Wait, What does it Print out?

The primitive type float is a single-precision floating-point format . According to the IEEE 754 standard, the mantissa holds 24 significant bits. However, 1234567890 in binary is 1001001100101100000001011010010 , too long (31 bits), and thus must be rounded into a scientific format-like $1.00100110010110000000110_2 \cdot 2^{30}$ . When cast back to an int , the above truncation gives 1234567936 , and thus (int)f - i = 1234567936 - 1234567890 which is $\boxed{46}$ .

The number is too big to store in a float primitive type. Some rounding occurs (this can be figured out by converting 1234567890 into binary and truncating some digits I think) and the end result is $46$ .

Shared to me by @Zong Zhang .