[Wien] openmpi bugs

[Laurence Marks]

Since there were a couple of recent posts about problems with openmpi,
and I ran in to some myself over the last week (on a new cluster), I
just want to post a little information.

Unfortunately, openmpi does not export environmental variables such as
LD_LIBRARY_PATH or the changes in the default stacksize from a command
such as "ulimit -s unlimited" (in bash, slightly different form is
needed for csh). As a consequence some of the larger Wien2k codes such
as lapw0_mpi may fail unless care has been taken with defaults when
the OS was setup. For some further reading see:

http://wiki.bath.ac.uk/display/HPC/OpenMPI
http://www.open-mpi.org/community/lists/users/2008/09/6491.php
http://www.open-mpi.org/faq/?category=openfabrics#ib-locked-pages-more

The second link gives a nice example of how to prove to whoever runs
your cluster that they have a problem (if you run into issues). In my
limited experience with openmpi it is lapw0_mpi which has most issues.

-- 
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2220 N Campus Drive
Northwestern University
Evanston, IL 60208, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
Web: www.numis.northwestern.edu
Chair, Commission on Electron Crystallography of IUCR
www.numis.northwestern.edu/
Electron crystallography is the branch of science that uses electron
scattering and imaging to study the structure of matter.