VASP编译安装

环境：242集群，Rocky Linux 8.8，Slurm作业管理系统，Intel oneAPI 2022.02编译器

下载需要的安装包¶

管理节点可上网，使用wget下载需要的安装包；直接使用wget下载https链接会报证书错误，需要加--no-check-certificate参数，或者进行如下操作：

[userA@matrix ~]$ echo "check-certificate = off" > ~/.wgetrc

部分集群管理节点wget下载不好使，可本地电脑下载好安装包，上传到用户目录。

[userA@matrix ~]$ mkdir temp
[userA@matrix ~]$ cd temp
[userA@matrix temp]$ wget https://support.hdfgroup.org/ftp/lib-external/szip/2.1.1/src/szip-2.1.1.tar.gz
[userA@matrix temp]$ wget https://www.zlib.net/zlib-1.3.tar.gz
[userA@matrix temp]$ wget https://support.hdfgroup.org/ftp/HDF5/releases/hdf5-1.12/hdf5-1.12.0/src/hdf5-1.12.0.tar.bz2
[userA@matrix temp]$ wget https://github.com/wannier-developers/wannier90/archive/v3.1.0.tar.gz
[userA@matrix temp]$ wget http://www.tddft.org/programs/libxc/down.php?file=6.2.2/libxc-6.2.2.tar.gz
[userA@matrix temp]$ wget http://theory.cm.utexas.edu/code/vtstcode-198.tgz

salloc申请编译计算节点¶

集群管理节点一般做了限制，不允许在上面直接运行作业或编译，我们也没专门设置编译节点，因此需要在计算节点上编译。为避免正常作业与编译抢占CPU，我们把编译也按正常作业提交即可。

[userA@matrix temp]$ salloc -N1 -n4 -p gao-grp
salloc: Granted job allocation 5165
salloc: Waiting for resource configuration
salloc: Nodes cu01 are ready for job

-N 申请的计算节点数目；-n 申请的任务核数；-p 申请的作业队列；-w 申请的计算节点名称。账号可使用队列的名称：可使用sacctmgr show ass user=$(whoami) format=part查看。队列所对应的节点可用sinfo或者pestat命令查看。这里我们申请一个计算节点，4核，使用gao-grp队列，申请计算节点为cu01.

HDF5安装¶

在计算节点编译安装，ssh到上面申请的cu01节点。

[userA@matrix temp]$ ssh cu01
[userA@cu01 ~]$ cd temp

szip安装

[userA@cu01 temp]$ mkdir -p /home/userA/soft/vasp/szip-2.1.1
[userA@cu01 temp]$ tar -zxvf szip-2.1.1.tar.gz
[userA@cu01 temp]$ cd szip-2.1.1/
[userA@cu01 szip-2.1.1]$ source /opt/intel/oneapi/setvars.sh
[userA@cu01 szip-2.1.1]$ export CC=icc
[userA@cu01 szip-2.1.1]$ export CXX=icpc
[userA@cu01 szip-2.1.1]$ export FC=ifort
[userA@cu01 szip-2.1.1]$ export CFLAGS='-O3 -xHost -ip'
[userA@cu01 szip-2.1.1]$ export CXXFLAGS='-O3 -xHost -ip'
[userA@cu01 szip-2.1.1]$ export FCFLAGS='-O3 -xHost -ip'
[userA@cu01 szip-2.1.1]$ ./configure --prefix=/home/userA/soft/vasp/szip-2.1.1
[userA@cu01 szip-2.1.1]$ make all install |tee make.log

zlib安装
为了使编译环境单一干净，上一步安装好szip后，exit退出cu01节点，再重新ssh进入cu01节点

[userA@cu01 szip-2.1.1]$ exit
[userA@matrix temp]$ ssh cu01
[userA@cu01 ~]$ cd temp
[userA@cu01 temp]$ tar -zxvf zlib-1.3.tar.gz
[userA@cu01 temp]$ cd zlib-1.3
[userA@cu01 zlib-1.3]$ mkdir -p /home/userA/soft/vasp/zlib-1.3
[userA@cu01 zlib-1.3]$ source /opt/intel/oneapi/setvars.sh
[userA@cu01 zlib-1.3]$ export CC=icc
[userA@cu01 zlib-1.3]$ export CFLAGS='-O3 -xHost -ip'
[userA@cu01 zlib-1.3]$ ./configure --prefix=/home/userA/soft/vasp/zlib-1.3
[userA@cu01 zlib-1.3]$ make all install |tee make.log

