This is the mail archive of the
glibc-bugs@sourceware.org
mailing list for the glibc project.
[Bug math/15359] Inaccurate cos for IBM long double
- From: "carlos at redhat dot com" <sourceware-bugzilla at sourceware dot org>
- To: glibc-bugs at sourceware dot org
- Date: Fri, 12 Apr 2013 13:52:24 +0000
- Subject: [Bug math/15359] Inaccurate cos for IBM long double
- Auto-submitted: auto-generated
- References: <bug-15359-131 at http dot sourceware dot org/bugzilla/>
http://sourceware.org/bugzilla/show_bug.cgi?id=15359
Carlos O'Donell <carlos at redhat dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |WAITING
AssignedTo|carlos at redhat dot com |unassigned at sourceware
| |dot org
--- Comment #5 from Carlos O'Donell <carlos at redhat dot com> 2013-04-12 13:52:24 UTC ---
(In reply to comment #4)
> > Therefore there is nothing wrong with the test, it is now correctly showing
> > that there are 376 ulps of error in the IBM long double implementation of
> > cos(x).
>
> Guess that's the point of this report.
It is. My thought was that perhaps I'd lost some precision in the constants
used for the test, and it was the case (this is something I'd been talking
about with Rich and David). After fixing the constants there can be no doubt,
unless I made a mistake with my analysis, that cos for IBM long double is
indeed inaccurate.
Would you mind updating the ulps file and adding a comment to this test to
indicate this BZ#?
--
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.