Well at the risk of repeating Derek's earlier instructions for building libmesh on a powerbook, I thought I would also post what I ended up doing here. I ran into some weird problems when PETSc was actually compiled with a fortran compiler, so I thought it might save someone else some time if they were about to go through the same exercise...
-J A list of steps required to build libmesh on a mac powerbook. .) Download, configure and install mpich (if you have g95, it will find it even if you don't specify) $ FC=g95 RSHCOMMAND=ssh ./configure --prefix=/usr/local/mpich-1.2.7 $ make $ sudo make install .) Download, configure, and install petsc. IMPORTANT: Note the C-BLAS and the explicit turning off of the Fortran interfaces on the configure line*. Also, you will get the debug version by default so I turned that off. $ ./config/configure.py --with-cc=mpicc --with-debugging=0 --with-mpi-dir=/usr/local/mpi --download-c-blas-lapack=1 --with-fortran=0 $ make all test .) Download, configure, and install libmesh. I couldn't get the shared libraries to build, this may be a subject of future work. $ CXX=mpicxx CC=mpicc ./configure --enable-second --enable-perflog --disable-shared $ make -j 2 * Note: when I attempted to compile PETSc with Fortran compilers, none of the resulting executables would actually run. The runtime error message was | 0 - <NO ERROR MESSAGE> : Could not convert index 12079072 into a pointer | The index may be an incorrect argument. | Possible sources of this problem are a missing "include 'mpif.h'", | a misspelled MPI object (e.g., MPI_COM_WORLD instead of MPI_COMM_WORLD) | or a misspelled user variable for an MPI object (e.g., | com instead of comm). | [0] Aborting program ! | [0] Aborting program! | p0_84805: p4_error: : 9039 And apparently has something to do with the MPI Fortran interface storing pointers as type INTEGER, which works on 32-bit machines but not 64. This website appears to have more info: http://www.pgroup.com/userforum/viewtopic.php?start=0&t=560 On Thu, Jun 12, 2008 at 3:22 PM, John Peterson <[EMAIL PROTECTED]> wrote: > Interesting, I guess because of the fact that I had a fortran > compiler in my path, it didn't like the --download-c-blas-lapack=1 > unless I explicitly said > > --with-fortran=0 > > as well. > > Just FYI for anyone trying to build on a Mac, you can get a Fortran95 > compiler in OSX binary form from g95.org > > -J > > > > On Thu, Jun 12, 2008 at 3:09 PM, Derek Gaston <[EMAIL PROTECTED]> wrote: >> Not if you don't specify --with-fc and you make sure to use >> --download-c-blas-lapack=1 (note the _c_ in the middle of that). >> >> Derek >> >> On Thu, Jun 12, 2008 at 2:05 PM, John Peterson <[EMAIL PROTECTED]> wrote: >>> Hey Derek, >>> >>> I'm curious, without a fortran compiler of any kind, didn't petsc's >>> configure script complain with >>> >>> ================================================================================= >>> TESTING: FortranMPICheck >>> from config.packages.MPI(python/BuildSystem/config/packages/MPI.py:523) >>> >>> ********************************************************************************* >>> UNABLE to CONFIGURE with GIVEN OPTIONS (see configure.log >>> for details): >>> --------------------------------------------------------------------------------------- >>> Fortran error! mpif.h could not be located at: ['/usr/local/mpi/include'] >>> >>> Thanks, >>> J >>> >>> On Thu, Jun 12, 2008 at 9:42 AM, Derek Gaston <[EMAIL PROTECTED]> wrote: >>>> For OSX 10.5.3 here's what I do. >>>> >>>> 1. Get MPICH1... compile and install into a place in my home directory. >>>> 2. Make sure my PATH is set so that the mpicxx/mpicc is the one that >>>> I just compiled and installed from MPICH1. >>>> 3. Compile PetsC using "./config/configure.py --with-cc=mpicc -- >>>> download-c-blas-lapack=1". Note the _c_ blas-lapack as I haven't >>>> taken the time to get a fortran