[gmx-users] Re: gromacs 4.0 parallelization
Justin A. Lemkul
jalemkul at vt.edu
Mon Nov 9 03:50:12 CET 2009
Justin A. Lemkul wrote:
> milad ekramnia wrote:
>> Dear gromacs users I am working with gromacs 4.0 and my system is a
>> protein under tension solvated in water under Lincs constraint .when I
>> run my mdrun over n nodes by mpirun I'll get n-1 trajectory files
>> (also the same happen for all other output files ) which as it appears
>> all are identical.
>> as far as I know each file shall represent a separate part of the box
>> but when I analyze the trrs the results are identical ( e.g for
>> g_gyrate task ) I want to now if they are exactly the same ? if yes ,
>> can I prevent the rest of them from being biult up ?
>> my mdrun command line is : mpirun -n 4 mdrun_mpi -v -s topol.tpr
> Is -n the appropriate option? Some MPI implementations are different
> (-np, -N, etc), and/or you may need to supply a hostfile.
>> and the grommp command is: grompp_mpi -v -c minimized.gro -p
>> topol.top -f pull.mdp -n index.ndx
>> I also added --enabled-mpi to my mdrun command line but it didn't work .
> If you did "--enabled-mpi" then that is incorrect, and you probably
> didn't actually compile with MPI support. The proper configuration
> option is "--enable-mpi" so you should check that you did not make a
> typo when installing Gromacs.
And additionally, I should read better before replying. The --enable-mpi option
is not an mpirun or mdrun option. It must be invoked when configuring Gromacs.
Please be sure to read the instructions here:
Justin A. Lemkul
ICTAS Doctoral Scholar
Department of Biochemistry
jalemkul[at]vt.edu | (540) 231-9080
More information about the gromacs.org_gmx-users