This is the mail archive of the
glibc-bugs@sourceware.org
mailing list for the glibc project.
[Bug math/13957] powl very inaccurate on powerpc
- From: "jsm28 at gcc dot gnu.org" <sourceware-bugzilla at sourceware dot org>
- To: glibc-bugs at sources dot redhat dot com
- Date: Tue, 06 Nov 2012 16:38:42 +0000
- Subject: [Bug math/13957] powl very inaccurate on powerpc
- Auto-submitted: auto-generated
- References: <bug-13957-131@http.sourceware.org/bugzilla/>
http://sourceware.org/bugzilla/show_bug.cgi?id=13957
Joseph Myers <jsm28 at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
Resolution| |INVALID
--- Comment #1 from Joseph Myers <jsm28 at gcc dot gnu.org> 2012-11-06 16:38:42 UTC ---
Actually, I don't think this case demonstrates a significant error, although
maybe other cases do.
You give an expected result of
2.49114074548890053222918160895394492812126078716433839487765876e-301. That
has binary exponent -999. The least normal binary exponent for IBM long double
is -969. So there are only 76 mantissa bits available for this number - that
is, 22 correct decimal digits after the point is the most you could expect.
Then you raise a number, with an error of about one part in 2^75 from the
mathematical value of x^y, to the power 1/13499, resulting in a number expected
to have an error of about one part in 2^89. That corresponds pretty well to
the 27 digits after the point.
--
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.