...
CP2K is an MPI-parallel application. You can use either mpirun or srun as the job starter for CP2K. If you opt for mpirun, then, apart from loading the corresponding impi or openmpi modules, CPU and/or GPU pinning should be carefully carried out.
...
| Codeblock |
|---|
| language | bash |
|---|
| title | Lise (using srun)For compute nodes with Rocky Linux 9 |
|---|
|
#!/bin/bash
#SBATCH --time=12:00:00
#SBATCH --partition=cpu-clx
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=24
#SBATCH --cpus-per-task=4
#SBATCH --job-name=cp2k
export SLURM_CPU_BIND=none
export OMP_NUM_THREADS=${SLURM_CPUS_PER_TASK}
# Binding OpenMP threads
export OMP_PLACES=cores
export OMP_PROC_BIND=close
# Binding MPI tasks
export I_MPI_PIN=yes
export I_MPI_PIN_DOMAIN=omp
export I_MPI_PIN_CELL=core
# Our tests have shown that CP2K has better performance with psm2 as libfabric provider
# Check if this also apply to your system
# To stick to the default provider, comment out the following line
export FI_PROVIDER=psm2
module load intel/2021.2 impi/2021.7.113 cp2k/20232024.21
srunmpirun cp2k.psmp input > output |
| Codeblock |
|---|
| language | bash |
|---|
| title | Lise For compute nodes with CentOS 7 (using mpirun) |
|---|
|
#!/bin/bash
#SBATCH --time=12:00:00
#SBATCH --partition standard96
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=24
#SBATCH --cpus-per-task=4
#SBATCH --job-name=cp2k
export SLURM_CPU_BIND=none
export OMP_NUM_THREADS=${SLURM_CPUS_PER_TASK}
# Binding OpenMP threads
export OMP_PLACES=cores
export OMP_PROC_BIND=close
# Binding MPI tasks
export I_MPI_PIN=yes
export I_MPI_PIN_DOMAIN=omp
export I_MPI_PIN_CELL=core
module load intel/2021.2 impi/2021.7.1 cp2k/2023.2
mpirun cp2k.psmp input > output |
...
| Codeblock |
|---|
| language | bash |
|---|
| title | Lise For compute nodes with CentOS 7 (using mpirun): on srun) |
|---|
|
#!/bin/bash
#SBATCH --time=12:00:00
#SBATCH --partition standard96
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=24
#SBATCH --cpus-per-task=4
#SBATCH --job-name=cp2k
export OMP_NUM_THREADS=${SLURM_CPUS_PER_TASK}
module load intel/2021.2 impi/2021.7.1 cp2k/2023.2
srun cp2k.psmp input > output |
| Codeblock |
|---|
| language | bash |
|---|
| title | For Nvidia A100 GPU nodes |
|---|
|
#!/bin/bash
#SBATCH --partition=gpu-a100
#SBATCH --time=12:00:00
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=4
#SBATCH --cpus-per-task=18
#SBATCH --job-name=cp2k
export SLURM_CPU_BIND=none
export OMP_NUM_THREADS=${SLURM_CPUS_PER_TASK}
export OMP_PLACES=cores
export OMP_PROC_BIND=close
module load gcc/11.3.0 openmpi/gcc.11/4.1.4 cuda/11.8 cp2k/2023.2
# gpu_bind.sh (see the following script) should be placed inside the same directory where cp2k will be executed
# Don't forget to make gpu_bind.sh executable by running: chmod +x gpu_bind.sh
mpirun --bind-to core --map-by numa:PE=${SLURM_CPUS_PER_TASK} ./gpu_bind.sh cp2k.psmp input > output |
...
| Codeblock |
|---|
| language | bash |
|---|
| title | gpu_bind.sh |
|---|
|
#!/bin/bash
export CUDA_VISIBLE_DEVICES=$OMPI_COMM_WORLD_LOCAL_RANK
$@ |
| HTML Kommentar |
|---|
Commenting out this block, as Lise and Emmy have separate documentation pages now. | Codeblock |
|---|
| language | bash |
|---|
| title | Emmy (using srun) |
|---|
| #!/bin/bash
#SBATCH --time=12:00:00
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=24
#SBATCH --cpus-per-task=4
#SBATCH --job-name=cp2k
export OMP_NUM_THREADS=${SLURM_CPUS_PER_TASK}
module load intel/2022.2 impi/2021.6 cp2k/2023.1
srun cp2k.psmp input > output |
|
Depending on the problem size, it may happen that the code stops may stop with a segmentation fault due to insufficient stack size or due to threads exceeding their stack space. To circumvent this, we recommend inserting in the jobscriptjob script:
| Codeblock |
|---|
|
export OMP_STACKSIZE=512M
ulimit -s unlimited |
...