compiler working in OSX. >>>> 4. Build PetsC >>>> 5. Set PETSC_HOME and PETS_ARCH. >>>> 6. Download libMesh and configure with "./configure --disable- >>>> shared". If you do it all right there is no reason to use --enable- >>>> mpi or --with-mpi. >>>> 7. Make libMesh. >>>> 8. Enjoy! >>>> >>>> I'm sure this same sequence will work with MPICH2. The biggest thing >>>> is to make sure that mpicc is in your path and to pass that as --with- >>>> cc to PetsC. If you do that (and you set PETSC_HOME and PETSC_ARCH >>>> correctly) libMesh will automatically pick up on PetsC and compile >>>> with mpi. >>>> >>>> Derek >>>> >>>> On Jun 12, 2008, at 1:51 AM, Coulaud Olivier wrote: >>>> >>>>> Hi, >>>>> >>>>> I try to install the last version on Mac OS X (10.5.3 and >>>>> gcc 4.0) with the following options >>>>> >>>>> ./configure --disable-triangle --enable-mpi --with-mpi=/opt/mpi/mpich2 >>>>> --disable-tecplot >>>>> >>>>> But it seems to have some problems when I want shared libraries. >>>>> >>>>> ... >>>>> >>>>> Compiling C (in optimized mode) timing.c... >>>>> >>>>> Compiling C (in optimized mode) util.c... >>>>> >>>>> Linking /Users/coulaudolivier/MesData/Recherche/DM-EF/libmultiscale/ >>>>> libmultiscale-externals/external/libmesh-0.6.2/contrib/lib/i686-apple- >>>>> darwin9.3.0_opt/libmetis.dylib >>>>> >>>>> Undefined symbols: >>>>> >>>>> "_main", referenced from: >>>>> >>>>> start in crt1.10.5.o >>>>> >>>>> ld: symbol(s) not found >>>>> >>>>> collect2: ld returned 1 exit status >>>>> >>>>> make[2]: *** [/Users/coulaudolivier/MesData/Recherche/DM-EF/ >>>>> libmultiscale/libmultiscale-externals/external/libmesh-0.6.2/contrib/ >>>>> lib/i686-apple-darwin9.3.0_opt/libmetis.dylib] Error 1 >>>>> >>>>> make[1]: *** [all] Error 2 >>>>> >>>>> make: *** [/Users/coulaudolivier/MesData/Recherche/DM-EF/ >>>>> libmultiscale/ >>>>> libmultiscale-externals/external/libmesh-0.6.2/lib/i686-apple- >>>>> darwin9.3.0_opt/libmesh.dylib] Error 2 >>>>> >>>>> >>>>> >>>>> I set in Make.common -dynamic rather -shared (non supported for mac os >>>>> x) but I have the same error. >>>>> >>>>> >>>>> >>>>> Any Idea? >>>>> >>>>> >>>>> >>>>> Moreover; is it possible to set a verbose mode to the compilation? >>>>> >>>>> >>>>> >>>>> Thanks in advance >>>>> >>>>> -- >>>>> >>>>> >>>>> >>>>> Olivier Coulaud >>>>> >>>>> >>>>> ----------------------------------------------------------------- >>>>> Scalaplix project - INRIA Bordeaux - Sud-Ouest >>>>> 351 cours de la libération - F-33400 Talence - France >>>>> >>>>> Tel: [33] (0) 5 24 57 40 80 >>>>> WEB: http://www.labri.fr/perso/coulaud/ >>>>> Mail: [EMAIL PROTECTED] >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> ------------------------------------------------------------------------- >>>>> Check out the new SourceForge.net Marketplace. >>>>> It's the best place to buy or sell services for >>>>> just about anything Open Source. >>>>> http://sourceforge.net/services/buy/index.php >>>>> _______________________________________________ >>>>> Libmesh-users mailing list >>>>> [email protected] >>>>> https://lists.sourceforge.net/lists/listinfo/libmesh-users >>>> >>>> >>>> ------------------------------------------------------------------------- >>>> Check out the new SourceForge.net Marketplace. >>>> It's the best place to buy or sell services for >>>> just about anything Open Source. >>>> http://sourceforge.net/services/buy/index.php >>>> _______________________________________________ >>>> Libmesh-users mailing list >>>> [email protected] >>>> https://lists.sourceforge.net/lists/listinfo/libmesh-users >>>> >>> >> > ------------------------------------------------------------------------- Check out the new SourceForge.net Marketplace. It's the best place to buy or sell services for just about anything Open Source. http://sourceforge.net/services/buy/index.php _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
