AMBER Archive (2003)

Subject: Script and MPICH

From: FyD (fyd_at_u-picardie.fr)
Date: Fri Mar 21 2003 - 08:01:40 CST

Dear All,
  
I use a script (see below) in order to follow each step of a multi-step simulation (AMBER7
under pc-linux cluster). If I run a non-parallel version of sander (i.e. without mpirun
calls), it works fine i.e. the terminal prints each line of the script following the order
of the simulation commands (as it should do...).
  
On the contray, if I ran 'sander' in parallel (i.e. with mpirun calls), only the 4 '1st
echo' and the first sander calculation are printed in the terminal and after the printing
is blocked but the calculations work fine (i.e. md2, md3, md4 continue...)
  
Any idea what is the problem ?
  
Thanks, regards, Francois
 
  
echo " igb=2 run "
echo " First minimization "
echo " position restraints: 5 kcal/mol/A^2, 9 A cutoff "
echo " "
/usr/pgi/linux86/bin/mpirun -np 2 -machinefile .machinefile7 /home/fyd/amber7/exe/sander
-O \
                -i min.in \
                -p prmtop \
                -c prmcrd \
              -ref prmcrd \
                -o min.out \
              -inf min.mdinfo \
                -r min.rst
  
echo " md1: slow warming up 0-> 298 K: "
echo " position restraints: 5 kcal/mol/A^2, 9 A cutoff "
echo " "
/usr/pgi/linux86/bin/mpirun -np 2 -machinefile .machinefile7 /home/fyd/amber7/exe/sander
-O \
                -i md1.in \
                -p prmtop \
                -c min.rst \
              -ref min.rst \
                -o md1.out \
                -x md1.mdcrd \
              -inf md1.mdinfo \
                -r md1.rst
  
 
echo " md2: equilibration 1 "
....
echo " md3: equilibration 2 "
....
echo " md4: productive MD "
....