T   9.2233720368547758E+018 -9223372036854775808 -9223372036854775808
                  400  9223372036854775407 T   9.2233720368547758E+018 -9223372036854775808 -9223372036854775808
                  500  9223372036854775307 T   9.2233720368547758E+018 -9223372036854775808 -9223372036854775808
                  512  9223372036854775295 F   9.2233720368547748E+018  9223372036854774784  9223372036854774784

so the signs of the integer variables are reversed for x equal to huge(i) but not huge(i)-512.