4 ms·
If your machine uses 80-bit floating point, then those extended-precision floats can exactly represent any 64-bit integer, provided that your language's standar
by wtallis 12y ago
If your machine uses 80-bit floating point, then those extended-precision floats can exactly represent any 64-bit integer, provided that your language's standard library/runtime correctly converts without an intermediate step using 64-bit floats (53-bit mantissa).
- jevinskie 12y agoSometimes called 'long double' by the compiler.
- saitohm 12y agoYes. You can see the compatibility between 80-bit floating points (long double in C) and 64-bit integer (long long int) with a code like this: long double x0,x1,y0,y1; x0 = 18446744073709551615.0L; /* 2^64 - 1 */ x1 = 18446744073709551616.0L; /* 2^64 */ y0 = x0 + 1.0L; y1 = x1 + 1.0L; printf("%d\n", x0 == y0); /* false (with enough precision) */ printf("%d\n", x1 == y1); /* true (precision being lost ) */