经过一个多星期的痛苦，终于将vasp编译通过了，期间经历的波折自不必说，在此总结一下VASP安装的详细过程，一来为将要或者还在为编译vasp痛苦的同志提供少许经验，二来为自己留个底，以后再编译的时候就简单了。

系统：64位ubuntu12.04

编译器：intel fortran和intel c++

数学库：intel MKL

并行软件：openmpi

傅里叶变换软件：fftw

1. 安装ubuntu12.04 详见

http://wenku.baidu.com/link?url=2C4ajX9AJdSbmBmLeSQ0U5qSjXIjh0N-Xd6nk5mYNQ3gu5XWd-JwnkYUQayhAuGgVYjNh_BlU_Lt2CH0hphmHkE8j6ktIzh9StARL1OS5-7

2. 打开终端输入：

sudo apt-get install build-essential

3. 安装编译器

(1) 从intel官网上申请下载非商业版本的编译器

https://software.intel.com/en-us/non-commercial-software-development

需要用邮箱申请一下，收到邮件后下载就行了

我现在用的版本是

l_ccompxe_2013_sp1.3.174.tgz和l_fcompxe_2013_sp1.3.174.tgz

(2) 解压缩

tar zxvf l_fcompxe_2013_sp1.3.174.tgz

tar zxvf l_ccompxe_2013_sp1.3.174.tgz

(3) 进入l_fcompxe_2013_sp1.3.174文件夹，输入

./install.sh

接下来按提示一步一步的来就行了

(4) 进入l_ccompxe_2013_sp1.3.174文件夹，输入

./install.sh

接下来按提示一步一步的来就行了

(5) 设置环境变量

在主文件夹下.bashrc中加入(用ctrl+h快捷键使得隐藏的文件显示出来)：

#intel compiler

source /opt/intel/composerxe/bin/compilervars.sh intel64

export PATH=/opt/intel/composer_xe_2013_sp1.3.174/bin/intel64/:$PATH

#intel mkl

export LD_LIBRARY_PATH=/opt/intel/composer_xe_2013_sp1.3.174/mkl/lib/intel64/:$LD_LIBRARY_PATH

(6) 测试

终端中输入which ifort和which icc看能否给出路径

4. 安装openmpi

(1) 解压缩openmpi-1.8.1.gz

tar zxvf openmpi-1.8.1.gz

(2) 进入openmpi-1.8.1文件夹，输入

./configure CC=icc CXX=icpc F77=ifort FC=ifort --prefix=/home/zx/programs/openmpi

make

make install

(3) 设置环境变量

在主文件夹下.bashrc中加入:

#openmpi

 

if [ -f /opt/intel/composerxe/bin/compilervars.sh ]

 then

  source /opt/intel/composerxe/bin/compilervars.sh intel64 2> /dev/null

  export MPI=/home/zx/programs/openmpi

  export PATH=$MPI/bin:$PATH

  export LD_LIBRARY_PATH=$MPI/lib:$LD_LIBRARY_PATH

  export MANPATH=$MANPATH:$MPI/share/man

  export INFOPATH=$INFOPATH:$MPI/share/man

fi

5. 安装fftw

(1) 解压缩fftw-3.3.4.gz

tar zxvf fftw-3.3.4.gz

(2) 打开fftw-3.3.4文件夹，在终端中输入：

./configure --prefix=/home/zx/programs/fftw GCC=ifort F77=ifort --enable-mpi

make

make install

6. 安装VASP库

(1) 解压缩vasp.5.lib.tar.gz

tar zxvf vasp.5.lib.tar.gz

(2) 进入vasp.5.lib文件夹

打开makefile.linux_ifc_P4，将FC=ifc修改为FC=ifort

(3) 在终端中输入

make -f makefile.linux_ifc_P4

得到libdmy.a

7. VASP并行编译

(1) 解压缩vasp.5.2.12.tar.gz

tar zxvf vasp.5.2.12.tar.gz

(2) 进入vasp.5.2文件夹，修改makefile.linux_ifc_P4（文章最后是我的makefile.linux_ifc_P4，修改的地方用红色标出了）