HDF5安装
同样，exit退出cu01节点，再重新ssh进入cu01节点

[userA@cu01 zlib-1.3]$ exit
[userA@matrix temp]$ ssh cu01
[userA@cu01 ~]$ cd temp
[userA@cu01 temp]$ tar -jxvf hdf5-1.12.0.tar.bz2
[userA@cu01 temp]$ cd hdf5-1.12.0
[userA@cu01 hdf5-1.12.0]$ mkdir -p /home/userA/soft/vasp/hdf5-1.12.0
[userA@cu01 hdf5-1.12.0]$ source /opt/intel/oneapi/setvars.sh
[userA@cu01 hdf5-1.12.0]$ ./configure --prefix=/home/userA/soft/vasp/hdf5-1.12.0 --enable-fortran --enable-hl --enable-shared CC=mpiicc FC=mpiifort CXX=mpiicpc CFLAGS="-fPIC -O3 -xHost -ip -fno-alias -align" FFLAGS="-fPIC -O3 -xHost -ip -fno-alias -align" CXXFLAGS="-fPIC -O3 -xHost -ip -fno-alias -align" FFLAGS="-I/opt/intel/oneapi/mpi/latest/include -L/opt/intel/oneapi/mpi/latest/lib" --with-szlib=/home/userA/soft/vasp/szip-2.1.1 --with-zlib=/home/userA/soft/vasp/zlib-1.3 --enable-parallel
[userA@cu01 hdf5-1.12.0]$ make -j 4 |tee make.log
[userA@cu01 hdf5-1.12.0]$ make install

VASP-2-Wannier90 interface¶

上传VASP6.4.2源码包到/home/userA/temp下

[userA@cu01 hdf5-1.12.0]$ exit
[userA@matrix temp]$ ssh cu01
[userA@cu01 ~]$ cd temp
[userA@cu01 temp]$ tar -zxvf vasp.6.4.2.tgz
[userA@cu01 temp]$ cd vasp.6.4.2
[userA@cu01 vasp.6.4.2]$ tar -zxvf ~/temp/v3.1.0.tar.gz
[userA@cu01 vasp.6.4.2]$ cd wannier90-3.1.0
[userA@cu01 wannier90-3.1.0]$ source /opt/intel/oneapi/setvars.sh
[userA@cu01 wannier90-3.1.0]$ cp config/make.inc.ifort make.inc

修改make.inc，内容如下：

[userA@cu01 wannier90-3.1.0]$ cat make.inc 

#=====================================================
# For Linux with intel version 11/12 on 64bit machines
#=====================================================
F90 = ifort
#COMMS=mpi
MPIF90=mpiifort
FCOPTS=-O3 -xHost
LDOPTS=-O3 -xHost

#========================================================
# Intel mkl libraries. Set LIBPATH if not in default path
#========================================================

LIBDIR = /opt/intel/oneapi/mkl/2022.2.0/lib/intel64
LIBS   =  -L$(LIBDIR) -lmkl_core -lmkl_intel_lp64 -lmkl_sequential -lpthread

#=======================
# ATLAS Blas and LAPACK
#=======================
#LIBDIR = /usr/local/lib
#LIBS = -L$(LIBDIR)  -llapack -lf77blas -lcblas -latlas

make编译，得到libwannier.a库文件
[userA@cu01 wannier90-3.1.0]$ make

Libxc编译安装¶

[userA@cu01 wannier90-3.1.0]$ exit
[userA@matrix temp]$ ssh cu01
[userA@cu01 ~]$ cd temp/
[userA@cu01 temp]$ tar -zxvf libxc-6.2.2.tar.gz
[userA@cu01 temp]$ cd libxc-6.2.2
[userA@cu01 libxc-6.2.2]$ mkdir -p /home/userA/temp/vasp.6.4.2/libxc-6.2.2
[userA@cu01 libxc-6.2.2]$ source /opt/intel/oneapi/setvars.sh
[userA@cu01 libxc-6.2.2]$ ./configure --prefix=/home/userA/temp/vasp.6.4.2/libxc-6.2.2 --enable-shared --disable-lxc CC=icc FC=ifort CFLAGS="-O3 -xHost" FCFLAGS="-O3 -xHost"
[userA@cu01 libxc-6.2.2]$ make install -j 4 |tee make.log

VTST代码修改¶

[userA@cu01 libxc-6.2.2]$ exit
[userA@matrix temp]$ ssh cu01
[userA@cu01 ~]$ cd temp/
[userA@cu01 temp]$ tar -zxvf vtstcode-198.tgz
[userA@cu01 temp]$ cp -R vtstcode-198/vtstcode6.4/* vasp.6.4.2/src/
[userA@cu01 temp]$ cd vasp.6.4.2/src/

