Install DFT-FE

These install scripts provide a set of executable functions that install the necessary dependencies of DFT-FE on OLCF Frontier machine.

To use these scripts, we assume you have cloned this repository onto a system where you intend to install DFT-FE. For example, I installed it into /lustre/orion/[projid]/scratch/$USER/install_DFTFE after cloning into the scatch directory

cd /lustre/orion/[projid]/scratch/$USER
git clone https://github.com/dftfeDevelopers/install_DFTFE.git install_DFTFE
cd install_DFTFE
git checkout frontierScriptROCM6Dealii9.5.2

Pre-requisites

Because it's a better shell, the scripts are written in the rc shell language. Install rc by running

cp src/rcrc $HOME/.rcrc
. ./bin/getrc.sh $HOME/$LMOD_SYSTEM_NAME
rc -l

Note that getrc installs into the $HOME/$LMOD_SYSTEM_NAME/bin directory, and adds that to your PATH. Also note that in rc shell, the export keyword is not used when setting environment variables.

Copying the rcrc startup file to your home directory provides the module command (in case your lmod version is old, and doesn't yet recognize the rc shell).

Module Environment

The module environment intended to run DFT-FE has been extracted into env2/env.rc. Edit this file before proceeding any further. Make sure that your module environment contains some version of the pre-requisites mentioned there (e.g. python3 and openblas). This environment file is used both by the install and run phases of DFT-FE.

Running the installation

The installation itself is contained within the functions in dftfe2.rc. Source this script using

. ./dftfe2.rc

and then run the functions listed in that file manually, in order. For example,

install_alglib
install_libxc
install_spglib
install_p4est
install_scalapack
install_ofi_rccl (optional)
install_elpa
install_kokkos
install_dealii
install_dftd4 (optional)
compile_dftfe

Each function follows a standard pattern - download source into $WD/src, patch, compile, and install into $INST. It is HIGHLY recommended to check all warnings and errors from these installs to be sure you have not ended up with broken packages.

Running DFT-FE

DFT-FE is built in real and cplx versions, depending on whether you want to enable k-points (implemented in the cplx version only).

Assuming you have already sourced env2/env.rc, an example batch script running GPU-enabled DFT-FE on 280 nodes is below:

#!$HOME/$LMOD_SYSTEM_NAME/bin/rc
#SBATCH -A spy007
#SBATCH -J dft14584
#SBATCH -t 00:25:00
#SBATCH -p batch
#SBATCH -N 280
#SBATCH --gpus-per-node 8
#SBATCH --ntasks-per-gpu 1
#SBATCH --gpu-bind closest

OMP_NUM_THREADS = 1
MPICH_VERSION_DISPLAY=1
MPICH_ENV_DISPLAY=1
MPICH_OFI_NIC_POLICY = NUMA 
MPICH_GPU_SUPPORT_ENABLED=1
MPICH_SMP_SINGLE_COPY_MODE=NONE

LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$INST/lib
LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$INST/lib/lib64
LD_LIBRARY_PATH=$CRAY_LD_LIBRARY_PATH:$LD_LIBRARY_PATH


BASE = $WD/src/dftfe/build/release/real
n=`{echo $SLURM_JOB_NUM_NODES '*' 8 | bc}

srun -n $n -c 7 --gpu-bind closest \
          $BASE/dftfe parameterFileGPU.prm > output

This uses SLURM_JOB_NUM_NODES to compute the number of MPI ranks to use as one per GCD (8 per node). If you wish to run on a different number of nodes, only the #SBATCH -N 280 needs to be changed.