...

CP2K is freely available for all users under the GNU General Public License (GPL).

Modules

...

CP2K is an MPI-parallel application. You can use either mpirun or srun as the job starter for CP2K. If you use the opt for mpirun command, then, apart from loading the corresponding impi or openmpi modules, CPU and/or GPU pinning should be carefully carried out.

CP2K Version	Modulefile	Requirement	Compute Partitions	Support	CPU/GPU	Lise/Emmy
7.1	cp2k/7.1	impi/2021.13	Rocky Linux 9	omp libint fftw3 libxc elpa parallel mpi3 scalapack xsmm spglib mkl	/	/
2022.2	cp2k/2022.2	intel/2021.2 (Lise) intel/2022.2 (Emmy)	CentOS 7	libint, fftw3, libxc, elpa, scalapack, cosma, xsmm, spglib, mkl, sirius, libvori and libbqb	/	/
2023.1	cp2k/2023.1	intel/2021.2 (Lise) intel/2022.2 (Emmy)	CentOS 7	Lise: libint, fftw3, libxc, elpa, scalapack, cosma, xsmm, spglib, mkl, sirius, libvori and libbqb. Emmy: libint, fftw3, libxc, elpa, scalapack, cosma, xsmm, spglib, mkl and sirius.	/	/
2023.1	cp2k/2023.1	openmpi/gcc.11/4.1.4 cuda/11.8	GPU A100	libint, fftw3, libxc, elpa, elpa_nvidia_gpu, scalapack, cosma, xsmm, dbcsr_acc, spglib, mkl, sirius, offload_cuda, spla_gemm, m_offloading, libvdwxc	/	/
2023.2	cp2k/2023.2	intel/2021.2 impi/2021.7.1	CentOS 7	libint, fftw3, libxc, elpa, scalapack, cosma, xsmm, spglib, mkl, sirius, libvori and libbqb	/	/
2023.2	cp2k/2023.2	openmpi/gcc.11/4.1.4 cuda/11.8	GPU A100	libint, fftw3, libxc, elpa, elpa_nvidia_gpu, scalapack, cosma, xsmm, dbcsr_acc, spglib, mkl, sirius, offload_cuda, spla_gemm, m_offloading, libvdwxc	/	/
2024.1	cp2k/2023.2	impi/2021.13	Rocky Linux 9	omp,libint,fftw3,fftw3_mkl,libxc,elpa,parallel,mpi_f08,scalapack,xsmm,spglib,mkl,sirius,hdf5	/	/

Remark: cp2k needs special attention when running on GPUs.

1. You need to check if, for your problem, a considerable acceleration is expected. E.g., for the following test cases, a performance degradation has been reported: https://www.cp2k.org/performance:piz-daint-h2o-64, https://www.cp2k.org/performance:piz-daint-h2o-64-ri-mp2, https://www.cp2k.org/performance:piz-daint-lih-hfx, https://www.cp2k.org/performance:piz-daint-fayalite-fist

2. GPU pinning is required (see the example of a job script below). Don't forget to make executable the script that takes care of the GPU pinning. In the example, this is achieved with:

chmod +x gpu_bind.sh

Using CP2K as a library

Starting from version 2023.2, CP2K has been compiled enabling the option that allows it to be used as a library: libcp2k.a can be found inside $CP2K_LIB_DIR. The header libcp2k.h is located in $CP2K_HEADER_DIR, and the module files (.mod), possibly needed by Fortran users, are in $CP2K_MOD_DIR.

For more details, please refer to the documentation.

Example Jobscripts

#!/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 --time=06:00:00
#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 intelimpi/2021.2 impi/2021.7.113 

# Select the appropriate version
module load cp2k/20232024.1

srunmpirun cp2k.psmp input > output

#!/bin/bash 
#SBATCH --partition=gpu-a100  
#SBATCH --time =12:00:00
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=244
#SBATCH --cpus-per-task=4
#SBATCH --time=06:00:0018
#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 intel/2022.2 impi/2021.6 cp2k/2023.1
srungcc/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

#!/bin/bash
export CUDA_VISIBLE_DEVICES=$OMPI_COMM_WORLD_LOCAL_RANK
$@

Remark on OpenMP

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:

export OMP_STACKSIZE=512M
ulimit -s unlimited

...