[Wien] Wien2k parallel compilation
Sergio Yanuen Rodriguez
srodriguez at physics.tamu.edu
Thu Aug 7 21:40:47 CEST 2008
Dear Laurence
do you still know what were your settings when you were using openmpi? I
am still getting the same errors, maybe I am missing some of the flags or
something. I already recompile openmpi with gcc+ifort but it didn't work.
Thank you
Sergio Y. Rodriguez
Physics Department
Texas A&M University
> I don't think just changing mpif90 to ifort from gfortran (which might
> be what you did) will work, you probably need to compile with ifort.
>
> For reference, I believe that for whatever reason you are not finding
> the mpi_init etc routines in openmpi. This probably has to do with the
> wrong number of "_" added at the end of the subroutine names. I have
> used mpich1 in the past compiled with gcc+ifort (icc+ifort is
> problematic) and currently use mvapich which seems to be better.
>
> On Fri, Aug 1, 2008 at 6:25 PM, Sergio Yanuen Rodriguez
> <srodriguez at physics.tamu.edu> wrote:
>> Dear Laurence
>>
>> Thanks for your suggestion the mpif90 was configured for gfortran
>> instead
>> of ifort, I already fixed that, but the same errors keep appearing.
>>
>> Also I want to ask you what are you using mpich1? Maybe I need to
>> switch
>> from openmpi to mpich.
>>
>> Thanks for your help
>>
>> Sergio Y. Rodriguez
>> Physics Department
>> Texas A&M Unversity
>>
>>> You have "some" incompatibiltiy between your openmpi libraries and
>>> your compilation. A suggestion, which may (or not) help. Your openmpi
>>> library is -I/usr/include/openmpi/1.2.4-gcc which implies that it was
>>> compiled with gcc not ifort. What mpif90 are you using, one that calls
>>> ifort or gfortran?
>>>
>>> The options I use are (not openmpi though)
>>> FOPT = -FR -opt-mem-bandwidth2 -mp1 -prec_div -r8 -pc80 -fpconstant
>>> -traceback -pad -align -O3 -ipo1 -xT -DINTEL_VML -thread -i-static
>>> -fminshared
>>> FPOPT = -FR -opt-mem-bandwidth2 -mp1 -prec_div -r8 -pc80 -fpconstant
>>> -traceback -pad -align -O3 -ipo1 -xT -DINTEL_VML -i-static -fminshared
>>> DParallel = '-DParallel'
>>> FGEN = $(PARALLEL)
>>> LDFLAGS = $(FOPT) -L/opt/intel/mkl/10.0.2.018/lib/em64t
>>> LDFLAGS = -L/opt/intel/mkl/10.0.3.020/lib/em64t
>>> R_LIBS = $(FOPT) -lmkl_lapack -lmkl_em64t -lmkl_core -liomp5 -pthread
>>> -static
>>> C_LIBS = $(R_LIBS)
>>> RP_LIBS = $(FPOPT) -lmkl_intel_lp64 -lmkl_scalapack_lp64
>>> -lmkl_blacs_lp64 -lmkl_sequential -lmkl_core -liomp5 -pthread
>>> -i-static
>>>
>>>
>>> On Fri, Aug 1, 2008 at 1:41 PM, Sergio Yanuen Rodriguez
>>> <srodriguez at physics.tamu.edu> wrote:
>>>> Dear wien2k users:
>>>>
>>>> I sent a message a few days ago but nobody has responded so far, I am
>>>> trying to compile wien2k in parallel on a intel core quad processor
>>>> with
>>>> fedora 8 and 8 GB in RAM but I am getting some errors. I am able to
>>>> install and run the serial version. In my earlier email I was mixing
>>>> the
>>>> mkl libraries versions 9.X and 10.0.X but I already fix that
