The ABINIT Users Mailing List ( CLOSED )

[Jyh-Shyong <c00jsh00@nchc.org.tw>]

张渟 提到:
> Dear all:
>
> I've got some trouble in compiling the abinit-5.8.4. After ./configure 
> --with-mpi-prefix=~/src/mpich-1.2.7p1, the "make" command yields error 
> messages(see below), and who can tell me what's wrong with it? My 
> compiler is pgi9.0-1 for opteron.
>
> Regards
>
> Zhang Ting
>
> ===============================
> make[6]: Leaving directory 
> `/home/phzting/src/abinit-5.8.4/plugins/bigdft/bigdft-1.2.0.2/src/profiling'
> make[6]: Entering directory 
> `/home/phzting/src/abinit-5.8.4/plugins/bigdft/bigdft-1.2.0.2/src'
> /home/phzting/src/mpich-1.2.7p1/bin/mpif90 -I. -Imodules -I. -IPSolver 
> -IPSolver/ABINIT-common -g -module 
> /home/phzting/src/abinit-5.8.4/tmp-modules -I/home/
> phzting/src/abinit-5.8.4/tmp-modules -Mextend -O2 -c -o BigDFT_API.o 
> BigDFT_API.f90
> /home/phzting/src/mpich-1.2.7p1/bin/mpif90 -I. -Imodules -I. -IPSolver 
> -IPSolver/ABINIT-common -g -module 
> /home/phzting/src/abinit-5.8.4/tmp-modules -I/home/
> phzting/src/abinit-5.8.4/tmp-modules -Mextend -O2 -c -o cluster.o 
> cluster.f90
> /home/phzting/src/mpich-1.2.7p1/bin/mpif90 -I. -Imodules -I. -IPSolver 
> -IPSolver/ABINIT-common -g -module 
> /home/phzting/src/abinit-5.8.4/tmp-modules -I/home/
> phzting/src/abinit-5.8.4/tmp-modules -Mextend -O2 -c -o 
> input_variables.o input_variables.f90
> /home/phzting/src/mpich-1.2.7p1/bin/mpif90 -I. -Imodules -I. -IPSolver 
> -IPSolver/ABINIT-common -g -module 
> /home/phzting/src/abinit-5.8.4/tmp-modules -I/home/
> phzting/src/abinit-5.8.4/tmp-modules -Mextend -O2 -c -o init.o init.f90
> /home/phzting/src/mpich-1.2.7p1/bin/mpif90 -I. -Imodules -I. -IPSolver 
> -IPSolver/ABINIT-common -g -module 
> /home/phzting/src/abinit-5.8.4/tmp-modules -I/home/
> phzting/src/abinit-5.8.4/tmp-modules -Mextend -O2 -c -o sumrho.o 
> sumrho.f90
> PGF90-S-0038-Symbol, mpi_in_place, has not been explicitly declared 
> (sumrho.f90)
> 0 inform, 0 warnings, 1 severes, 0 fatal for sumrho
> PGF90-S-0121-Operation requires logical or integer data types 
> (init.f90: 348)
> 0 inform, 0 warnings, 1 severes, 0 fatal for createprojectorsarrays
> make[6]: *** [sumrho.o] Error 2
> make[6]: *** Waiting for unfinished jobs....
> make[6]: *** [init.o] Error 2
> make[6]: Leaving directory 
> `/home/phzting/src/abinit-5.8.4/plugins/bigdft/bigdft-1.2.0.2/src'
> make[5]: *** [all-recursive] Error 1
> make[5]: Leaving directory 
> `/home/phzting/src/abinit-5.8.4/plugins/bigdft/bigdft-1.2.0.2/src'
> make[4]: *** [all-recursive] Error 1
> make[4]: Leaving directory 
> `/home/phzting/src/abinit-5.8.4/plugins/bigdft/bigdft-1.2.0.2'
> make[3]: *** [all] Error 2
> make[3]: Leaving directory 
> `/home/phzting/src/abinit-5.8.4/plugins/bigdft/bigdft-1.2.0.2'
> make[2]: *** [build-stamp] Error 2
> make[2]: Leaving directory `/home/phzting/src/abinit-5.8.4/plugins/bigdft'
> make[1]: *** [package-ready] Error 2
> make[1]: Leaving directory `/home/phzting/src/abinit-5.8.4/plugins/bigdft'
> make: *** [multi] Error 2
Perhaps you can try mpich2 instead of mpich, mpich2 is more robust than 
mpich.

I encountered the same problem a few weeks ago, here is what I did:

I checked the argument list of this subroutine (MPI_ALLREDUCE) called by 
all *.f90 files in
src directory, and found that except file sumrho.f90, the first argument 
of this subroutine called
by all other files is declared as real type, but there is no declaration 
of MPI_IN_PLACE in sumrho.f90,
so I added a declaration statement for MPI_IN_PLACE in file sumrho.f90, 
and recompiled the program.
I was using ifort, not pgi compiler.

I also tried mpich2, and the program compiled without modifying sumrho.f90.


Jyh-Shyong Ho