Some problem during compiling cp2k-4.1 using intel compiler

Alfio Lazzaro alfio.... at gmail.com
Mon Apr 9 07:40:26 UTC 2018


In principle, you can add

DFLAGS += -D__HAS_NO_MPI_MOD

in your arch file. 

Anyway, it can be that there is something wrong with the MPI installation 
on your system...

Alfio


Il giorno domenica 8 aprile 2018 15:49:45 UTC+2, Tianshu Jiang in Beijing 
ha scritto:
>
> Hello Alfio,
> I did as you suggested, but there is still error. 
>
> USE mpi  ! compiler *errors* mean mpi installation and fortran compiler 
> misma
>       1
> Fatal Error: Can't open module file 'mpi.mod' for reading at (1): No such 
> file or directory
> make[3]: *** [message_passing.o] Error 1
> make[3]: *** Waiting for unfinished jobs....
> ar: creating 
> /share/home/tshjiang/cp2k-4.1/lib/Linux-x86-64-gfortran/sopt/libdbcsrerr.a
> make[2]: *** [all] Error 2
> make[1]: *** [sopt] Error 2
> make: *** [all] Error 2
>
> It seems that the library installed in our cluster did't have such a file.
> I use the path set in the system. But the file still cannot be found.
> Where may be the problem?
>
>
>
> 在 2018年4月8日星期日 UTC+8下午3:20:57,Alfio Lazzaro写道:
>>
>> I'm confused... You are not using Intel compiler, you are using GNU?
>> Anyway, you should use the MPI wrapper for the compilation, i.e.
>>
>> FC = mpif90
>> LD = mpif90
>>
>> Alfio
>>
>>
>> Il giorno sabato 7 aprile 2018 13:52:30 UTC+2, Tianshu Jiang in Beijing 
>> ha scritto:
>>>
>>> Hi, Alfio, 
>>>
>>> This error 
>>>
>>> ld: skipping incompatible /share/soft/intel_2017/
>>> compilers_and_libraries_2017.1.132/linux/mpi/mic/lib/libmpifort.a when 
>>> searching for -lmpifort
>>> ld: cannot find -lmpifort
>>> ld: skipping incompatible /share/soft/intel_2017/
>>> compilers_and_libraries_2017.1.132/linux/mpi/mic/lib/debug_mt/libmpi.so 
>>> when searching for -lmpi
>>> ld: skipping incompatible /share/soft/intel_2017/
>>> compilers_and_libraries_2017.1.132/linux/mpi/mic/lib/debug_mt/libmpi.a 
>>> when searching for -lmpi
>>> ld: skipping incompatible /share/soft/intel_2017/
>>> compilers_and_libraries_2017.1.132/linux/mpi/mic/lib/libmpi.so when 
>>> searching for -lmpi
>>> ld: skipping incompatible /share/soft/intel_2017/
>>> compilers_and_libraries_2017.1.132/linux/mpi/mic/lib/libmpi.a when 
>>> searching for -lmpi
>>> ld: cannot find -lmpi
>>> ld: skipping incompatible /share/soft/intel_2017/
>>> compilers_and_libraries_2017.1.132/linux/mpi/mic/lib/libmpigi.a when 
>>> searching for -lmpigi
>>> ld: cannot find -lmpigi
>>> make[3]: *** [/share/home/tshjiang/intel_cp2k/exe/Linux-x86-64-gfortran/
>>> graph.psmp] Error 1
>>> make[3]: *** Waiting for unfinished jobs....
>>> make[2]: *** [all] Error 2
>>> make[1]: *** [psmp] Error 2
>>> make: *** [all] Error 2
>>>
>>>
>>> is related the version of Intel compiler ? The mismatching version of 
>>> cp2k and Intel compiler lead  to the final incompatible error?
>>> By the way, if I make cp2k using GNU's compiler, my arch file is as 
>>> follows
>>>
>>> # Tested with: GFortran 4.9.2, MPICH 3.1, LAPACK 3.5.0, ScaLAPACK 2.0.2
>>> CC         = /soft/gcc-6.4.0/bin/gcc
>>> CPP        =
>>> FC         = gfortran
>>> LD         = gfortran
>>> AR         = ar -r
>>> ACML_INC   = /share/home/tshjiang/cp2k-4.1/tools/toolchain/install/
