VASP

Overview

From the VASP website

The Vienna Ab initio Simulation Package (VASP) is a computer program for atomic scale materials modelling, e.g. electronic structure calculations and quantum-mechanical molecular dynamics, from first principles.

VASP computes an approximate solution to the many-body Schroedinger equation, either within density functional theory (DFT), solving the Kohn-Sham equations, or within the Hartree-Fock (HF) approximation, solving the Roothaan equations. Hybrid functionals that mix the Hartree-Fock approach with density functional theory are implemented as well. Furthermore, Green functions methods (GW quasiparticles, and ACFDT-RPA) and many-body perturbation theory (2nd-order Moller-Plesset) > are available in VASP.

Restrictions on Use

The use of VASP is restricted to particular research groups and different versions may be restricted to different research groups. This is via UNIX groups, and your user ID must be a member of the relevant group and you must ensure you have changed to that group via the Linux command

newgrp vasp

Job Scripts

For maximum speed this can be run on multiple nodes, but to do so requires entries both in the job script and also the input files.

The following script is a basic one for the compute partition using the basic vasp_std binary ( vasp_gam and vasp_ncl are also available). You can change the level of parallelism by changing the value in –ntasks=.

#!/bin/bash -l
#SBATCH --account={youraccountnamehere}
#SBATCH --time={timerequiredtorunyourjobhere}
module purge
module load vasp/5.4.1.03Aug16
cd $1
newgrp vasp
export OMP_NUM_THREADS=1
mpirun -np ${SLURM_NTASKS} vasp_std

You must also set the value of NPAR in INCAR in the directory where the data is, which is {/path/to/where/your/data/is} in the submission examples below:

NPAR = 28

Assuming the above job script is called vasp.sh you can then submit this to one node with 28 cores with

sbatch --ntasks=28 --partition=compute vasp.sh {/path/to/where/your/data/is}

and over two nodes

sbatch --ntasks=56 --partition=compute vasp.sh {/path/to/where/your/data/is}

and so on, also with NPAR=28

For further advice on setting both NPAR and NCORE see NPAR and NCORE advice

Web Page

Official website

Versions

5.4.4

Flavours

2

Architecture

x86_64

Group
module load site-local
Build Details

Compiled with:

  • intel/compiler/64/2016.3.210
  • intel/mkl/64/2016.3.210
  • intel/mpi/64/2016.3.210
  • CPP_OPTIONS=’-DHOST=”LinuxIFC” -DMPI -DMPI_BLOCK=8000 -Duse_collective -DscaLAPACK -DCACHE_SIZE=4000 -Davoidalloc -Duse_bse_te -Dtbdyn -Duse_shmem’
  • FC=mpiifort
  • FCL=’mpiifort -mkl=sequential -lstdc++’
  • FFLAGS=’-assume byterecl -w’
  • OFLAG=’-O2 -xCORE-AVX2′
  • DEBUG=’-O0 -xCORE-AVX2′
  • MKL_PATH=$(MKLROOT)/lib/intel64
  • BLAS=
  • LAPACK=
  • BLACS=-lmkl_blacs_intelmpi_lp64
  • SCALAPACK=’$(MKL_PATH)/libmkl_scalapack_lp64.a $(BLACS)’
  • INCS=-I$(MKLROOT)/include/fftw
  • LLIBS=’ $(SCALAPACK) $(LAPACK) $(BLAS)’
  • CFLAGS_LIB=’-O -xCORE-AVX2′
  • FFLAGS_LIB=’-O1 -xCORE-AVX2′
Usage
module load vasp/5.4.4/2
Example Job Scripts

None yet.

Status

Run all examples except:

  • Non collinear examples, which report ERROR: non collinear calculations require that VASP is compiled without the flag -DNGXhalf and -DNGZhalf
  • Some examples that only work with low parallelism.

It has also been tested on some longer examples (‘FM’ from Pooja Panchmatia at Loughborough) with a run time on 8 cores of ~700 minutes.

5.4.1.03Aug16

Fully patched version of 5.4.1.

Flavours

2

Architecture

x86_64

Group
module load site-local
Build Details

This version uses an out-of-date version of MLNX_OFED and should not be used. The module will be withdrawn in future.

Compiled with:

Usage
module load vasp/5.4.1.03Aug16/2
Example Job Scripts

None yet.

Status

Run all examples except:

  • Non collinear examples, which report ERROR: non collinear calculations require that VASP is compiled without the flag -DNGXhalf and -DNGZhalf
  • Some examples that only work with low parallelism.

It has also been tested on some longer examples (‘FM’ from Pooja Panchmatia at Loughborough) with a run time on 8 cores of ~700 minutes.

3

Architecture

x86_64

Group
module load site-local
Build Details

This version uses an out-of-date version of MLNX_OFED and should not be used. The module will be withdrawn in future.

Compiled with:

Usage
module load vasp/5.4.1.03Aug16/3
Example Job Scripts

None yet.

Status

Run all examples except:

  • Non collinear examples, which report ERROR: non collinear calculations require that VASP is compiled without the flag -DNGXhalf and -DNGZhalf
  • Some examples that only work with low parallelism.

It has also been tested on some longer examples (‘FM’ from Pooja Panchmatia at Loughborough) with a run time on 8 cores of ~700 minutes.