This is the mail archive of the
mailing list for the glibc project.
Re: Fix libm-test.inc ulps calculation for subnormals (bug 14064)
On Sun, 6 May 2012, Andreas Jaeger wrote:
> On 05/06/2012 02:44 PM, Joseph S. Myers wrote:
> > When fixing problems with exp10 I noted a problem, bug 14064, with how
> > libm-test.inc calculates ulps values for subnormal expected results,
> > so causing last-bit errors to be wrongly reported as huge numbers of
> > ulps.
> > I propose this patch to fix this libm-test.inc bug. Tested x86_64,
> > x86 and (together with my exp10 patch
> > <http://sourceware.org/ml/libc-alpha/2012-05/msg00201.html>, where the
> > error is now reported as 1ulp not 35184372088832ulp) powerpc.
> Does this fix any other values?
I didn't spot expected errors for subnormal values in a glance over some
libm-test-ulps files. There aren't many tests expecting subnormal results
(this particular one is only subnormal for IBM long double because of how
that type has a normal range narrower than the normal range for double).
Joseph S. Myers