修改main.F代码
备份main.F

[userA@cu01 src]$ cp main.F main.F-org

使用vim把main.F的3654-3655行

      CALL CHAIN_FORCE(T_INFO%NIONS,DYN%POSION,TOTEN,TIFOR, &
           LATT_CUR%A,LATT_CUR%B,IO%IU6)

修改为

      CALL CHAIN_FORCE(T_INFO%NIONS,DYN%POSION,TOTEN,TIFOR, &
           TSIF,LATT_CUR%A,LATT_CUR%B,IO%IU6)

并且把969行

      IF (LCHAIN) CALL chain_init( T_INFO, IO)

修改为

      CALL chain_init( T_INFO, IO)

修改.objects

[userA@cu01 src]$ vi .objects

在126行chain.o前面加上

        bfgs.o \
        dynmat.o \
        instanton.o \
        lbfgs.o \
        sd.o \
        cg.o \
        dimer.o \
        bbm.o \
        fire.o \
        lanczos.o \
        neb.o qm.o \
        pyamff_fortran/*.o \
        ml_pyamff.o \
        opt.o \

修改src/makefile
把17行LIB=lib parser修改为LIB=lib parser pyamff_fortran

VASPsol溶剂化代码修改¶

参考VASP6.1.0以上版本中VASPsol的修改及编译方法修改
下载https://github.com/henniggroup/VASPsol/releases/tag/V1.0源代码VASPsol-1.0.tar.gz

备份solvation.F

[userA@cu01 src]$ mv solvation.F solvation.F-org

上传VASPsol-1.0.tar.gz压缩包src/solvation.F到/home/userA/temp/vasp.6.4.2/src目录下

修改mpi.F代码
备份mpi.F

[userA@cu01 src]$ cp mpi.F mpi.F-org

在2169行#endif // MPI_INPLACE后添加：

! Subroutine for VASPsol - Added by D. Gaissmaier
!----------------------------------------------------------------------
!
! Global sum function - required for VASPsol
!
!----------------------------------------------------------------------

      SUBROUTINE M_sum_1(COMM, v1)
      USE mpimy
      IMPLICIT NONE

      TYPE(communic) COMM
      REAL(q) vec(1),v1

      vec(1)=v1

      CALL M_sumb_d(COMM, vec, 1)

      v1=vec(1)

      RETURN
      END SUBROUTINE
! End of VASPsol subroutine

修改solvation.F代码
把2206行CALLMPI( M_sum_s(SP_GRIDC%COMM,1,Ecorr3,0,0,0))
替换为CALLMPI( M_sum_1(SP_GRIDC%COMM,Ecorr3))
把2217行CALLMPI( M_sum_s(SP_GRIDC%COMM,1,Ecorr4,0,0,0))
替换为CALLMPI( M_sum_1(SP_GRIDC%COMM,Ecorr4))

修改pot.F代码¶

备份pot.F

[userA@cu01 src]$ cp pot.F pot.F-org

在409行CALL POT_FLIP(CVTOT, GRIDC,WDES%NCDIJ )后面添加

! solvation__
! vaclevel
      IF (LSOL) CALL GET_FERMISHIFT(LATT_CUR, T_INFO, GRIDC, WDES, CWORK, CHTOT)
! vaclevel
! solvation__

VASP编译安装¶

编写makefile.include

[userA@cu01 src]$ cd ../
[userA@cu01 vasp.6.4.2]$ cat makefile.include 
# Default precompiler options
CPP_OPTIONS = -DHOST=\"LinuxIFC\" \
              -DMPI -DMPI_BLOCK=8000 -Duse_collective \
              -DscaLAPACK \
              -Dsol_compat \
              -DCACHE_SIZE=4000 \
              -Davoidalloc \
              -Dvasp6 \
              -Duse_bse_te \
              -Dtbdyn \
              -Dfock_dblbuf

CPP         = fpp -f_com=no -free -w0  $*$(FUFFIX) $*$(SUFFIX) $(CPP_OPTIONS)

FC          = mpiifort
FCL         = mpiifort

FREE        = -free -names lowercase

FFLAGS      = -assume byterecl -w -heap-arrays 64

OFLAG       = -O3
OFLAG_IN    = $(OFLAG)
DEBUG       = -O0

OBJECTS     = fftmpiw.o fftmpi_map.o fftw3d.o fft3dlib.o
OBJECTS_O1 += fftw3d.o fftmpi.o fftmpiw.o
OBJECTS_O2 += fft3dlib.o

# For what used to be vasp.5.lib
CPP_LIB     = $(CPP)
FC_LIB      = $(FC)
CC_LIB      = icc
CFLAGS_LIB  = -O
FFLAGS_LIB  = -O1
FREE_LIB    = $(FREE)