>>> acml_install/gfortran64/include
>>> ACML_LIB   = /share/home/tshjiang/cp2k-4.1/tools/toolchain/install/
>>> acml_install/gfortran64/lib
>>> FFTW_INC   = /share/home/tshjiang/cp2k-4.1/tools/toolchain/install/fftw-
>>> 3.3.4/include
>>> FFTW_LIB   = /share/home/tshjiang/cp2k-4.1/tools/toolchain/install/fftw-
>>> 3.3.4/lib
>>> LIBINT_INC = /share/home/tshjiang/cp2k-4.1/tools/toolchain/install/
>>> libint-1.1.4/include
>>> LIBINT_LIB = /share/home/tshjiang/cp2k-4.1/tools/toolchain/install/
>>> libint-1.1.4/lib
>>> LIBXC_INC  = /share/home/tshjiang/cp2k-4.1/tools/toolchain/install/libxc
>>> /include
>>> LIBXC_LIB  = /share/home/tshjiang/cp2k-4.1/tools/toolchain/install/libxc
>>> /lib
>>> DFLAGS     = -D__FFTW3 -D__LIBINT -D__LIBXC2 -D__MPI_VERSION=3\
>>>              -D__LIBINT_MAX_AM=7 -D__LIBDERIV_MAX_AM1=6 -D__MAX_CONTR=4\
>>>              -D__parallel -D__SCALAPACK
>>> CPPFLAGS   =
>>> FCFLAGS    = $(DFLAGS) -O2 -ffast-math -ffree-form -ffree-line-length-
>>> none\
>>>              -fopenmp -ftree-vectorize -funroll-loops\
>>>              -mtune=native\
>>>              -I$(ACML_INC) -I$(FFTW_INC) -I$(LIBINT_INC) -I$(LIBXC_INC)
>>> LDFLAGS    = $(FCFLAGS) -static-libgfortran
>>> LIBS       = /share/home/tshjiang/cp2k-4.1/tools/toolchain/install/
>>> libscalapack.a\
>>>              /share/home/tshjiang/cp2k-4.1/tools/toolchain/install/acml-
>>> 5.3.1/gfortran64/lib/libacml.a\
>>>              $(FFTW_LIB)/libfftw3.a\
>>>              $(FFTW_LIB)/libfftw3_threads.a\
>>>              $(LIBXC_LIB)/libxcf90.a\
>>>              $(LIBXC_LIB)/libxc.a\
>>>
>>>              $(LIBINT_LIB)/libderiv.a\
>>>              $(LIBINT_LIB)/libint.a
>>>
>>> The version of my GNU compiler is 6.4.0.
>>> Does there exists some obvious error in this arch file ?
>>> Thanks for your patience!
>>>
>>> 在 2018年4月7日星期六 UTC+8下午2:06:17,Alfio Lazzaro写道:
>>>>
>>>> Note that Intel 17.0.1 is not supported (see 
>>>> https://www.cp2k.org/dev:compiler_support ). Make sure that your CP2K 
>>>> installation goes through the regtest (
>>>> https://www.cp2k.org/dev:regtesting ).
>>>>
>>>> Alfio
>>>>
>>>>
>>>>
>>>> Il giorno venerdì 6 aprile 2018 14:11:57 UTC+2, Tianshu Jiang in 
>>>> Beijing ha scritto:
>>>>>
>>>>> Dear all, 
>>>>>
>>>>> I have been compiling cp2k-4.1 using intel compiler for several days 
>>>>> but there is always problems.
>>>>> I edit both Linux-x86-64-gfortran.psmp and Linux-x86-64-intel-mic.psmp 
>>>>> file according to the tutorial searched from the internet.
>>>>> But in my latest trial, the beginning is OK, then the error emerge. 
>>>>> The compiling environment is on our cluster, the version of Intel 
>>>>> compiler is 17.0.1, libint 1.1.4 and libxc 2.2.2.
>>>>> The error information is as follows
>>>>>
>>>>>
>>>>> mpiifort -c -D__MKL -D__FFTW3 -D__LIBINT -D__LIBXC2 -D__MPI_VERSION=3 
>>>>> -D__LIBINT_MAX_AM=7 -D__LIBDERIV_MAX_AM1=6 -D__MAX_CONTR=4 -D__parallel 
>>>>> -D__SCALAPACK -O2 -g -traceback -fpp -free -I/soft/intel_2017/
>>>>> compilers_and_libraries_2017/linux/mkl/include -I/soft/intel_2017/
>>>>> compilers_and_libraries_2017/linux/mkl/include/fftw -I/share/home/
>>>>> tshjiang/intel_cp2k/tools/toolchain/install/libint-1.1.4/include -I/
>>>>> share/home/tshjiang/intel_cp2k/tools/toolchain/install/libxc-2.2.2/include 
