overflow and underflow in floating point representation