Hi All,
I know this is low-priority right now but I hit a small snag in compiling
sander.RISM.MPI under cygwin. It appears the problem is in compiling
FFTW-2.1.5:
make[2]: Entering directory `/home/Dan/Amber/CVS/amber11/src/fftw-2.1.5'
Making install in fftw
...
mkdir .libs
libtool: link: warning: undefined symbols not allowed in i686-pc-cygwin
shared libraries [This message occurs several times]
...
mpicc -DCYGWIN -DBINTRAJ test_sched.c -o test_sched
/tmp/ccBMABYM.o:test_sched.c:(.text+0x104): undefined reference to
`_make_comm_schedule'
/tmp/ccBMABYM.o:test_sched.c:(.text+0x153): undefined reference to
`_check_comm_schedule'
/tmp/ccBMABYM.o:test_sched.c:(.text+0x18f): undefined reference to
`_print_comm_schedule'
/tmp/ccBMABYM.o:test_sched.c:(.text+0x1c5): undefined reference to
`_sort_comm_schedule'
...
and ultimately the compile fails. So it seems that FFTW-2.1.5 does something
with libraries that cygwin doesn't like. A web search on this is pretty
inconclusive - it seems other people have had similar issues but no one
really know what the deal is. However on the FFTW under their release notes
for FFTW 3 (
http://www.fftw.org/release-notes.html) it seems that certain
things under cygwin have been fixed. Since we already seem to have
FFTW-3.2.2 in the /src directory, would it be a simple thing to tell RISM to
use this instead? Just out of curiosity why do we have 2 versions of FFTW
anyway?
Like I said, not a big deal if it doesn't get sorted out - I'd be willing to
bet I'm one of very few who will try to compile parallel under cygwin...
-Dan
--
-------------------------
Daniel R. Roe
Postdoctoral Associate
SAS - Chemistry & Chemical Biology
610 Taylor Road
Piscataway, NJ 08854
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Mar 26 2010 - 07:30:02 PDT
