:construction_site: under construction, please consult [this](https://crimson.oca.eu/fr/acces-fonctionnement/gestion-des-jobs/1340-exemples-de-scripts) in the meantime
[**M**essage **P**assing **I**nterface](https://www.mpi-forum.org/)(MPI) can be used to distribute parallel aplication on more than one node. We support the Intel implementation.[^1]
[**M**essage **P**assing **I**nterface](https://www.mpi-forum.org/)(MPI) can be used to distribute parallel aplication on more than one node. We support the Intel implementation.[^1]
15:08:47 [alainm@pollux MPI]# more log-dice-33544995.out
Hi, this is processs 0 from node x001.cluster.local
...
Hi, this is processs 3 from node x002.cluster.local
..
Let's play dices.
Who's de best ?
This is 0, I got 2, I'm such a looser...
...
This is 3, I got 4, I'm so talented
15:08:54 [alainm@pollux MPI]#
```
## Script explanation
We are now describing the Slurm script [run.slurm](SLURM/MPI/build.sh). Note that we will run 2 task per node to illustrate the fact that the job will use more than one node, although this toy example would be happy with one job per core.
```
[^1]:which has an impact in term of Slurm integration.
```
[^1]:which has an impact in term of Slurm integration.