[petsc-dev] testing of nvcc incomplete?

[Matthew Knepley]

On Sat, Oct 1, 2011 at 10:34 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:

>
>   Matt,
>
>    It looks like the tests of nvcc are incomplete? That is it checks that
> the nvcc compiler works but does not check if the the resulting .o can be
> linked successfully. That is it never checks linking of a nvcc compiled
> thing??[see attached file: configure.log]
>
>    I think this is wrong. See the attached configure.log it claims that
> nvcc works but then conks out when checking the cuda version (because it
> cannot link a nvcc program). This is the wrong place to conk out and
> misleading message to user. It should conk when checking linking of nvcc
> program with a useful error message about not being able to link.
>

http://petsc.cs.iit.edu/petsc/BuildSystem/annotate/7fc6b5eafc08/config/setCompilers.py#l415

   Matt


>    Thanks
>
>     Barry
>
>
>


-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20120423/2f95b7bf/attachment.html>