Date: Thu, 17 Aug 2023 17:57:12 +0200 From: Szabolcs Nagy <nsz@...t70.net> To: Paul Zimmermann <Paul.Zimmermann@...ia.fr> Cc: musl@...ts.openwall.com Subject: Re: musl 1.2.4 * Paul Zimmermann <Paul.Zimmermann@...ia.fr> [2023-08-16 16:39:25 +0200]: > while updating our comparison of mathematical functions with Musl 1.2.4, > I noticed the following two issues with acoshl and exp10l in double extended > precision: > > zimmerma@...iandre:~/svn/tbd/20/src/binary80$ VERBOSE=-v ./doit.musl acosh 1000 > Checking acosh with musl-1.2.4 > Using seed 2804715 > Using 6 threads > NEW acosh 0 -1 -0x6.e2368c0ed74e5698p+16l [-nan] [inf] inf inf > libm gives -0x4.b4d6a621e8e631f8p+0l > mpfr gives nanl yes it seems acoshl(x) does not handle x<0 (and can go wrong for x in [-2,-0x1p32]) > > zimmerma@...iandre:~/svn/tbd/20/src/binary80$ VERBOSE=-v ./doit.musl exp10 1000 > Checking exp10 with musl-1.2.4 > Using seed 2807610 > Using 6 threads > exp10 0 -1 0x2.68826a13ef3fde64p+16376l [-nan] [inf] inf inf > libm gives nanl > mpfr gives infl this seems to be a bug in powl: the reducl function can overflow. thanks for the reports.
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.