https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94586
John David Anglin <danglin at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Attachment #48295|0 |1
is obsolete| |
--- Comment #26 from John David Anglin <danglin at gcc dot gnu.org> ---
Created attachment 48327
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=48327&action=edit
Patch to fix float128 node selection on hpux
This is my latest patch to fix the float128. In this iteration, I have
attempted
to create a target hook to allow REAL16 to be float128 instead of long double.
Previous version had a problem with iroundq. Need to use long double builtins.
This version is untested. With previous version, there was one new test fail
(probably it didn't run before) and one XPASS. So, using libquadmath is
working out quite well.
Have no objection to having configure checks for fmaf and fma. That's better
than my POSIX check.
Sorry, this is going slowly but testing takes awhile.
