How can I fill values greater than FFFFFFFFh or
4294967295 in registers?
It is possible to put two 32bit registers together?

No you cant, 32Bit = 0ffffffffh different combonations, meaning you cant get greater values then that using registers.
However, if you use floating point numbers, for a 32Bit (called a REAL4), you get a much larger range due to special ordering of the bits etc. You can extend the range and acuracy with a REAL8 or a REAL10 varibles (64Bit and 80Bit Respecitivy).
There is a links to good floating point tutorials on the message board.

You could write some code for handling large numbers yourself.
However, this becomes sorta tricky (and tedious) if you want
to do more than addition and subtraction. Multiplication and
division is definitely not fun.
If you can live with using other people's code (which might not
be the fastest out there), try looking for miracl, freelip and
crypto++ - afaik these all have "large number routines".