Hello, I just gave gcc 4.9.0 a try and the mpi_f09 module is there but it seems to miss some functions:
mpifort test.f90 /tmp/ccHCEbXC.o: In function `MAIN__': test.f90:(.text+0x35a): undefined reference to `mpi_win_lock_all_' test.f90:(.text+0x373): undefined reference to `mpi_win_lock_all_' test.f90:(.text+0x964): undefined reference to `mpi_win_sync_' test.f90:(.text+0x978): undefined reference to `mpi_win_sync_' test.f90:(.text+0xb20): undefined reference to `mpi_win_sync_' test.f90:(.text+0xb34): undefined reference to `mpi_win_sync_' test.f90:(.text+0x1772): undefined reference to `mpi_win_unlock_all_' test.f90:(.text+0x1786): undefined reference to `mpi_win_unlock_all_' collect2: error: ld returned 1 exit status Here also some configure log content: OMPI_BUILD_FORTRAN_USEMPI_IGNORE_TKR_BINDINGS_FALSE='#' OMPI_BUILD_FORTRAN_USEMPI_IGNORE_TKR_BINDINGS_TRUE='' OMPI_BUILD_FORTRAN_USEMPI_TKR_BINDINGS_FALSE='' OMPI_BUILD_FORTRAN_USEMPI_TKR_BINDINGS_TRUE='#' OMPI_FORTRAN_IGNORE_TKR_PREDECL='!GCC$ ATTRIBUTES NO_ARG_CHECK ::' OMPI_FORTRAN_IGNORE_TKR_TYPE='type(*), dimension(*)' OMPI_FORTRAN_USEMPI_DIR='mpi/fortran/use-mpi-ignore-tkr' OMPI_FORTRAN_USEMPI_LIB='-lmpi_usempi_ignore_tkr' libmpi_usempi_ignore_tkr_so_version='' libmpi_usempi_tkr_so_version='4:2:3' #define OMPI_FORTRAN_IGNORE_TKR_PREDECL "!GCC$ ATTRIBUTES NO_ARG_CHECK ::" #define OMPI_FORTRAN_IGNORE_TKR_TYPE #define OMPI_FORTRAN_HAVE_IGNORE_TKR 1 configure:10267: result: yes (mpif.h, mpi and mpi_f08 modules) configure:10417: checking which 'use mpi_f08' implementation to use configure:58804: checking which mpi_f08 implementation to build configure:58845: checking if building Fortran 'use mpi_f08' bindings OMPI_F08_SUFFIX='_f08' Regards Christoph Niethammer -- Christoph Niethammer High Performance Computing Center Stuttgart (HLRS) Nobelstrasse 19 70569 Stuttgart Tel: ++49(0)711-685-87203 email: nietham...@hlrs.de http://www.hlrs.de/people/niethammer ----- Original Message ----- From: "Marcus G Daniels" <mdani...@lanl.gov> To: "Jeff Squyres (jsquyres)" <jsquy...@cisco.com>, "Open MPI User's List" <us...@open-mpi.org> Sent: Thursday, August 14, 2014 8:13:27 AM Subject: Re: [OMPI users] GCC 4.9 and MPI_F08? Hi Jeff, Works for me! (With mpi_f08, GCC 4.9.1 absolutely insists on getting the finer details right on things like MPI_User_function types for MPI_Op_create. So I'll assume the rest of the type checking is just as good, and be glad I took that minor detour..) Thanks, Marcus -----Original Message----- From: Jeff Squyres (jsquyres) [mailto:jsquy...@cisco.com] Sent: Wednesday, August 13, 2014 10:00 AM To: Open MPI User's List Cc: Daniels, Marcus G Subject: Re: [OMPI users] GCC 4.9 and MPI_F08? Marcus -- The fix was applied yesterday to the v1.8 branch. Would you mind testing the v1.8 nightly tarball from last night, just to make sure it works for you? http://www.open-mpi.org/nightly/v1.8/ On Aug 12, 2014, at 2:54 PM, Jeff Squyres (jsquyres) <jsquy...@cisco.com> wrote: > @#$%@#$%#@$% > > I was very confused by this bug report, because in my mind, a) this > functionality is on the SVN trunk, and b) we had moved the gcc functionality > to the v1.8 branch long ago. > > I just checked the SVN/Trac records: > > 1. I'm right: this functionality *is* on the trunk. If you build the OMPI > SVN trunk with gcc/gfortran 4.9, you get the ignore TKR "mpi" module and the > mpi_f08 module. I just tried it myself to verify this. > > 2. I'm (sorta) right: we CMR'ed the "add the GCC ignore TKR functionality" in > https://svn.open-mpi.org/trac/ompi/ticket/4058 for v1.7.5, but it looks like > that CMR was botched somehow and the functionality wasn't applied (!) ...even > though the log message says it was. Sad panda. > > I'll open a ticket to track this. We'll have to see how the RM feels about > putting this in at the last second; it may or may not make the cutoff for > 1.8.2 (the freeze has already occurred). > > > > On Aug 12, 2014, at 12:32 PM, Daniels, Marcus G <mdani...@lanl.gov> wrote: > >> Hi Jeff, >> >> On Tue, 2014-08-12 at 16:18 +0000, Jeff Squyres (jsquyres) wrote: >>> Can you send the output from configure, the config.log file, and the >>> ompi_config.h file? >> >> Attached. configure.log comes from >> >> (./configure --prefix=/usr/projects/eap/tools/openmpi/1.8.2rc3 2>&1) >> > configure.log >> >> Seems fishy that there is no "checking for Fortran compiler support of ! >> GCC$ ATTRIBUTES NO_ARG_CHECK". >> >> checking for Fortran compiler module include flag... -I checking >> Fortran compiler ignore TKR syntax... not cached; checking variants >> checking for Fortran compiler support of TYPE(*), DIMENSION(*)... no >> checking for Fortran compiler support of !DEC$ ATTRIBUTES >> NO_ARG_CHECK... no checking for Fortran compiler support of !$PRAGMA >> IGNORE_TKR... no checking for Fortran compiler support of !DIR$ >> IGNORE_TKR... no checking for Fortran compiler support of !IBM* >> IGNORE_TKR... no checking Fortran compiler ignore TKR syntax... >> 0:real:! >> checking if Fortran compiler supports ISO_C_BINDING... yes checking >> if building Fortran 'use mpi' bindings... yes checking if building >> Fortran 'use mpi_f08' bindings... no >> >> Marcus >> >> >> <openmpi-1.8.2rc3-config.tar.gz> > > > -- > Jeff Squyres > jsquy...@cisco.com > For corporate legal information go to: > http://www.cisco.com/web/about/doing_business/legal/cri/ > > _______________________________________________ > users mailing list > us...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/users > Link to this post: > http://www.open-mpi.org/community/lists/users/2014/08/25004.php -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/ _______________________________________________ users mailing list us...@open-mpi.org Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/users Link to this post: http://www.open-mpi.org/community/lists/users/2014/08/25009.php
