Re: kiss-fft's builds failure on rawhide after upgrading of gcc and glibc
Il giorno mar, 21/04/2020 alle 21.27 +0200, Guido Aulisi ha scritto: > Il giorno mar, 21/04/2020 alle 11.19 -0700, Samuel Sieb ha scritto: > > On 4/21/20 2:30 AM, Guido Aulisi wrote: > > > I'm trying to understand why kiss-fft (an FFT library) is failing > > > in > > > rawhide only on aarch64. > > > The failing test does some fft calculation and compares them with > > > the > > > same done with numpy. > > > For waht I understood the upgrade of gcc and/or glibc makes this > > > test > > > fail only on aarch64, and I cannot undestand why gcc or glibc > > > changed > > > the maths or the precision used on this arch. > > > > Can you get the output to see what's happening in the failing > > test? > > That might help to figure out what's going wrong. > > Yes, of course, sorry I forgot to link the Koschei failed build: > > https://koschei.fedoraproject.org/package/kiss-fft?collection=f33 > > And this is the build.log: > https://kojipkgs.fedoraproject.org/work/tasks/8318/43588318/build.log > With the latest update of gcc and glibc kiss-fft's builds are back to normal now, thanks gcc folks! Ciao Guido ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
Re: kiss-fft's builds failure on rawhide after upgrading of gcc and glibc
Il giorno mar, 21/04/2020 alle 11.19 -0700, Samuel Sieb ha scritto: > On 4/21/20 2:30 AM, Guido Aulisi wrote: > > I'm trying to understand why kiss-fft (an FFT library) is failing > > in > > rawhide only on aarch64. > > The failing test does some fft calculation and compares them with > > the > > same done with numpy. > > For waht I understood the upgrade of gcc and/or glibc makes this > > test > > fail only on aarch64, and I cannot undestand why gcc or glibc > > changed > > the maths or the precision used on this arch. > > Can you get the output to see what's happening in the failing test? > That might help to figure out what's going wrong. Yes, of course, sorry I forgot to link the Koschei failed build: https://koschei.fedoraproject.org/package/kiss-fft?collection=f33 And this is the build.log: https://kojipkgs.fedoraproject.org/work/tasks/8318/43588318/build.log Thanks Ciao Guido FAS: tartina signature.asc Description: This is a digitally signed message part ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
Re: kiss-fft's builds failure on rawhide after upgrading of gcc and glibc
On 4/21/20 2:30 AM, Guido Aulisi wrote: I'm trying to understand why kiss-fft (an FFT library) is failing in rawhide only on aarch64. The failing test does some fft calculation and compares them with the same done with numpy. For waht I understood the upgrade of gcc and/or glibc makes this test fail only on aarch64, and I cannot undestand why gcc or glibc changed the maths or the precision used on this arch. Can you get the output to see what's happening in the failing test? That might help to figure out what's going wrong. ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
kiss-fft's builds failure on rawhide after upgrading of gcc and glibc
Hello, I'm trying to understand why kiss-fft (an FFT library) is failing in rawhide only on aarch64. The failing test does some fft calculation and compares them with the same done with numpy. For waht I understood the upgrade of gcc and/or glibc makes this test fail only on aarch64, and I cannot undestand why gcc or glibc changed the maths or the precision used on this arch. I could disable tests, but it's not a great solution. Can any gcc expert tell what changed between the release about math calculation and maybe how to fix them or are these tests only a waste of time and they should be disbled? Thank you very much for any help. Ciao Guido FAS: tartina signature.asc Description: This is a digitally signed message part ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
