User`s guide

Using the Intel(R) MPI Library

2.7. Running an MPI Program

To launch programs linked with the Intel® MPI Library use the mpiexec command:

> mpiexec.exe -n <# of processes> myprog.exe

The wmpiexec utility is a GUI wrapper for mpiexec.exe. See the Intel® MPI Library Reference

Manual for more details.

To set the number of processes on the local node, use the only required mpiexec -n option.

To set names of hosts and number of processes, use the -hosts option :

> mpiexec.exe -hosts 2 host1 2 host2 2 myprog.exe

If you are using a network fabric as opposed to the default fabric, use the -genv option to set the

I_MPI_FABRICS variable.

For example, to run an MPI program using the shm fabric, type in the following command:

> mpiexec.exe -genv I_MPI_FABRICS shm -n <# of processes> \

myprog.exe

To run the program, you may use the -configfile option :

> mpiexec.exe -configfile config_file

The configuration file contains:

-host host1 -n 1 -genv I_MPI_FABRICS shm:dapl myprog.exe

-host host2 -n 1 -genv I_MPI_FABRICS shm:dapl myprog.exe

For the rdma capable fabric, use the following command:

> mpiexec.exe -hosts 2 host1 1 host2 1 -genv I_MPI_FABRICS dapl myprog.exe

You can select any supported device. For more information, see Section Selecting a Network

Fabric.

If you successfully run your application using the Intel® MPI Library, you can move your

application from one cluster to another and use different fabrics between the nodes without re-

linking. If you encounter problems, see Troubleshooting for possible solutions.

2.8. Controlling MPI Process Placement

The mpiexec command controls how the ranks of the processes are allocated to the

nodes of the cluster. By default, the mpiexec command uses group round-robin

assignment, putting consecutive MPI process on all processor ranks of a node. This

placement algorithm may not be the best choice for your application, particularly for

clusters with symmetric multi-processor (SMP) nodes.