Differences

This shows you the differences between two versions of the page.

--- slurm [2025/01/13 12:44] – [Output] amcguire
+++ slurm [2025/06/30 17:59] (current) – amcguire
@@ Line 39: / Line 39: @@
 ===== Where to begin =====
-Slurm is a set of command line utilities that can be accessed via the command line from **most** any computer science system you can login to. Using our main shell servers (linux.cs.uchicago.edu) is expected to be our most common use case, so you should start there.
+Slurm is a set of command line utilities that can be accessed via the command line from **most** any computer science system you can login to. Using our main shell servers (''linux.cs.uchicago.edu'') is expected to be our most common use case, so you should start there.
   ssh user@linux.cs.uchicago.edu
@@ Line 88: / Line 88: @@
 === Default Quotas ===
 By default we set a job to be run on one CPU and allocate 100MB of RAM. If you require more than that you should specify what you need. Using the following options will do: ''%%--mem-per-cpu%%'', ''%%--nodes%%'', ''%%--ntasks%%''.
+=== MPI Usage ===
+The AI cluster supports the use of MPI. The following example illustrates its basic use.
+<code>
+amcguire@fe01:~$ cat mpi-hello.c
+#include <mpi.h>
+#include <unistd.h>
+#include <stdio.h>
+int main(int argc, char **argv) {
+    // Initialize MPI
+    MPI_Init(&argc, &argv);
+    // Get the number of processes in the global communicator
+    int count;
+    MPI_Comm_size(MPI_COMM_WORLD, &count);
+    // Get the rank of the current process
+    int rank;
+    MPI_Comm_rank(MPI_COMM_WORLD, &rank);
+    // Get the current hostname
+    char hostname[1024];
+    gethostname(hostname, sizeof(hostname));
+    // Print a hello world message for this rank
+    printf("Hello from process %d of %d on host %s\n", rank, count, hostname);
+    // Finalize the MPI environment before exiting
+    MPI_Finalize();
+}
+amcguire@fe01:~$ cat hello-job.sh
+#!/bin/bash
+#SBATCH -J mpi-hello            # Job name
+#SBATCH -n 2                    # Number of processes
+#SBATCH -t 0:10:00              # Max wall time
+#SBATCH -o hello-job.out        # Output file name
+# Disable the Infiniband transport for OpenMPI (not present on all clusters)
+#export OMPI_MCA_btl="^openib"
+# Run the job (assumes the batch script is submitted from the same directory)
+mpirun -np 2 ./mpi-hello
+amcguire@fe01:~$ mpicc -o mpi-hello mpi-hello.c
+amcguire@fe01:~$ ls -l mpi-hello
+-rwxrwx--- 1 amcguire amcguire 16992 Jun 30 10:49 mpi-hello
+amcguire@fe01:~$ sbatch -w p001,p002 -p peanut-cpu hello-job.sh
+Submitted batch job 1196702
+amcguire@fe01:~$ cat hello-job.out
+Hello from process 0 of 2 on host p001
+Hello from process 1 of 2 on host p002
+</code>
 === Exclusive access to a node ===