(3) 修改完后在终端输入make -f makefile.linux_ifc_P4即可以得到vasp的可执行文件

(4) 把vasp可执行文件考到计算的文件夹中，输入

mpirun -n 8 ./vasp

即可以算了，如果出现段错误，可以在.bashrc中加入ulimit -s unlimited

 

下面是我的makefile.linux_ifc_P4

# all CPP processed fortran files have the extension .f90

SUFFIX=.f90

 

#-----------------------------------------------------------------------

# fortran compiler and linker

#-----------------------------------------------------------------------

FC=ifort

# fortran linker

FCL=$(FC)

 

 

#-----------------------------------------------------------------------

# whereis CPP ?? (I need CPP, can't use gcc with proper options)

# that's the location of gcc for SUSE 5.3

#

#  CPP_   =  /usr/lib/gcc-lib/i486-linux/2.7.2/cpp -P -C

#

# that's probably the right line for some Red Hat distribution:

#

#  CPP_   =  /usr/lib/gcc-lib/i386-redhat-linux/2.7.2.3/cpp -P -C

#

#  SUSE X.X, maybe some Red Hat distributions:

 

CPP_ =  ./preprocess <$*.F | /usr/bin/cpp -P -C -traditional >$*$(SUFFIX)

 

#-----------------------------------------------------------------------

# possible options for CPP:

# NGXhalf             charge density   reduced in X direction

# wNGXhalf            gamma point only reduced in X direction

# avoidalloc          avoid ALLOCATE if possible

# PGF90               work around some for some PGF90 / IFC bugs

# CACHE_SIZE          1000 for PII,PIII, 5000 for Athlon, 8000-12000 P4, PD

# RPROMU_DGEMV        use DGEMV instead of DGEMM in RPRO (depends on used BLAS)

# RACCMU_DGEMV        use DGEMV instead of DGEMM in RACC (depends on used BLAS)

# tbdyn                 MD package of Tomas  Bucko

#-----------------------------------------------------------------------

 

CPP     = $(CPP_)  -DHOST="LinuxIFC" \

          -DCACHE_SIZE=12000 -DPGF90 -Davoidalloc -DNGXhalf \

#          -DRPROMU_DGEMV  -DRACCMU_DGEMV

 

#-----------------------------------------------------------------------

# general fortran flags  (there must a trailing blank on this line)

# byterecl is strictly required for ifc, since otherwise

# the WAVECAR file becomes huge

#-----------------------------------------------------------------------

 

FFLAGS =  -FR -lowercase -assume byterecl

 

#-----------------------------------------------------------------------

# optimization

# we have tested whether higher optimisation improves performance

# -axK  SSE1 optimization,  but also generate code executable on all mach.

#       xK improves performance somewhat on XP, and a is required in order

#       to run the code on older Athlons as well

# -xW   SSE2 optimization

# -axW  SSE2 optimization,  but also generate code executable on all mach.

# -tpp6 P3 optimization

# -tpp7 P4 optimization

#-----------------------------------------------------------------------

 

# ifc.9.1, ifc.10.1 recommended

OFLAG=-O2 -ip -ftz

 

OFLAG_HIGH = $(OFLAG)

OBJ_HIGH =

OBJ_NOOPT =

DEBUG  = -FR -O0

INLINE = $(OFLAG)

 

#-----------------------------------------------------------------------

# the following lines specify the position of BLAS  and LAPACK

# VASP works fastest with the libgoto library

# so that's what we recommend

#-----------------------------------------------------------------------

 

# mkl.10.0

# set -DRPROMU_DGEMV  -DRACCMU_DGEMV in the CPP lines

#BLAS=-L/opt/intel/mkl100/lib/em64t -lmkl -lpthread

BLAS=-L/opt/intel/composer_xe_2013_sp1.3.174/mkl/lib/intel64 -lmkl_intel_lp64 -lmkl_core -lmkl_sequential -lpthread

 

