The calculated number is indeed 56.99999999999999289457264239899814128875732421875 which is not same to 57 (IEEE 754 binary32 or binary64 can represent this integer exactly). You can verify this with the following:

    printf("%f\n", 57.0 / 100.0 * 100.0 - 57.0);

It just happens that `%f` in C defaults to 6 fractional digits.