Dear Amber users,
I have the following problem when trying to compile Amber 6.0 on a Debian
Cluster:
With Machine.pgf77, I get the following problem:
pgf77 -o rdparm main.o rdparm.o dispatch.o help.o utility.o io.o trajectory.o
torsion.o rms.o display.o interface.o experimental.o ptraj.o actions.o
analyze.o pdb/libpdb.a -lm -lm
main.o: In function `main':
main.o(.text+0x5c): multiple definition of `main'
/usr/local/pgi/linux86/lib/pgfmain.o(.text+0x10): first defined here
/usr/bin/ld: Warning: size of symbol `main' changed from 155 to 293 in main.o
/usr/local/pgi/linux86/lib/pgfmain.o: In function `main':
/usr/local/pgi/linux86/lib/pgfmain.o(.text+0x8a): undefined reference to
`MAIN_'
make[1]: *** [rdparm] Error 1
make[1]: Leaving directory `/usr/local/amber6/src/ptraj'
make: *** [install] Error 2
There is one main.o in ptraj, one in carnal, and one in
$PGI/linux86/lib/pgfmain.o. This error occurs even if I specify the complete
name of main.o in the Makefile. Does anybody have an idea?
With Machine.g77_mpich, the compilation runs to the end, but mpirun ...
$AMBERHOME/exe/sander don't work with any number of processors, giving the
error:
/usr/bin/mpirun.mpich -machinefile /usr/spool/PBS/aux/129.icmbcu001.epfl.ch
-np 3 /usr/local/amber6/exe/sander -O -i ./md/dy300_vel.in -o ./md/dy300.out
-p ./prep/TETA_prm.top -c ./md/TETA_rel.rst -r ./md/MD300.rst -inf
./md/MD.lis -x ./md/MD300.g12 -v ./md/MD300.vel
p0_24460: p4_error: : 1
bm_list_24461: p4_error: interrupt SIGINT: 2
p0_24460: (1.298727) net_send: could not write to fd=4, errno = 32
For this I inspired from an nice answer of David Konerding, with a program to
test the mpich system. I also introduced the environment variable
P4_GLOBALMEMSIZE=1000000 to be sure there is enough memory. My MPICH system
works correctly. But amber does not work...
Still the same error. Where should I search?
1 precision: with Machine.g77, everything works fine. No problem of file
access or so...
Thanks for help. I'm lost
Fabrice
Received on Wed Jun 05 2002 - 01:31:26 PDT