# even faster for VASP Kazushige Goto's BLAS

# http://www.cs.utexas.edu/users/kgoto/signup_first.html

# parallel goto version requires sometimes -libverbs

#BLAS=  /opt/libs/libgoto/libgoto.so

 

# LAPACK, simplest use vasp.5.lib/lapack_double

#LAPACK= ../vasp.5.lib/lapack_double.o

 

# use the mkl Intel lapack

LAPACK=-L/opt/intel/composer_xe_2013_sp1.3.174/mkl/lib/intel64 -lmkl_intel_lp64 -lmkl_core -lmkl_sequential -lpthread

 

#-----------------------------------------------------------------------

 

LIB  = -L../vasp.5.lib -ldmy \

     ../vasp.5.lib/linpack_double.o $(LAPACK) \

     $(BLAS)

 

# options for linking, nothing is required (usually)

LINK    =

 

#-----------------------------------------------------------------------

# fft libraries:

# VASP.5.2 can use fftw.3.1.X (http://www.fftw.org)

# since this version is faster on P4 machines, we recommend to use it

#-----------------------------------------------------------------------

 

#FFT3D   = fft3dfurth.o fft3dlib.o

 

# alternatively: fftw.3.1.X is slighly faster and should be used if available

FFT3D   = fftw3d.o fft3dlib.o   /home/zx/programs/fftw/lib/libfftw3.a

 

 

#=======================================================================

# MPI section, uncomment the following lines until

#    general  rules and compile lines

# presently we recommend OPENMPI, since it seems to offer better

# performance than lam or mpich

#

# !!! Please do not send me any queries on how to install MPI, I will

# certainly not answer them !!!!

#=======================================================================

#-----------------------------------------------------------------------

# fortran linker for mpi

#-----------------------------------------------------------------------

 

FC=mpif90

FCL=$(FC)

 

#-----------------------------------------------------------------------

# additional options for CPP in parallel version (see also above):

# NGZhalf               charge density   reduced in Z direction

# wNGZhalf              gamma point only reduced in Z direction

# scaLAPACK             use scaLAPACK (usually slower on 100 Mbit Net)

# avoidalloc          avoid ALLOCATE if possible

# PGF90               work around some for some PGF90 / IFC bugs

# CACHE_SIZE          1000 for PII,PIII, 5000 for Athlon, 8000-12000 P4, PD

# RPROMU_DGEMV        use DGEMV instead of DGEMM in RPRO (depends on used BLAS)

# RACCMU_DGEMV        use DGEMV instead of DGEMM in RACC (depends on used BLAS)

# tbdyn                 MD package of Tomas  Bucko

#-----------------------------------------------------------------------

 

#-----------------------------------------------------------------------

 

CPP    = $(CPP_) -DMPI  -DHOST="LinuxIFC" -DIFC \

     -DCACHE_SIZE=4000 -DPGF90 -Davoidalloc -DNGZhalf \

     -DMPI_BLOCK=8000

##    -DRPROMU_DGEMV  -DRACCMU_DGEMV

 

#-----------------------------------------------------------------------

# location of SCALAPACK

# if you do not use SCALAPACK simply leave that section commented out

#-----------------------------------------------------------------------

 

BLACS=$(HOME)/archives/SCALAPACK/BLACS/

SCA_=$(HOME)/archives/SCALAPACK/SCALAPACK

 

SCA= $(SCA_)/libscalapack.a  \

 $(BLACS)/LIB/blacsF77init_MPI-LINUX-0.a $(BLACS)/LIB/blacs_MPI-LINUX-0.a $(BLACS)/LIB/blacsF77init_MPI-LINUX-0.a

 

SCA=

 

#-----------------------------------------------------------------------

# libraries for mpi

#-----------------------------------------------------------------------

 

LIB     = -limf -lm -L../vasp.5.lib -ldmy  \

      ../vasp.5.lib/linpack_double.o $(LAPACK) \

      $(SCA) $(BLAS)

 

