Cubic uses long double which is not standardized. On x86-64 it provides only 80 bits of precision, but on other architectures it provides either 64 bits or 128 bits of precision. This not only mean...