Hi All,
Recently when testing the Amber RCs I've come across this kind of
error during RISM parallel tests; they all fail with:
mdfil: Error unknown flag: -xvv
Looking into this I found that this happens if a configure that
enables RISM by default (e.g. 'configure intel') occurs after a
configure that disables it by default (e.g. 'configure -mpi intel').
Because these tests depend on TESTRISMSANDER being properly set in
config.h, if the value in config.h doesn't match the binary, the tests
can be run when they shouldn't be. Having the tests depend on
variables in the config.h file seems like bad practice anyway because
config.h can be so volatile. I think it would be better if the tests
could somehow be modified so they detect whether RISM is available
(e.g. see if the -xvv flag is supported). Something that could be
sourced from test run scripts that will skip the test cleanly like:
touch foo
HAS_RISM=`$TESTsander -O -xvv foo 2> /dev/null | grep flag`
/bin/rm foo mdout
if [ -n "$HAS_RISM" ] ; then
echo "$TESTsander compiled without RISM support."
exit 0
fi
In the long run it may be worthwhile to add something like the cpptraj
'--defines' flag to sander/pmemd for easier inquiry of what
capabilities are supported by a given binary.
Of course for this release it might just be too much work for too
little gain, but maybe something to keep in mind for the future.
Thoughts?
-Dan
--
-------------------------
Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 307
Salt Lake City, UT 84112-5820
http://home.chpc.utah.edu/~cheatham/
(801) 587-9652
(801) 585-6208 (Fax)
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Apr 07 2015 - 14:30:03 PDT
