https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118580
--- Comment #8 from anlauf at gcc dot gnu.org ---
It appears that the issue is solved with r15-7165-g3cef53a4d4ff44 .
Can you please check?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118580
--- Comment #7 from federico ---
Thanks Harald and Steve, great hints. Looking at the `-fdump-tree-original`, I
see:
- the on-the-fly expression seems the difference between a complex(8) and a
complex(4) expressions:
```
__builtin_cabs ( COMP
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118580
--- Comment #6 from anlauf at gcc dot gnu.org ---
The issue can be reduced to the case k=1, i.e. 1x1 matrices.
I then get:
maxval(abs(A - eye)) = -Infinity
Looks like the initial condition of the maximum search plays a role.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118580
--- Comment #5 from kargls at comcast dot net ---
(In reply to federico from comment #3)
> Sorry, I introduced a typo while editing the post.
> Here is a slightly shorter version of the sample:
>
Given your track record in fortran-lang, I ass
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118580
--- Comment #4 from anlauf at gcc dot gnu.org ---
Adding -ffinite-math-only to the compile flag seems to "fix" the problem.
This might give a hint.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118580
--- Comment #3 from federico ---
Sorry, I introduced a typo while editing the post.
Here is a slightly shorter version of the sample:
```
program complex_eye
implicit none
integer, parameter :: k = 2
integer :: j
complex, dimension(k,
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118580
kargls at comcast dot net changed:
What|Removed |Added
CC||kargls at comcast dot net
--
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118580
Richard Biener changed:
What|Removed |Added
Keywords||wrong-code
Status|UNCONFIR