>>>>> -D__COMPILE_ARCH="\"Linux-x86-64-gfortran\"" -D__COMPILE_DATE="\"Fri 
>>>>> Apr  6 19:45:36 CST 2018\"" -D__COMPILE_HOST="\"ln02\"" -
>>>>> D__COMPILE_REVISION="\"svn:17462\"" -D__DATA_DIR=
>>>>> "\"/share/home/tshjiang/intel_cp2k/data\"" -D__SHORT_FILE__=
>>>>> "\"eri_mme/eri_mme_lattice_summation.F\"" /share/home/tshjiang/
>>>>> intel_cp2k/src/eri_mme/eri_mme_lattice_summation.F
>>>>> ld: group ended before it began (--help for usage)
>>>>> make[3]: *** [/share/home/tshjiang/intel_cp2k/exe/Linux-x86-64-
>>>>> gfortran/graph.psmp] Error 1
>>>>> make[3]: *** Waiting for unfinished jobs....
>>>>> make[2]: *** [all] Error 2
>>>>> make[1]: *** [psmp] Error 2
>>>>> make: *** [all] Error 
>>>>>
>>>>>
>>>>>
>>>>> I even can not figure out where is the error. The command of make is:
>>>>>
>>>>> make -j 8 ARCH=Linux-x86-64-gfortran VERSION=psmp
>>>>>
>>>>> the arch file looks like
>>>>>
>>>>> # Tested with: GFortran 4.9.2, MPICH 3.1, LAPACK 3.5.0, ScaLAPACK 2.0.2
>>>>> CC         = icc
>>>>> CPP        =
>>>>> FC         = mpiifort
>>>>> LD         = mpiifort
>>>>> AR         = ar -r
>>>>>
>>>>> LIBINT_INC = /share/home/tshjiang/intel_cp2k/tools/toolchain/install/
>>>>> libint-1.1.4/include
>>>>>
>>>>> LIBINT_LIB = /share/home/tshjiang/intel_cp2k/tools/toolchain/install/
>>>>> libint-1.1.4/lib
>>>>> LIBXC_INC  = /share/home/tshjiang/intel_cp2k/tools/toolchain/install/
>>>>> libxc-2.2.2/include
>>>>> LIBXC_LIB  = /share/home/tshjiang/intel_cp2k/tools/toolchain/install/
>>>>> libxc-2.2.2/lib
>>>>> DFLAGS     = -D__MKL -D__FFTW3 -D__LIBINT -D__LIBXC2 -D__MPI_VERSION=3
>>>>> \
>>>>>              -D__LIBINT_MAX_AM=7 -D__LIBDERIV_MAX_AM1=6 -D__MAX_CONTR=
>>>>> 4\
>>>>>              -D__parallel -D__SCALAPACK
>>>>> CFLAGS     = $(DFLAGS)
>>>>> CPPFLAGS   =
>>>>> MKLROOT    = /soft/intel_2017/compilers_and_libraries_2017/linux/mkl
>>>>> FCFLAGS    = $(DFLAGS) -O2 -g -traceback -fpp -free \
>>>>>              -I$(MKLROOT)/include -I$(MKLROOT)/include/fftw -I$(
>>>>> LIBINT_INC) -I$(LIBXC_INC)
>>>>> FCFLAGS2   = $(DFLAGS) -O0 -g -traceback -fpp -free \
>>>>>              -I$(MKLROOT)/include -I$(MKLROOT)/include/fftw -I$(
>>>>> LIBINT_INC) -I$(LIBXC_INC)
>>>>> LDFLAGS    = $(FCFLAGS) -static-intel
>>>>> LDFLAGS_C  = $(FCFLAGS) -static-intel -nofor_main
>>>>> LIBS       = $(MKLROOT)/lib/intel64/libmkl_scalapack_lp64.a \
>>>>>              -WI,--start-group $(MKLROOT)/lib/intel64/
>>>>> libmkl_intel_lp64.a \
>>>>>              $(MKLROOT)/lib/intel64/libmkl_sequential.a \
>>>>>              $(MKLROOT)/lib/intel64/libmkl_core.a \
>>>>>              $(MKLROOT)/lib/intel64/libmkl_blacs_intelmpi_lp64.a -Wl
>>>>> ,--end-group \
>>>>>              -lpthread -lm \
>>>>>              -L$(LIBXC_LIB) -lxcf90 -lxc \
>>>>>              $(LIBINT_LIB)/libderiv.a \
>>>>>              $(LIBINT_LIB)/libint.a -lstdc++
>>>>> mp2_optimize_ri_basis.o:mp2_optimize_ri_basis.F
>>>>>         $(FC) -c $(FCFLAGS2) $<
>>>>> hfx_contraction_methods.o:hfx_contraction_methods.F
>>>>>         $(FC) -c $(FCFLAGS2) $<
>>>>>
>>>>> Can anyone encounter the similar situation and give some advice. 
>>>>> I will be appreciated for your suggestion.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180409/6bc3cd23/attachment.htm>


More information about the CP2K-user mailing list