Skip to content

Argonne Leadership Computing Facility

VASP

VASP 6.x.x in Polaris (NVHPC+OpenACC+OpenMP+CUDA math+CrayMPI)

The Vienna Ab initio Simulation Package (VASP) is a software package for performing electronic structure calculations with periodic boundary conditions. It is most commonly used that to perform density functional theory (DFT) calculations in a planewave basis using the projector augemented wave (PAW) method. A more complete description of VASP can be found here.

Users must have a license to use this code on ALCF systems. More information on how to get access to VASP binaries can be found here.

General compiling/installing instructions provided by VASP support

Instructions and samples of makefile.include could be found in the vasp.at wiki page.

The follow makefile.include was tailored for Polaris, originally taken from here.

# Precompiler options
CPP_OPTIONS = -DHOST=\"LinuxNV\" \
              -DMPI -DMPI_BLOCK=8000 -Duse_collective \
              -DscaLAPACK \
              -DCACHE_SIZE=4000 \
              -Davoidalloc \
              -Dvasp6 \
              -Duse_bse_te \
              -Dtbdyn \
              -Dqd_emulate \
              -Dfock_dblbuf \
              -D_OPENMP \
              -D_OPENACC \
              -DUSENCCL -DUSENCCLP2P\

CPP        = nvfortran -Mpreprocess -Mfree -Mextend -E $(CPP_OPTIONS) $*$(FUFFIX)  > $*$(SUFFIX)

FC         = ftn -acc -gpu=cc80 -mp -target-accel=nvidia80
FCL        = ftn -acc -gpu=cc80 -c++libs -target-accel=nvidia80

FREE       = -Mfree

FFLAGS     = -Mbackslash -Mlarge_arrays

OFLAG      = -fast

DEBUG      = -Mfree -O0 -traceback

# Specify your NV HPC-SDK installation, try to set NVROOT automatically
NVROOT     =$(shell which nvfortran | awk -F /compilers/bin/nvfortran '{ print $$1 }')
# ...or set NVROOT manually
NVHPC      ?= /opt/nvidia/hpc_sdk
NVVERSION  = 20.9
#NVROOT     = $(NVHPC)/Linux_x86_64/$(NVVERSION)

# Use NV HPC-SDK provided BLAS and LAPACK libraries
LIBAOCL=/soft/libraries/aocl/3.2.0
BLAS       = ${LIBAOCL}/lib/libblis-mt.a
LAPACK     = ${LIBAOCL}/lib/libflame.a

BLACS      =
SCALAPACK  =
#SCALAPACK  = -Mscalapack
#SCALAPACK  = ${LIBAOCL}/lib/libscalapack.a

CUDA       = -cudalib=cublas,cusolver,cufft,nccl -cuda

LLIBS      = $(SCALAPACK) $(LAPACK) $(BLAS) $(CUDA)

# Software emulation of quadruple precsion
QD         ?= $(NVROOT)/compilers/extras/qd
LLIBS      += -L$(QD)/lib -lqdmod -lqd
INCS       += -I$(QD)/include/qd

#INCS       += -I/usr/include/linux 
#INCS       += -I/usr/include/c++/7/tr1 
#INCS       += -I/usr/include/c++/7 
#INCS       += -I/usr/include/x86_64-linux-gnu/c++/7
#INCS       += -I/lus/theta-fs0/software/spack/spack-dev/opt/spack/linux-sles15-x86_64/gcc-9.3.0/gcc-10.2.0-r7v3naxd5xgzzaqxoe73jj2ytwuddamr/lib/gcc/x86_64-pc-linux-gnu/10.2.0/include/

# Use the FFTs from fftw
FFTW       ?= ${LIBAOCL}
LLIBS      += -L$(FFTW)/lib -lfftw3 -lfftw3_omp -lomp
#INCS       += -I/soft/libraries/aocl/3.2.0/include_LP64/
INCS       += -I$(FFTW)/include

OBJECTS    = fftmpiw.o fftmpi_map.o fftw3d.o fft3dlib.o

# Redefine the standard list of O1 and O2 objects
SOURCE_O1  := pade_fit.o
SOURCE_O2  := pead.o

# For what used to be vasp.5.lib
CPP_LIB    = $(CPP)
FC_LIB     = nvfortran
CC_LIB     = cc
CFLAGS_LIB = -O $(INCS) -c++libs -cuda
FFLAGS_LIB = -O1 -Mfixed
FREE_LIB   = $(FREE)

OBJECTS_LIB= linpack_double.o getshmem.o

# For the parser library
#CXX_PARS   = nvc++ --no_warnings -I/lus/theta-fs0/software/spack/spack-dev/opt/spack/linux-sles15-x86_64/gcc-9.3.0/gcc-10.2.0-r7v3naxd5xgzzaqxoe73jj2ytwuddamr/include/c++/10.2.0/ -I/lus/theta-fs0/software/spack/spack-dev/opt/spack/linux-s
les15-x86_64/gcc-9.3.0/gcc-10.2.0-r7v3naxd5xgzzaqxoe73jj2ytwuddamr/include/c++/10.2.0/x86_64-pc-linux-gnu -I/lus/theta-fs0/software/spack/spack-dev/opt/spack/linux-sles15-x86_64/gcc-9.3.0/gcc-10.2.0-r7v3naxd5xgzzaqxoe73jj2ytwuddamr/lib/gcc
/x86_64-pc-linux-gnu/10.2.0/include -I/lus/theta-fs0/software/spack/spack-dev/opt/spack/linux-sles15-x86_64/gcc-9.3.0/gcc-10.2.0-r7v3naxd5xgzzaqxoe73jj2ytwuddamr/lib/gcc/x86_64-pc-linux-gnu/10.2.0/include-fixed/
CXX_PARS   = nvc++ --no_warnings 

