Just remember that 2^(10*x) is approximately 10^(3*x) - you're probably already used to this with kilobytes/kibibytes etc. That is:
2^10 = 1024 ~= one thousand
2^20 = 1024^2 = 1048576 ~= one million
2^30 = 1024^3 = 1073741824 ~= one billion
Since an int uses 31 bits (+ ~1 bit for the sign), just double 2^30 to get approximately 2 billion. For an unsigned int using 32 bits, double again for 4 billion. The error factor gets higher the larger you go of course, but you don't need the exact value memorised (If you need it, you should be using a pre-defined constant for it anyway). The approximate value is good enough for noticing when something might be a dangerously close to overflowing.