# FFT: fftmpi.o with fft3dlib of Juergen Furthmueller

#FFT3D   = fftmpi.o fftmpi_map.o fft3dfurth.o fft3dlib.o

 

# alternatively: fftw.3.1.X is slighly faster and should be used if available

FFT3D   = fftmpi.o fftmpi_map.o fftw3d.o fft3dlib.o  /home/zx/programs/fftw/lib/libfftw3.a

 

#-----------------------------------------------------------------------

# general rules and compile lines

#-----------------------------------------------------------------------

BASIC=   symmetry.o symlib.o   lattlib.o  random.o  

 

 

SOURCE=  base.o     mpi.o      smart_allocate.o      xml.o  \

         constant.o jacobi.o   main_mpi.o  scala.o   \

         asa.o      lattice.o  poscar.o   ini.o  mgrid.o  xclib.o  vdw_nl.o  xclib_grad.o \

         radial.o   pseudo.o   gridq.o     ebs.o  \

         mkpoints.o wave.o     wave_mpi.o  wave_high.o  \

         $(BASIC)   nonl.o     nonlr.o    nonl_high.o dfast.o    choleski2.o \

         mix.o      hamil.o    xcgrad.o   xcspin.o    potex1.o   potex2.o  \

         constrmag.o cl_shift.o relativistic.o LDApU.o \

         paw_base.o metagga.o  egrad.o    pawsym.o   pawfock.o  pawlhf.o   rhfatm.o  paw.o   \

         mkpoints_full.o       charge.o   Lebedev-Laikov.o  stockholder.o dipol.o    pot.o \

         dos.o      elf.o      tet.o      tetweight.o hamil_rot.o \

         steep.o    chain.o    dyna.o     sphpro.o    us.o  core_rel.o \

         aedens.o   wavpre.o   wavpre_noio.o broyden.o \

         dynbr.o    rmm-diis.o reader.o   writer.o   tutor.o xml_writer.o \

         brent.o    stufak.o   fileio.o   opergrid.o stepver.o  \

         chgloc.o   fast_aug.o fock.o     mkpoints_change.o sym_grad.o \

         mymath.o   internals.o dynconstr.o dimer_heyden.o dvvtrajectory.o vdwforcefield.o \

         hamil_high.o nmr.o    pead.o     mlwf.o     subrot.o   subrot_scf.o \

         force.o    pwlhf.o  gw_model.o optreal.o   davidson.o  david_inner.o \

         electron.o rot.o  electron_all.o shm.o    pardens.o  paircorrection.o \

         optics.o   constr_cell_relax.o   stm.o    finite_diff.o elpol.o    \

         hamil_lr.o rmm-diis_lr.o  subrot_cluster.o subrot_lr.o \

         lr_helper.o hamil_lrf.o   elinear_response.o ilinear_response.o \

         linear_optics.o linear_response.o   \

         setlocalpp.o  wannier.o electron_OEP.o electron_lhf.o twoelectron4o.o \

         ratpol.o screened_2e.o wave_cacher.o chi_base.o wpot.o local_field.o \

         ump2.o bse_te.o bse.o acfdt.o chi.o sydmat.o dmft.o \

         rmm-diis_mlr.o  linear_response_NMR.o

 

vasp: $(SOURCE) $(FFT3D) $(INC) main.o

       rm -f vasp

       $(FCL) -o vasp main.o  $(SOURCE)   $(FFT3D) $(LIB) $(LINK)

makeparam: $(SOURCE) $(FFT3D) makeparam.o main.F $(INC)

       $(FCL) -o makeparam  $(LINK) makeparam.o $(SOURCE) $(FFT3D) $(LIB)

zgemmtest: zgemmtest.o base.o random.o $(INC)

       $(FCL) -o zgemmtest $(LINK) zgemmtest.o random.o base.o $(LIB)

dgemmtest: dgemmtest.o base.o random.o $(INC)

       $(FCL) -o dgemmtest $(LINK) dgemmtest.o random.o base.o $(LIB)

ffttest: base.o smart_allocate.o mpi.o mgrid.o random.o ffttest.o $(FFT3D) $(INC)

