You apparently have a perl binary that only supports 32-bit integers.  And the value you compute (2348273369088) is much larger than 32 bits can hold. Represented as binary, it is:

   10_00100010_11000000_00000000_00000000_00000000
[download]

So, it gets truncated

               11000000_00000000_00000000_00000000
[download]

and as the leftmost (most significant) bit happens to be 1, and you used %d (= signed int) in printf, it is interpreted as a negative value (see Two's Complement). Effectively the same as

printf "%d", 0b11000000_00000000_00000000_00000000 - 2**32;    # -1073
+741824
[download]

with a perl that supports ints larger than 32 bit.