Re: [AMBER] Problems during installation of amber-14 with cuda-5 on GTX 670

From: Josh Berryman <the.real.josh.berryman.gmail.com>
Date: Fri, 2 May 2014 08:45:19 +0200

Oh, and send the -noemil flag when you run ./configure !




On 2 May 2014 08:43, Josh Berryman <the.real.josh.berryman.gmail.com> wrote:

> > testing [C++ / fortran] cross-compile with MPI libs
> > Error! Could not link C++ mpi code using mpi fortran linker:
> > "mpif90 -lstdc++ -lmpi_cxx"
> > [C++ / fortran / MPI] cross-linking is needed for cuda_parallel and EMIL
> > builds.
> > Try: "mpif90 -show"
> > Configure failed due to the errors above!
> >
> > own.arrakis ~/Documents/simulations/restart_holo $ mpif90 -show
> > gfortran -I/usr/lib/openmpi/include -pthread -I/usr/lib/openmpi/lib
> > -L/usr/lib/openmpi/lib -lmpi_f90 -lmpi_f77 -lmpi -lopen-rte -lopen-pal
> -ldl
> > -Wl,--export-dynamic -lnsl -lutil -lm -ldl
> >
> > What should be done here?
>
> >>Install a working MPI.
>
> Or try commenting out the check in AmberTools/src/configure2
>
> "
> echo "[C++ / fortran / MPI] cross-linking is needed for
> cuda_parallel and EMIL builds."
> echo "Try: \"mpif90 -show\""
> this line-----> exit 1
> fi
> "
> and hope that the functionality which you need will then still compile
> OK. Not ideal, I know, but if you are building for execution on your local
> machine then CUDA_PARALLEL and EMIL are probably not going to be used
> anyway these are heavy-duty crunching utilities that would normally be on a
> cluster.
>
> Josh
>
>
>
>
>
> On 1 May 2014 15:15, Jason Swails <jason.swails.gmail.com> wrote:
>
>> On Thu, 2014-05-01 at 13:05 +0400, James Starlight wrote:
>> > Jason,
>> >
>> > mpirun -np 4 MMPBSA.py.MPI
>> > resulted in
>> > mpirun was unable to launch the specified application as it could not
>> find
>> > an executable:
>> >
>> > Executable: MMPBSA.py.MPI
>>
>> Then you do not have AmberTools installed in parallel.
>>
>> >
>> > Previously I faced with the problems of compilation of amber using
>> > ./configure -mpi
>> >
>> > resulted in
>> > testing [C++ / fortran] cross-compile with MPI libs
>> > Error! Could not link C++ mpi code using mpi fortran linker:
>> > "mpif90 -lstdc++ -lmpi_cxx"
>> > [C++ / fortran / MPI] cross-linking is needed for cuda_parallel and EMIL
>> > builds.
>> > Try: "mpif90 -show"
>> > Configure failed due to the errors above!
>> >
>> > own.arrakis ~/Documents/simulations/restart_holo $ mpif90 -show
>> > gfortran -I/usr/lib/openmpi/include -pthread -I/usr/lib/openmpi/lib
>> > -L/usr/lib/openmpi/lib -lmpi_f90 -lmpi_f77 -lmpi -lopen-rte -lopen-pal
>> -ldl
>> > -Wl,--export-dynamic -lnsl -lutil -lm -ldl
>> >
>> > What should be done here?
>>
>> Install a working MPI. The configure_openmpi and configure_mpich
>> scripts can help. Getting a working MPI can be tricky and requires a
>> bit of Linux savvy to get your environment set up correctly, however.
>> If you can't install one from your package manager or get one working
>> yourself, I suggest trying to find an experienced Linux user near you to
>> help.
>>
>> Or, of course, just run in serial.
>>
>> Good luck,
>> Jason
>>
>> --
>> Jason M. Swails
>> BioMaPS,
>> Rutgers University
>> Postdoctoral Researcher
>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri May 02 2014 - 00:00:03 PDT
Custom Search