From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 19340 invoked by alias); 9 Nov 2011 00:49:17 -0000 Received: (qmail 19328 invoked by uid 22791); 9 Nov 2011 00:49:16 -0000 X-SWARE-Spam-Status: No, hits=-2.8 required=5.0 tests=ALL_TRUSTED,AWL,BAYES_00 X-Spam-Check-By: sourceware.org Received: from localhost (HELO sourceware.org) (127.0.0.1) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Wed, 09 Nov 2011 00:49:03 +0000 From: "vincent-srcware at vinc17 dot net" To: glibc-bugs@sources.redhat.com Subject: [Bug math/13381] rounding problem wih sincosl Date: Wed, 09 Nov 2011 00:49:00 -0000 X-Bugzilla-Reason: CC X-Bugzilla-Type: changed X-Bugzilla-Watch-Reason: None X-Bugzilla-Product: glibc X-Bugzilla-Component: math X-Bugzilla-Keywords: X-Bugzilla-Severity: normal X-Bugzilla-Who: vincent-srcware at vinc17 dot net X-Bugzilla-Status: NEW X-Bugzilla-Priority: P2 X-Bugzilla-Assigned-To: aj at suse dot de X-Bugzilla-Target-Milestone: --- X-Bugzilla-Changed-Fields: CC Message-ID: In-Reply-To: References: X-Bugzilla-URL: http://sourceware.org/bugzilla/ Auto-Submitted: auto-generated Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Mailing-List: contact glibc-bugs-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Post: List-Help: , Sender: glibc-bugs-owner@sourceware.org X-SW-Source: 2011-11/txt/msg00016.txt.bz2 http://sourceware.org/bugzilla/show_bug.cgi?id=3D13381 Vincent Lef=C3=A8vre changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |vincent-srcware at vinc17 | |dot net --- Comment #5 from Vincent Lef=C3=A8vre 2011-11-09 00:48:59 UTC --- The AMD Opteron doesn't have this "bug". Now, there's a much more important problem than the rounding, on both Intel= and AMD processors: the fact that their implementation is broken due to the lac= k of accurate range reduction. Try with X =3D 1.0e22 and compare the results obt= ained with -O0 (hardware implementation) and -O3 (obtained with MPFR, I suppose, = thus correct). In short, the hardware implementation must not be used. If the user wants a fast implementation but incorrect results on large values, he can still use compiler options like -ffast-math, with which the compiler would generate hardware instructions directly. --=20 Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=3Demail ------- You are receiving this mail because: ------- You are on the CC list for the bug.