https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125169
Drea Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed| |2026-05-03
Ever confirmed|0 |1
Keywords| |needs-source
Status|UNCONFIRMED |WAITING
--- Comment #1 from Drea Pinski <pinskia at gcc dot gnu.org> ---
Can you provide a full testcase that is compilable instead of just a snipit?
As I tried:
```
#include <math.h>
#define PI2 6.28318530717958647692528676654f
float pwr[2];
float f() {
float cPwr = 0.0266058277;
pwr[1] = 0.00304872775;
float s = sqrtf(1/(cPwr * cPwr * PI2));
float s2 = 1/(cPwr * cPwr * PI2);
float res = pwr[1] - expf(-(1 * 1)/(2*s2)) / (s*sqrtf(PI2));
return res;
}
int main()
{
__builtin_printf("%f\n", f());
}
```
And it prints out -0.023498 on all versions I tried.