[mesa-users] Additional chi2 variables in astero
paxton at kitp.ucsb.edu
Mon Feb 24 13:17:02 EST 2014
I only checked 5995 on my mac with gfortran since it wasn't a public release, so perhaps there is a new bit-for-bit problem showing up. I'll check before doing a public release, but in the meantime, you might update to 6019 and try that. It just passed the full test suite last night.
On Feb 23, 2014, at 11:11 PM, Warrick Ball wrote:
> I just did a fresh checkout of revision 5995 and tried to compile with the latest version of the SDK (December 23, 2013) on openSUSE 12.2. I ran into a failed check when creating the nuclear reaction rates:
> write ../../data/rates_data/cache/rni57na_aux_2.bin
> < o16 6.7982636015615083D-01
> < mg24 9.3550987171147307D-02
> < si28 1.9759879793152474D-01
> < s32 1.8174827115480761D-02
>> o16 6.7982636013724751D-01
>> mg24 9.3550987149650516D-02
>> si28 1.9759879792773372D-01
>> s32 1.8174827118491894D-02
> TEST FAILED -- compare test_output to tmp.txt
> ./build_and_test FAILED
> I realise this isn't a supported release but I thought I'd let you (and everyone else?) know anyway. The differences are very small, so I guess this might be a hangover of the (phenomenal!) work you did on getting bit-for-bit reproduceability?
> I'll compile again and just let it warn me about these errors, rather than halting the compilation. They don't look big enough to be serious issues.
> Warrick Ball
> Postdoc, Institut für Astrophysik Göttingen
> wball at astro.physik.uni-goettingen.de
> +49 (0) 551 39 5069
More information about the Mesa-users