Hi developers,
I tried to compile a fresh (as of this morning) checkout of Amber11 on my
desktop machine. Several problems occured... Serial compilation configured
with
./configure -nopar -nobintraj intel
works after I remove the -DBINTRAJ flags in config.h at:
FPP=cpp -traditional -P -DNO_SANDER_DIVCON -DBINTRAJ
FPPFLAGS=-P -DNO_SANDER_DIVCON -DBINTRAJ
is there any reason those are still set when nobintraj is specified?
When I afterwards try a parallel compile with
/configure -mpi -nobintraj intel && make -f Makefile_at clean && make -f
Makefile_at parallel
it dies at the ptraj step with this message:
make[2]: Leaving directory
`/home/steinbrt/programs/amber11_new/amber11/src/f2c'
mpicc -c -DMPI -o main.o main.c
mpicc -c -DMPI -o rdparm.o rdparm.c
mpicc -c -DMPI -o dispatch.o dispatch.c
mpicc -c -DMPI -o help.o help.c
mpicc -c -DMPI -o utility.o utility.c
mpicc -c -DMPI -o second.o second.c
mpicc -c -DMPI -o io.o io.c
mpicc -c -DMPI -o trajectory.o trajectory.c
trajectory.c: In function ‘dumpAmberTrajectory’:
trajectory.c:1390: error: ‘MPI_File’ undeclared (first use in this function)
trajectory.c:1390: error: (Each undeclared identifier is reported only once
trajectory.c:1390: error: for each function it appears in.)
trajectory.c:1390: error: expected expression before ‘)’ token
mpicc: No such file or directory
make[1]: *** [trajectory.o] Error 1
make[1]: Leaving directory
`/home/steinbrt/programs/amber11_new/amber11/src/ptraj'
make: *** [parallel] Error 2
any suggestion what that means/where to look? The additional (superfluous)
-DBINTRAJ flags are present in config.h again, but removing them gives the
same error for a parallel compile of AT. If I comment out ptraj in
Makefile_at and finish the AT compile, a subsequent compilation of
sander.MPI afterwards dies as well with this:
mv f2c.a /home/steinbrt/programs/amber11_new/amber11/lib
make[1]: Leaving directory
`/home/steinbrt/programs/amber11_new/amber11/src/f2c'
mpif77 -o sander.MPI constants.o memory_module.o stack.o nose_hoover.o
[many many object files...]
hooks.o ncsu-umbrella.o ncsu-utils.o ncsu-value.o
/opt/intel/Compiler/11.0/083/lib/ia32/libifcoremt.a(for_mi_int.o): In
function `i_dnnt':
for_mi_int.c:(.text+0x100): multiple definition of `i_dnnt'
/home/steinbrt/programs/amber11_new/amber11/lib/f2c.a(i_dnnt.o):i_dnnt.c:(.text+0x0):
first defined here
mpif77: No such file or directory
make: *** [sander.MPI] Error 1
apparently i_dnnt is defined in ambers f2c.a as well as in my compiler
includes. Is this a problem with intels fortran compiler? Anything else I
doing wrong here?
Any comments/suggestions are appreciated,
Kind Regards,
Thomas
Dr. Thomas Steinbrecher
BioMaps Institute
Rutgers University
610 Taylor Rd.
Piscataway, NJ 08854
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Sep 15 2009 - 07:30:02 PDT