# Normally no need to change this
SRCDIR     = ../../src
BINDIR     = ../../bin

Setting up compiler and libraries with module

The follow modules will update the include and libraries paths used by the Cray compiler wrapper ftn to load additional math libraries for the CPU.

module purge
module load PrgEnv-nvhpc
module load cray-libsci
module load craype-accel-nvidia8

Compiling VASP

Once the modules are loaded and a makefile.include is in the vasp folder, compiling all the object files and binaries is done with:

make -j1

Running VASP in Polaris

An example of a submission script could be found here /soft/applications/vasp/vasp.6.3.2/submit-polaris2023.sh , which would looks something similar to:

#!/bin/sh
#PBS -l select=1:system=polaris  
#PBS -l place=scatter
#PBS -l walltime=0:30:00
#PBS -l filesystems=home:grand:eagle
#PBS -q debug
#PBS -A Catalyst

module load PrgEnv-nvhpc
module load cray-libsci

export MPICH_GPU_SUPPORT_ENABLED=1
NNODES=`wc -l < $PBS_NODEFILE`
NRANKS=4
NDEPTH=8
NTHREADS=1
NGPUS=4
NTOTRANKS=$(( NNODES * NRANKS ))

mpiexec -n ${NTOTRANKS} --ppn ${NRANKS} --depth ${NDEPTH} --cpu-bind depth --env OMP_NUM_THREADS=${NTHREADS} /path_to_vasp/bin/vasp_std

Submission scripts should have executable attibutes to be used with qsub script mode.

chmod +x example-script.sh
qsub  example-script.sh

Known issues

  • Undefined MPIX_Query_cuda_support function at linking binary: This function is called in src/openacc.F. The MPIX_Query_cuda_support is not included incray-mpich. One workaround to this issue is to comment this function call. See the follow suggested changes marked by !!!!!CHANGE HERE in the file:src/openacc.F
+!!!!!CHANGE HERE 
-      INTERFACE
-        INTEGER(c_int) FUNCTION MPIX_Query_cuda_support() BIND(C, name="MPIX_Query_cuda_support")
-        END FUNCTION
-      END INTERFACE

       CHARACTER(LEN=1) :: ENVVAR_VALUE
       INTEGER :: ENVVAR_STAT

       ! This should tell us if MPI is CUDA-aware
+!!!!!CHANGE HERE 
-       CUDA_AWARE_SUPPORT = MPIX_Query_cuda_support() == 1
+       CUDA_AWARE_SUPPORT = .TRUE.
       ! However, for OpenMPI some env variables can still deactivate it even though the previous
       ! check was positive
       CALL GET_ENVIRONMENT_VARIABLE("OMPI_MCA_mpi_cuda_support", ENVVAR_VALUE, STATUS=ENVVAR_STAT)
       IF (ENVVAR_STAT==0 .AND. ENVVAR_VALUE=='0') CUDA_AWARE_SUPPORT = .FALSE.
       CALL GET_ENVIRONMENT_VARIABLE("OMPI_MCA_opal_cuda_support", ENVVAR_VALUE, STATUS=ENVVAR_STAT)
       IF (ENVVAR_STAT==0 .AND. ENVVAR_VALUE=='0') CUDA_AWARE_SUPPORT = .FALSE.
       ! Just in case we might be non-OpenMPI, and their MPIX_Query_cuda_support behaves similarly
       CALL GET_ENVIRONMENT_VARIABLE("MV2_USE_CUDA", ENVVAR_VALUE, STATUS=ENVVAR_STAT)
       IF (ENVVAR_STAT==0 .AND. ENVVAR_VALUE=='0') CUDA_AWARE_SUPPORT = .FALSE.
       CALL GET_ENVIRONMENT_VARIABLE("MPICH_RDMA_ENABLED_CUDA", ENVVAR_VALUE, STATUS=ENVVAR_STAT)
       IF (ENVVAR_STAT==0 .AND. ENVVAR_VALUE=='0') CUDA_AWARE_SUPPORT = .FALSE.
       CALL GET_ENVIRONMENT_VARIABLE("PMPI_GPU_AWARE", ENVVAR_VALUE, STATUS=ENVVAR_STAT)
       IF (ENVVAR_STAT==0) CUDA_AWARE_SUPPORT =(ENVVAR_VALUE == '1')
+!!!!!CHANGE HERE 
+       CALL GET_ENVIRONMENT_VARIABLE("MPICH_GPU_SUPPORT_ENABLED", ENVVAR_VALUE, STATUS=ENVVAR_STAT)
+       IF (ENVVAR_STAT==0) CUDA_AWARE_SUPPORT =(ENVVAR_VALUE == '1')