VASP
To access the staff-supported VASP on the HPC, every individual user must provide documentation of proof that they are on a valid VASP license, and which versions of VASP the license is for. VASP has an online portal that allows license holders to see the list of all individuals on their license, and which versions the license is good for. Please send screenshots to HPC showing this information to oit_hpc@help.ncsu.edu. The Vienna Ab initio Simulation Package (VASP), is a package for performing ab initio quantum mechanical calculations using either Vanderbilt pseudopotentials, or the projector augmented wave method, and a plane wave basis set. See the online Manual for VASP documentation. Here we provide sample files, and give instructions how to run it on the CPUs and GPUs with the staff supported version on Henry2 HPC.
External Links:
VASP website
Version 5.4.1 runs on CPUs and the newer GPUs (NVIDIA):
Table of Contents - Select for a Particular Topic
- VASP input sample
- VASP with GPUs
- VASP with CPUs
- VASP input sample
Note that the HPC staff needs to provide you with permission to run these tests by adding you to the vasp software group.
The input files to run these test can be obtained with:
cp /usr/local/apps/vasp/examples/Si_Liquid_Freezing.tar .
tar -xvf Si_Liquid_Freezing.tar
cd Si_Liquid_Freezing
This example is the same Si_Liquid_Freezing example shown on VASP's online documentation:
VASP Si_Liquid_Freezing example
It does 400 steps of Molecular Dynamics. The example, which takes about 26 minutes, can be shortened by changing NSW in the INCAR file.
VASP for GPUs
To interactively test VASP for GPUs (not recommended for production jobs) do
bsub -Is -n 1 -W 10 -q gpu -R "select[k20m]" -gpu "num=1:mode=shared:mps=yes" tcsh
module load vasp
mpirun vasp_gpu
The interactive way above is NOT recommended for running production jobs. However, because there are many VASP input parameters that conflict with each other, it is useful for quickly debugging input scripts.
The P100 GPU also works - the P100 GPU has the highest double-precision performance. Both the K20M and the P100 GPUs have high double-precison performance, so it is recommended to only run VASP on those nodes.
To test in batch mode (the recommended way), create "testGPU.bsub" with the following lines:
#!/bin/tcsh
#BSUB -n 1
#BSUB -W 20
#BSUB -q gpu
#BSUB -R "select[k20m]"
#BSUB -gpu "num=1:mode=shared:mps=yes"
#BSUB -o out.%J
#BSUB -e err.%J
module load vasp
mpirun vasp_gpu
Run this batch job with:
bsub < testGPU.bsub
Sample output is shown below, and there were no errors.
VASP for CPUs
Note that the HPC staff needs to provide you with permission to run these tests, by adding you to the vasp software group. To test VASP CPUs (interactively) do
bsub -Is -n 1 -W 10 tcsh
module load vasp
mpirun vasp_std
To do the same thing in batch mode, create "testCPU.bsub" with the following lines:
#!/bin/tcsh
#BSUB -n 1
#BSUB -W 10
#BSUB -o out.%J
#BSUB -e err.%J
module load vasp
mpirun vasp_std
Run this batch job with:
bsub < testCPU.bsub
Last modified: February 17 2022 10:23:56.