THIS IS THE INSTRUCTIONS FOR RUNNING CODES IN PARALLEL ON MPI

After logging in:

1) Create a file, say 'mpihosts', or any other name, and put 
in it the name of the computers we want to use.

node0
node1
node3
node4
node5
node6
node7

2) Save, and on the command line run:

$ lamboot -v mpihosts

This starts the lam dameon on each of the nodes.  These let 
you start programs on each node and communicate between them.

3) Compile your program.

$ mpicc -o program program.c

This is a wrapper for gcc, and automatically includes 
whichever libraries and such you need for lam/mpi programs.  
You can pass any normal gcc arguments, for example, if you 
need the math library, put '-lm' at the end.

4) Run your program.

$ mpirun -np 7 program
or
$ mpirun N program

This runs your program on the specified number of 
processors.  'N' is a shortcut for running on all available 
processors.  You can pass your program arguments like normal 
after your program name.

$ mpirun N program arg1 arg2 ...

5) At the end of your session, halt the lam nodes.

$ mpihalt -v

This kills all of the dameons.