>>>>
>>>> My software is:
>>>>
>>>> Kernel version 2.6.25
>>>> gcc version 4.1.2
>>>> Intel Fortran compiler 10.1.015
>>>> Intel MKL 10.0.1.014
>>>> OpenMPI 1.2.4
>>>> WIEN2k_08.2
>>>>
>>>> The compilation options are:
>>>> Current settings:
>>>> O Compiler options: -FR -mp1 -w -prec_div -pc80 -pad -ip
>>>> -DINTEL_VML -O3 -xW
>>>> L Linker Flags: $(FOPT)
>>>> -L/opt/intel/mkl/10.0.1.014/lib/em64t
>>>> -lpthread
>>>> P Preprocessor flags '-DParallel'
>>>> R R_LIB (LAPACK+BLAS): -L/opt/intel/mkl/10.0.1.014/lib/em64t
>>>> -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -lguide -lpthread
>>>>
>>>> Current settings:
>>>> RP RP_LIB(SCALAPACK+PBLAS): -L
>>>> /opt/intel/mkl/10.0.1.014/lib/em64t
>>>> -L/usr/lib64/openmpi/1.2.4-gcc -lmkl_intel_lp64 -lmkl_scalapack_lp64
>>>> -lmkl_blacs_lp64 -lmpi_f90 -lmkl_lapack
>>>> FP FPOPT(par.comp.options): -FR -mp1 -w -prec_div -pc80 -pad -ip
>>>> -DINTEL_VML -traceback -I/usr/include/openmpi/1.2.4-gcc
>>>> -I/usr/include/openmpi/1.2.4-gcc/64
>>>> MP MPIRUN commando : mpirun -np _NP_ -machinefile
>>>>
>>>> I am trying to compile lapw0 first and then when this is alright I
>>>> will
>>>> try to compile the whole program, lapw0 compiles something but then
>>>> some
>>>> errors appear, the errors I am getting are:
>>>>
>>>> modules.o: In function `parallel_mp_begend_':
>>>> modules.F:(.text+0x328): undefined reference to `mpi_comm_split_'
>>>> modules.o: In function `parallel_mp_init_parallel_':
>>>> modules.F:(.text+0x447): undefined reference to `mpi_init_'
>>>> modules.F:(.text+0x46b): undefined reference to `mpi_comm_size_'
>>>> modules.F:(.text+0x48f): undefined reference to `mpi_comm_rank_'
>>>> energy.o: In function `energy_':
>>>> energy.F:(.text+0x46db): undefined reference to `mpi_allreduce_'
>>>> energy.F:(.text+0x470d): undefined reference to `mpi_allreduce_'
>>>> gtfnam.o: In function `gtfnam_':
>>>> gtfnam.F:(.text+0x65): undefined reference to `mpi_bcast_'
>>>> gtfnam.F:(.text+0x8f): undefined reference to `mpi_bcast_'
>>>> lapw0.o: In function `MAIN__':
>>>> lapw0.F:(.text+0xa77): undefined reference to `mpi_abort_'
>>>> lapw0.F:(.text+0x7d89): undefined reference to `mpi_reduce_'
>>>> lapw0.F:(.text+0x7e33): undefined reference to `mpi_send_'
>>>> lapw0.F:(.text+0x93ce): undefined reference to `mpi_recv_'
>>>> lapw0.F:(.text+0xb1e1): undefined reference to `mpi_send_'
>>>> lapw0.F:(.text+0xb67b): undefined reference to `mpi_barrier_'
>>>> lapw0.F:(.text+0xb687): undefined reference to `mpi_finalize_'
>>>> lapw0.F:(.text+0xb770): undefined reference to `mpi_recv_'
>>>> lapw0.F:(.text+0xdfba): undefined reference to `mpi_allreduce_'
>>>> lapw0.F:(.text+0x11002): undefined reference to `mpi_recv_'
>>>> lapw0.F:(.text+0x11d31): undefined reference to `mpi_send_'
>>>> lapw0.F:(.text+0x11dcd): undefined reference to `mpi_address_'
>>>> lapw0.F:(.text+0x11df4): undefined reference to `mpi_address_'
>>>> lapw0.F:(.text+0x11e1b): undefined reference to `mpi_address_'
>>>> lapw0.F:(.text+0x11e42): undefined reference to `mpi_address_'
