Ilmar Wilbers wrote: > Version 1.2.7 > > So I suppose we encourage openmpi, then? >
MPICH 2 should work. PETSc uses MPICH 2 if you ask it to download MPI. Garth > ilmar > > Garth N. Wells wrote: >> >> Ilmar Wilbers wrote: >>> When trying to compile dolfin with mpich on 64-bit, I get the >>> following error: >>> cc1plus: warnings being treated as errors >>> /usr/lib/mpich-shmem/include/mpi2c++/pcomm.h:59: warning: 'class >>> PMPI::Comm' has virtual functions but non-virtual destructor >>> /usr/lib/mpich-shmem/include/mpi2c++/pintracomm.h:30: warning: 'class >>> PMPI::Intracomm' has virtual functions but non-virtual destructor >>> /usr/lib/mpich-shmem/include/mpi2c++/ptopology.h:31: warning: 'class >>> PMPI::Cartcomm' has virtual functions but non-virtual destructor >>> /usr/lib/mpich-shmem/include/mpi2c++/ptopology.h:80: warning: 'class >>> PMPI::Graphcomm' has virtual functions but non-virtual destructor >>> /usr/lib/mpich-shmem/include/mpi2c++/pintercomm.h:31: warning: 'class >>> PMPI::Intercomm' has virtual functions but non-virtual destructor >>> /usr/lib/mpich-shmem/include/mpi2c++/comm.h:106: warning: 'class >>> MPI::Comm' has virtual functions but non-virtual destructor >>> /usr/lib/mpich-shmem/include/mpi2c++/intracomm.h:30: warning: 'class >>> MPI::Intracomm' has virtual functions but non-virtual destructor >>> /usr/lib/mpich-shmem/include/mpi2c++/topology.h:31: warning: 'class >>> MPI::Cartcomm' has virtual functions but non-virtual destructor >>> /usr/lib/mpich-shmem/include/mpi2c++/topology.h:112: warning: 'class >>> MPI::Graphcomm' has virtual functions but non-virtual destructor >>> /usr/lib/mpich-shmem/include/mpi2c++/intercomm.h:31: warning: 'class >>> MPI::Intercomm' has virtual functions but non-virtual destructor >>> scons: *** [dolfin/main/MPI.os] Error 1 >>> >>> See http://fenics.org:8010/dolfin/linux_64/builds/22/step-compile_3/0 >>> for further details. >>> >>> Should dolfin support this, or should we stick to openmpi solely? >>> >> This is an MPI error so I don't know what we can really do about and I >> don't think that we should try to get around it. >> >> Are you using MPICH 1 or 2? Last time I tried MPICH 1 it gave loads of >> C++ errors. >> >> Garth >> >>> ilmar >>> _______________________________________________ >>> DOLFIN-dev mailing list >>> [email protected] >>> http://www.fenics.org/mailman/listinfo/dolfin-dev >> > _______________________________________________ > DOLFIN-dev mailing list > [email protected] > http://www.fenics.org/mailman/listinfo/dolfin-dev _______________________________________________ DOLFIN-dev mailing list [email protected] http://www.fenics.org/mailman/listinfo/dolfin-dev
