fdim(x, y) is not folded at the compilation time when (x, y) is constant

[k-arrows]

Consider the following example.
https://godbolt.org/z/15Ef6W1Es

#include <cmath>

double f1() { return std::fdim(-1.0, 0.0); }

double f2() { return std::fdim(0.0, 0.0); }

double f3() { return std::fdim(1.0, 0.0); }

[braw-lee]

i can try this