> On Nov 28, 2014, at 10:08 AM, Dave Love <d.l...@liverpool.ac.uk> wrote: > > Gilles Gouaillardet <gilles.gouaillar...@gmail.com> writes: > >> It could be because configure did not find the knem headers and hence knem >> is not supported and hence this mca parameter is read-only > > Yes, in that case (though knem was meant to be used and it's annoying > that configure doesn't abort if it doesn't find something you've > explicitly asked for, and I didn't immediately need it).
FWIW: I dug into this and found why we weren’t aborting if knem was specified and not found. This obviously isn’t the specified OMPI behavior, so I fixed it in the trunk and will port it back to v1.8.4 > However, I got > the same for at least mpi_abort_print_stack with that parameter set. I’m not entirely familiar with the thinking behind it, but it appears that some MCA params are intended solely for reporting purposes and are therefore not really “settable”. The “have_knem_support” is one such example, and it sounds like you may have found some others. I’ll ask around and see why we did that as I can understand how it would be confusing. > > This didn't happen with OMPI 1.6 and there's no obvious way to turn it > off. > > _______________________________________________ > users mailing list > us...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/users > Link to this post: > http://www.open-mpi.org/community/lists/users/2014/11/25887.php
