This is the mail archive of the
libc-hacker@sourceware.cygnus.com
mailing list for the glibc project.
Re: Failing tests in glibc
Kars de Jong <jongk@cs.utwente.nl> writes:
|> On Wed, Nov 18, 1998 at 11:05:22AM +0100, Andreas Schwab wrote:
|> > Kars de Jong <jongk@cs.utwente.nl> writes:
|> >
|> > |> >Submitter-Id: net
|> > |> >Originator: Kars de Jong
|> > |> >Organization:
|> > |> >Confidential: no
|> > |> >Synopsis: glibc fails math tests
|> > |> >Severity: serious
|> > |> >Priority: medium
|> > |> >Category: libc
|> > |> >Class: sw-bug
|> > |> >Release: libc-2.0.102
|> > |> >Environment:
|> > |> Host type: m68k-unknown-linux-gnu
|> > |> System: Linux rincewind 2.1.127 #2 Mon Nov 16 21:51:19 CET 1998 m68k unknown
|> > |> Architecture: m68k
|> > |> CPU: 68060 with FPU and MMU
|> >
|> > I cannot reproduce the problems on my 68030/68882. It may be a bug in the
|> > FPU emulation for the 68060. Could you please run the math tests
|> > (test-{float,double,ldouble}) manually with -v2, so that we can see how
|> > much the computed values differ from the expected ones?
|>
|> Okay, here comes the output:
Thanks. The differences look small enough to be ignored as round-off
errors. I'll update the epsilons in the tests. It seems like the fpu
emulation isn't very accurate in general.
--
Andreas Schwab "And now for something
schwab@issan.cs.uni-dortmund.de completely different"
schwab@gnu.org