This is the mail archive of the
glibc-bugs@sourceware.org
mailing list for the glibc project.
[Bug math/20033] [x86_64] vectorized math function don't call the __finite versions
- From: "jtaylor.debian at googlemail dot com" <sourceware-bugzilla at sourceware dot org>
- To: glibc-bugs at sourceware dot org
- Date: Thu, 28 Jul 2016 13:36:44 +0000
- Subject: [Bug math/20033] [x86_64] vectorized math function don't call the __finite versions
- Auto-submitted: auto-generated
- References: <bug-20033-131@http.sourceware.org/bugzilla/>
https://sourceware.org/bugzilla/show_bug.cgi?id=20033
--- Comment #7 from Julian Taylor <jtaylor.debian at googlemail dot com> ---
oh its a glibc feature? So one could do the aliasing to direct to _ZGV*_finite
under ffast-math but leave it intact and svml compatible if that is not set?
I think it severely limits the usability of these functions when they can only
be used with ffast-math, that is a catch all optimization flag unsuitable for a
lot of applications.
Slight numerical differences are ok, but some of the other stuff this flag
enables not (like the change of non-finites, poor complex division etc.)
--
You are receiving this mail because:
You are on the CC list for the bug.