>>>> lapw0.F:(.text+0x11e69): undefined reference to `mpi_address_'
>>>> lapw0.o:lapw0.F:(.text+0x11e90): more undefined references to
>>>> `mpi_address_' follow
>>>> lapw0.o: In function `MAIN__':
>>>> lapw0.F:(.text+0x120ce): undefined reference to `mpi_type_struct_'
>>>> lapw0.F:(.text+0x120e1): undefined reference to `mpi_type_commit_'
>>>> lapw0.F:(.text+0x1218b): undefined reference to `mpi_send_'
>>>> lapw0.F:(.text+0x121d7): undefined reference to `mpi_send_'
>>>> lapw0.F:(.text+0x12234): undefined reference to `mpi_send_'
>>>> lapw0.F:(.text+0x12280): undefined reference to `mpi_send_'
>>>> lapw0.F:(.text+0x122dd): undefined reference to `mpi_send_'
>>>> lapw0.o:lapw0.F:(.text+0x12329): more undefined references to
>>>> `mpi_send_'
>>>> follow
>>>> lapw0.o: In function `MAIN__':
>>>> lapw0.F:(.text+0x12538): undefined reference to `mpi_recv_'
>>>> lapw0.F:(.text+0x1258b): undefined reference to `mpi_recv_'
>>>> lapw0.F:(.text+0x125ef): undefined reference to `mpi_recv_'
>>>> lapw0.F:(.text+0x12642): undefined reference to `mpi_recv_'
>>>> lapw0.F:(.text+0x126a6): undefined reference to `mpi_recv_'
>>>> lapw0.o:lapw0.F:(.text+0x126f9): more undefined references to
>>>> `mpi_recv_'
>>>> follow
>>>> rean0.o: In function `rean0_':
>>>> rean0.F:(.text+0x14d3): undefined reference to `mpi_allreduce_'
>>>> make[1]: *** [lapw0_mpi] Error 1
>>>> make[1]: Leaving directory `/root/WIEN/SRC_lapw0'
>>>> make: *** [para] Error 2
>>>>
>>>> Any help is greatly appreciated
>>>>
>>>> Sergio Y. Rodriguez
>>>> Physics Department
>>>> Texas A&M University
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Wien mailing list
>>>> Wien at zeus.theochem.tuwien.ac.at
>>>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>>>>
>>>
>>>
>>>
>>> --
>>> Laurence Marks
>>> Department of Materials Science and Engineering
>>> MSE Rm 2036 Cook Hall
>>> 2220 N Campus Drive
>>> Northwestern University
>>> Evanston, IL 60208, USA
>>> Tel: (847) 491-3996 Fax: (847) 491-7820
>>> email: L-marks at northwestern dot edu
>>> Web: www.numis.northwestern.edu
>>> Commission on Electron Diffraction of IUCR
>>> www.numis.northwestern.edu/IUCR_CED
>>> _______________________________________________
>>> Wien mailing list
>>> Wien at zeus.theochem.tuwien.ac.at
>>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>>>
>>>
>>
>>
>> _______________________________________________
>> Wien mailing list
>> Wien at zeus.theochem.tuwien.ac.at
>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>>
>
>
>
> --
> Laurence Marks
> Department of Materials Science and Engineering
> MSE Rm 2036 Cook Hall
> 2220 N Campus Drive
> Northwestern University
> Evanston, IL 60208, USA
> Tel: (847) 491-3996 Fax: (847) 491-7820
> email: L-marks at northwestern dot edu
> Web: www.numis.northwestern.edu
> Commission on Electron Diffraction of IUCR
> www.numis.northwestern.edu/IUCR_CED
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
>
More information about the Wien
mailing list