This is the mail archive of the
glibc-bugs@sourceware.org
mailing list for the glibc project.
[Bug math/22227] expm1f gives inaccurate results
- From: "joseph at codesourcery dot com" <sourceware-bugzilla at sourceware dot org>
- To: glibc-bugs at sourceware dot org
- Date: Thu, 28 Sep 2017 17:14:38 +0000
- Subject: [Bug math/22227] expm1f gives inaccurate results
- Auto-submitted: auto-generated
- References: <bug-22227-131@http.sourceware.org/bugzilla/>
https://sourceware.org/bugzilla/show_bug.cgi?id=22227
--- Comment #1 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
Having a result that is not correctly rounded is not a bug (please see the
documented accuracy goals in the glibc manual), and tests involving random
inputs are not suitable for bug reports.
If there is a particular input for which expm1f produces a result with an
error of more than a few ulps, please specify the exact input (as a hex
float constant) that gives the largest error you observed, stating the
expected result and the actual result you observe, as hex floats.
If the error is more than 1ulp from the correctly rounded result, that
input could then be added to auto-libm-test-in so the libm-test-ulps files
become more accurate. It would need to be several ulp to consider there
to possibly be a bug.
--
You are receiving this mail because:
You are on the CC list for the bug.