Gilles, Jeff, In Open MPI 1.6 days, MPI_ARGVS_NULL and MPI_STATUSES_IGNORE were defined as double precision and MPI_Comm_spawn_multiple and MPI_Waitall etc. interfaces had two subroutines each.
https://github.com/open-mpi/ompi-release/blob/v1.6/ompi/include/mpif-common.h#L148 https://github.com/open-mpi/ompi-release/blob/v1.6/ompi/mpi/f90/scripts/mpi-f90-interfaces.h.sh#L9568 https://github.com/open-mpi/ompi-release/blob/v1.6/ompi/mpi/f90/scripts/mpi-f90-interfaces.h.sh#L8639 This situation was changed in Open MPI 1.8 but perhaps use-mpi-tkr interface was not correctly adapted. Though I don't know the background, the files which caused compilation errors may be unnecessary. > Jeff, > > so it seems NAG uses the use-mpi-tkr interface. > do you have any recollection of why > - in mpi_waitall_f90.f90, MPI_Waitall and friends have two implementations > (e.g. MPI_WaitallI and MPI_WaitallS) > especially, MPI_WaitallI declares array_of_statuses as a double precision > > - in mpi-f90-interfaces.h, MPI_Waitall has only one prototype (no double > precision) > > Cheers, > > Gilles > > > On Monday, August 15, 2016, Franz-Joseph Barthold <franz-joseph.barthold@tu- > dortmund.de > <javascript:_e(%7B%7D,'cvml','franz-joseph.barth...@tu-dortmund.de');>> > wrote: > > > Dear developers, > > > > I actually compiled Open MPI 2.0.0 using the NAG compiler (nagfor) in its > > recent version. > > > > The following error occur: > > > > make[2]: Entering directory '/home/barthold/devel/prgs/ext > > ernal/openmpi-2.0.0/ > > ompi/mpi/fortran/use-mpi-tkr' > > FC mpi_comm_spawn_multiple_f90.lo > > NAG Fortran Compiler Release 6.1(Tozai) Build 6106 > > Error: mpi_comm_spawn_multiple_f90.f90: Argument 3 to > > MPI_COMM_SPAWN_MULTIPLE > > has data type DOUBLE PRECISION in reference from MPI_COMM_SPAWN_MULTIPLEN > > and > > CHARACTER in reference from MPI_COMM_SPAWN_MULTIPLEA > > [NAG Fortran Compiler error termination, 1 error] > > Makefile:1896: recipe for target 'mpi_comm_spawn_multiple_f90.lo' failed > > make[2]: [mpi_comm_spawn_multiple_f90.lo] Error 1 (ignored) > > FC mpi_testall_f90.lo > > NAG Fortran Compiler Release 6.1(Tozai) Build 6106 > > Error: mpi_testall_f90.f90: Argument 4 to MPI_TESTALL has data type DOUBLE > > PRECISION in reference from MPI_TESTALLI and INTEGER in reference from > > MPI_TESTALLS > > [NAG Fortran Compiler error termination, 1 error] > > Makefile:1896: recipe for target 'mpi_testall_f90.lo' failed > > make[2]: [mpi_testall_f90.lo] Error 1 (ignored) > > FC mpi_testsome_f90.lo > > NAG Fortran Compiler Release 6.1(Tozai) Build 6106 > > Error: mpi_testsome_f90.f90: Argument 5 to MPI_TESTSOME has data type > > DOUBLE > > PRECISION in reference from MPI_TESTSOMEI and INTEGER in reference from > > MPI_TESTSOMES > > [NAG Fortran Compiler error termination, 1 error] > > Makefile:1896: recipe for target 'mpi_testsome_f90.lo' failed > > make[2]: [mpi_testsome_f90.lo] Error 1 (ignored) > > FC mpi_waitall_f90.lo > > NAG Fortran Compiler Release 6.1(Tozai) Build 6106 > > Error: mpi_waitall_f90.f90: Argument 3 to MPI_WAITALL has data type DOUBLE > > PRECISION in reference from MPI_WAITALLI and INTEGER in reference from > > MPI_WAITALLS > > [NAG Fortran Compiler error termination, 1 error] > > Makefile:1896: recipe for target 'mpi_waitall_f90.lo' failed > > make[2]: [mpi_waitall_f90.lo] Error 1 (ignored) > > FC mpi_waitsome_f90.lo > > NAG Fortran Compiler Release 6.1(Tozai) Build 6106 > > Error: mpi_waitsome_f90.f90: Argument 5 to MPI_WAITSOME has data type > > DOUBLE > > PRECISION in reference from MPI_WAITSOMEI and INTEGER in reference from > > MPI_WAITSOMES > > [NAG Fortran Compiler error termination, 1 error] > > Makefile:1896: recipe for target 'mpi_waitsome_f90.lo' failed > > > > Other compiler (Absoft 16.0, Gfortran 6.1, Intel 16 ifort) are less > > sensitive > > and compile. But may fail during runtime. > > > > My experience is that NAG (nagfor) is the most sensitive Fortran compiler. > > Thus, I recommend its usage within the development phase. _______________________________________________ devel mailing list devel@lists.open-mpi.org https://rfd.newmexicoconsortium.org/mailman/listinfo/devel