OBJECTS_LIB = linpack_double.o

# For the parser library
CXX_PARS    = icpc
LLIBS       = -lstdc++

##
## Customize as of this point! Of course you may change the preceding
## part of this file as well if you like, but it should rarely be
## necessary ...
##

# When compiling on the target machine itself, change this to the
# relevant target when cross-compiling for another architecture
VASP_TARGET_CPU ?= -xHOST
FFLAGS     += $(VASP_TARGET_CPU)

# Intel MKL (FFTW, BLAS, LAPACK, and scaLAPACK)
# (Note: for Intel Parallel Studio's MKL use -mkl instead of -qmkl)
FCL        += -qmkl=sequential
MKLROOT    ?= /opt/intel/oneapi/mkl/2022.2.0
LLIBS      += -L$(MKLROOT)/lib/intel64 -lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_lp64
INCS        =-I$(MKLROOT)/include/fftw

# HDF5-support (optional but strongly recommended)
CPP_OPTIONS+= -DVASP_HDF5
HDF5_ROOT  ?= /home/userA/soft/vasp/hdf5-1.12.0
LLIBS      += -L$(HDF5_ROOT)/lib -lhdf5_fortran
INCS       += -I$(HDF5_ROOT)/include


# For the VASP-2-Wannier90 interface (optional)
CPP_OPTIONS    += -DVASP2WANNIER90
WANNIER90_ROOT ?= /home/userA/temp/vasp.6.4.2/wannier90-3.1.0
LLIBS          += -L$(WANNIER90_ROOT) -lwannier

# for libxc
CPP_OPTIONS +=  -DUSELIBXC
LIBXCPATH=/home/userA/temp/vasp.6.4.2/libxc-6.2.2
INCS += -I$(LIBXCPATH)/include
LLIBS += $(LIBXCPATH)/lib/libxcf03.a $(LIBXCPATH)/lib/libxc.a

编译安装

[userA@cu01 vasp.6.4.2]$ source /opt/intel/oneapi/setvars.sh
[userA@cu01 vasp.6.4.2]$ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/soft/vasp/hdf5-1.12.0/lib:/opt/soft/vasp/zlib-1.3/lib
[userA@cu01 vasp.6.4.2]$ make |tee make.log

把生成的二进制可执行程序拷贝到/home/userA/soft/vasp下

[userA@cu01 vasp.6.4.2]$ exit
[userA@matrix temp]# mkdir -p /hme/userA/soft/vasp/
[userA@matrix temp]# cd /hme/userA/soft/vasp/
[userA@matrix vasp]# cp /home/userA/temp/vasp.6.4.2/bin/vasp_std vasp_std_642_avx512
[userA@matrix vasp]# cp /home/userA/temp/vasp.6.4.2/bin/vasp_gam vasp_gam_642_avx512
[userA@matrix vasp]# cp /home/userA/temp/vasp.6.4.2/bin/vasp_ncl vasp_ncl_642_avx512

VASP的提交脚本¶

[userA@matrix vasp]# cd /home/userA/test
[userA@matrix test]$ cat vasp6.slurm
#!/bin/bash
# Job submission script for SLURM:
# Usage: sbatch <this_script>
#
#SBATCH -J vasp6
#SBATCH -o vasp6.o%J
##SBATCH -e vasp6.e
#SBATCH -N 1
#SBATCH --ntasks-per-node=16
#SBATCH -t 1206:04:02
#SBATCH -p public
#############################################################################
#    -J: name of job
#    -o: file for batch script's standard output
#    -e: file for batch script's standard error
#    -N: number of nodes on which to run (N = min[-max])
#    -n: number of tasks to run
#    -t: time limit
#    -p: partition requested
#    -w: request a specific list of hosts
#############################################################################

# go to work dir
cd $SLURM_SUBMIT_DIR

# The program we want to execute
EXEC=/home/userA/soft/vasp/vasp_std_642_avx512

# module load intel oneapi
source /opt/intel/oneapi/compiler/2022.2.0/env/vars.sh
source /opt/intel/oneapi/mkl/2022.2.0/env/vars.sh
source /opt/intel/oneapi/mpi/2021.7.0/env/vars.sh
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/userA/soft/vasp/hdf5-1.12.0/lib:/home/userA/soft/vasp/szip-2.1.1/lib:/home/userA/soft/vasp/zlib-1.3/lib

echo "Starting VASP run at" `date`

# running program
time mpiexec.hydra -n $SLURM_NTASKS $EXEC > ${SLURM_SUBMIT_DIR}/run.log

echo "Finished VASP run at" `date`
echo "Work Dir is : $SLURM_SUBMIT_DIR"

参考¶

本文阅读量次
本站总访问量次