Compiling and running MPI codes

To illustrate the procedure we will compile and run a MPI hello world example from mpitutorial.com. First we download the source code:
$ wget https://raw.githubusercontent.com/mpitutorial/mpitutorial/gh-pages/tutorials/mpi-hello-world/code/mpi_hello_world.c

 

Compiling with GCC

To compile the code, we first need to load the gcc and mvapich2 modules:

$ module load gcc
$ module load mvapich2                                                                                                                                       
Then we can produce the executable called mpi_hello_world by compiling the source code mpi_hello_world.c:
$ mpicc mpi_hello_world.c -o mpi_hello_world
The mpicc tool is a wrapper around the gcc compiler that adds the correct options for linking MPI codes and if you are curious you can run mpicc -show to see what it does.
To run the executable we create a Slurm submission script called run_mpi_hello_world.sh, where we ask to run a total of 4 MPI tasks with (at max) 2 tasks per node:
#!/bin/bash

#SBATCH --time 00-00:05:00
#SBATCH --mem=2G
#SBATCH --ntasks 4
#SBATCH --ntasks-per-node 2
#SBATCH --cpus-per-task 1

module purge
module load gcc
module load mvapich2
module list

EXE=mpi_hello_world
[ ! -f  $EXE ] && echo "EXE $EXE not found." && exit 1

srun  $EXE
Finally, we submit our MPI job with:
$ sbatch run_mpi_hello_world.sh

Upon completion you should get something like:
...

Hello world from processor dna001.curnagl, rank 1 out of 4 processors
Hello world from processor dna001.curnagl, rank 3 out of 4 processors
Hello world from processor dna004.curnagl, rank 0 out of 4 processors
Hello world from processor dna004.curnagl, rank 2 out of 4 processors

It is important to check is that you have a single group of 4 processors and not 4 groups of 1 processor. If that's the case, you can now compile and run your own MPI application.

The important bit of the script is the srun $EXE as MPI jobs but be started with a job launcher in order to run multiple processes on multiple nodes. 

Compiling with Intel

Rather than compiling with GCC and MVAPICH2, you can compile and run your MPI application with the tools from Intel. So, instead of loading the modules gcc and mpich, you load the modules intel and intel-oneapi-mpi:

$ module load intel
$ module load intel-oneapi-mpi 

To compile, use the Intel compiler wrapper mpiicc (rather than mpiic which is a wrapper for gcc):

$ mpiicc mpi_hello_world.c -o mpi_hello_world

And to run, simply load the right modules accordingly:

#!/bin/bash

#SBATCH --time 00-00:05:00
#SBATCH --mem=2G
#SBATCH --ntasks 4
#SBATCH --ntasks-per-node 2
#SBATCH --cpus-per-task 1

module purge
module load intel
module load intel-oneapi-mpi
module list

EXE=mpi_hello_world
[ ! -f  $EXE ] && echo "EXE $EXE not found." && exit 1

srun $EXE

 


Révision #5
Créé 25 mai 2021 12:50:11 par Ewan Roche
Mis à jour 14 juillet 2021 07:54:41 par Ewan Roche