job_cpu.sh

#!/bin/bash

# The budget account to use
#SBATCH --account=tra210016p

# The partition to use
#SBATCH --partition=RM

# Jobs are capped at 1 minute (Your code should run for ~10 seconds anyway)
#SBATCH --time=00:01:00

# The name of the output file
#SBATCH --output=output.txt
# The name of the error file
#SBATCH --error=error.txt

# The number of nodes (max. 4)
#SBATCH --nodes=4

# The number of MPI processes per node
#SBATCH --ntasks-per-node=50

# The number of OpenMP threads per MPI process
#SBATCH --cpus-per-task=1


# The number of OpenMP threads. If using MPI, it is the number of OpenMP threads per MPI process
export OMP_NUM_THREADS=$SLURM_CPUS_PER_TASK

# Place OpenMP threads on cores
export OMP_PLACES=cores

# Keep the OpenMP threads where they are
export OMP_PROC_BIND=true

# Load the modules needed (do not change the modules)
ml purge
module load cuda/12.6.1 nvhpc/22.9 openmpi/4.0.5-nvhpc22.9

# Compile everything (do not change the makefile)
make

# Execute the program 
mpirun -n $SLURM_NTASKS --bind-to none --report-bindings ./bin/main