       $(FCL) -o ffttest $(LINK) ffttest.o mpi.o mgrid.o random.o smart_allocate.o base.o $(FFT3D) $(LIB)

kpoints: $(SOURCE) $(FFT3D) makekpoints.o main.F $(INC)

       $(FCL) -o kpoints $(LINK) makekpoints.o $(SOURCE) $(FFT3D) $(LIB)

 

clean:    

       -rm -f *.g *.f *.o *.L *.mod ; touch *.F

 

main.o: main$(SUFFIX)

       $(FC) $(FFLAGS)$(DEBUG)  $(INCS) -c main$(SUFFIX)

xcgrad.o: xcgrad$(SUFFIX)

       $(FC) $(FFLAGS) $(INLINE)  $(INCS) -c xcgrad$(SUFFIX)

xcspin.o: xcspin$(SUFFIX)

       $(FC) $(FFLAGS) $(INLINE)  $(INCS) -c xcspin$(SUFFIX)

 

makeparam.o: makeparam$(SUFFIX)

       $(FC) $(FFLAGS)$(DEBUG)  $(INCS) -c makeparam$(SUFFIX)

 

makeparam$(SUFFIX): makeparam.F main.F

#

# MIND: I do not have a full dependency list for the include

# and MODULES: here are only the minimal basic dependencies

# if one strucuture is changed then touch_dep must be called

# with the corresponding name of the structure

#

base.o: base.inc base.F

mgrid.o: mgrid.inc mgrid.F

constant.o: constant.inc constant.F

lattice.o: lattice.inc lattice.F

setex.o: setexm.inc setex.F

pseudo.o: pseudo.inc pseudo.F

poscar.o: poscar.inc poscar.F

mkpoints.o: mkpoints.inc mkpoints.F

wave.o: wave.F

nonl.o: nonl.inc nonl.F

nonlr.o: nonlr.inc nonlr.F

 

$(OBJ_HIGH):

       $(CPP)

       $(FC) $(FFLAGS) $(OFLAG_HIGH) $(INCS) -c $*$(SUFFIX)

$(OBJ_NOOPT):

       $(CPP)

       $(FC) $(FFLAGS) $(INCS) -c $*$(SUFFIX)

 

fft3dlib_f77.o: fft3dlib_f77.F

       $(CPP)

       $(F77) $(FFLAGS_F77) -c $*$(SUFFIX)

 

.F.o:

       $(CPP)

       $(FC) $(FFLAGS) $(OFLAG) $(INCS) -c $*$(SUFFIX)

.F$(SUFFIX):

       $(CPP)

$(SUFFIX).o:

       $(FC) $(FFLAGS) $(OFLAG) $(INCS) -c $*$(SUFFIX)

 

# special rules

#-----------------------------------------------------------------------

# these special rules are cummulative (that is once failed

#   in one compiler version, stays in the list forever)

# -tpp5|6|7 P, PII-PIII, PIV

# -xW use SIMD (does not pay of on PII, since fft3d uses double prec)

# all other options do no affect the code performance since -O1 is used

 

fft3dlib.o : fft3dlib.F

       $(CPP)

       $(FC) -FR -lowercase -O2 -c $*$(SUFFIX)

 

fft3dfurth.o : fft3dfurth.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

fftw3d.o : fftw3d.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

wave_high.o : wave_high.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

radial.o : radial.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

symlib.o : symlib.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

symmetry.o : symmetry.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

wave_mpi.o : wave_mpi.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

wave.o : wave.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

dynbr.o : dynbr.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

asa.o : asa.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

broyden.o : broyden.F

       $(CPP)

       $(FC) -FR -lowercase -O2 -c $*$(SUFFIX)

 

us.o : us.F

       $(CPP)

       $(FC) -FR -lowercase -O1 -c $*$(SUFFIX)

 

LDApU.o : LDApU.F

       $(CPP)

       $(FC) -FR -lowercase -O2 -c $*$(SUFFIX)