Before running an MPI program, place it to a shared location and make sure it is accessible from all cluster nodes. Alternatively, you can have a local copy of your program on all the nodes. In this case, make sure the paths to the program match.
Run the MPI program using the mpirunexec command. The command line syntax is as follows:
$ mpirun > mpiexec -n <# of processes> -ppn <# of processes per node> -f <hostfile> ./myprog.exe
For example:
$ mpirun > mpiexec -n 4 -ppn 2 -f hosts ./myprog.exe
The mpiexec command launches the Hydra process manager, which controls the execution of your MPI program on the cluster.
In the command line above:
-n sets the number of MPI processes to launch; if the option is not specified, the process manager pulls the host list from a job scheduler, or uses the number of cores on the machine.
-ppn sets the number of processes to launch on each node; if the option is not specified, processes are assigned to the physical cores on the first node; if the number of cores is exceeded, the next node is used.
-f specifies the path to the host file listing the cluster nodes; alternatively, you can use the -hosts option to specify a comma-separated list of nodes; if hosts are not specified, the local node is used.
myprog.exe is the name of your MPI program.
The mpirun command is a wrapper around the mpiexec.hydra command, which invokes the Hydra process manager. Consequently, you can use all mpiexec.hydra options with the mpirun command.
For the list of all available options, run mpirunexec with the -help option, or see the Intel® MPI Library Developer Reference, section Command Reference > Hydra Process Manager Command.
You can specify all mpiexec.hydra options with the mpirun command.
The mpirun command detects if the MPI job is submitted from within a session allocated using a job scheduler like PBS Pro* or LSF*. Thus, you are recommended to use mpirun when an MPI program is running under a batch scheduler or job manager.