[PATCH] add inputs to auto-libm-test-in yielding larger errors
Paul Zimmermann
Paul.Zimmermann@inria.fr
Thu Nov 26 08:16:48 GMT 2020
> Date: Wed, 25 Nov 2020 22:05:20 +0000
> From: Joseph Myers <joseph@codesourcery.com>
>
> On Wed, 25 Nov 2020, Paul Zimmermann wrote:
>
> > Hi,
> >
> > the following patch adds inputs to auto-libm-test-in that yield larger
> > errors for double precision (binary64) on x86_64. In particular for exp
> > and log the maximal ulp error increases from 0 to 1.
>
> OK.
>
> --
> Joseph S. Myers
> joseph@codesourcery.com
thank you Joseph, I will commit that patch.
I also found a case where lgamma() yields more than 10 ulps of error:
Checking lgamma
x=-0x1.f60c969a239f2p+1 d=1.0410549778851831e+01
This should be considered as a bug and entered into bugzilla?
Paul
More information about the Libc-alpha
mailing list