What version of cmake are you using? I had a similar problem for metis 5.0.1: I got metis compiled by using version of cmake 2.6 which is older than the requested one by dropping the request in Cmakelists
2011/12/1 Jack Poulson <jack.poulson at gmail.com> > I'm not too surprised; I haven't looked at how ParMETIS is being used > within PETSc. The problem is definitely that ParMETIS isn't properly > locating its GKlib subproject, and I think that the issue is in the lines I > pointed out, but I'm not sure as to what the appropriate replacement for > i"${CMAKE_CURRENT_SOURCE_DIR}" isin your situation. > > Jack > > > On Wed, Nov 30, 2011 at 5:18 PM, Mark F. Adams <mark.adams at > columbia.edu>wrote: > >> This did not seem to help. >> Mark >> >> On Nov 30, 2011, at 5:35 PM, Jack Poulson wrote: >> >> I have actually run into a similar problem with ParMetis. I fixed it on >> my end by changing the ParMetis CMakeLists.txt lines >> >> set(GKLIB_PATH METIS/GKlib CACHE PATH "path to GKlib") >> set(METIS_PATH METIS CACHE PATH "path to METIS") >> >> to >> >> set(GKLIB_PATH ${CMAKE_CURRENT_SOURCE_DIR}/metis/GKlib CACHE PATH "path >> to GKlib >> ") >> set(METIS_PATH ${CMAKE_CURRENT_SOURCE_DIR}/metis CACHE PATH "path to >> METIS") >> >> Your mileage may vary. >> >> Jack >> >> On Wed, Nov 30, 2011 at 4:32 PM, Aron Ahmadia <aron.ahmadia at >> kaust.edu.sa>wrote: >> >>> Mark, >>> >>> Sorry, no help here, except that cmake is pretty easy to install from >>> source, so that might be my recommendation if you don't want to wait for >>> help from their systems team. >>> >>> A >>> >>> On Thu, Dec 1, 2011 at 1:30 AM, Mark F. Adams <mark.adams at >>> columbia.edu>wrote: >>> >>>> Thanks Tim, I do use gcc. And this is on Franklin at NERSC. >>>> >>>> Cmake is just kicking my ass, now I get (appended). >>>> >>>> I do have an old version of cmake here: >>>> >>>> madams/petsc-dev> which cmake >>>> /usr/common/usg/cmake/2.8.3/bin/cmake >>>> >>>> any ideas? >>>> Mark >>>> >>>> =============================================================================== >>>> Configuring METIS; this may take >>>> several minutes >>>> >>>> =============================================================================== >>>> >>>> >>>> ******************************************************************************* >>>> UNABLE to CONFIGURE with GIVEN OPTIONS (see configure.log >>>> for details): >>>> >>>> ------------------------------------------------------------------------------- >>>> Error running configure on METIS: Could not execute "cd >>>> /global/u2/m/madams/petsc-dev/externalpackages/metis-5.0.2 && make >>>> distclean && make config >>>> prefix=/global/homes/m/madams/petsc-dev/arch-xt5-opt cc="cc" >>>> doubleprecision=1": >>>> rm -rf build/Linux-x86_64 >>>> rm -rf build/Linux-x86_64 >>>> mkdir -p build/Linux-x86_64 >>>> cd build/Linux-x86_64 && cmake >>>> /global/u2/m/madams/petsc-dev/externalpackages/metis-5.0.2 >>>> -DCMAKE_VERBOSE_MAKEFILE=1 -DGKLIB_PATH= >>>> -DCMAKE_INSTALL_PREFIX=/global/homes/m/madams/petsc-dev/arch-xt5-opt >>>> -DCMAKE_C_COMPILER=cc -DMETIS_USE_DOUBLEPRECISION=1 >>>> -- The C compiler identification is GNU >>>> -- The CXX compiler identification is GNU >>>> -- Check for working C compiler: /opt/cray/xt-asyncpe/5.01/bin/cc >>>> -- Check for working C compiler: /opt/cray/xt-asyncpe/5.01/bin/cc -- >>>> works >>>> -- Detecting C compiler ABI info >>>> -- Detecting C compiler ABI info - done >>>> -- Check for working CXX compiler: /opt/gcc/4.5.3/bin/c++ >>>> -- Check for working CXX compiler: /opt/gcc/4.5.3/bin/c++ -- works >>>> -- Detecting CXX compiler ABI info >>>> -- Detecting CXX compiler ABI info - done >>>> -- Configuring incomplete, errors occurred! >>>> CMake Error at CMakeLists.txt:21 (include): >>>> include could not find load file: >>>> >>>> /GKlibSystem.cmake >>>> >>>> >>>> make: *** [config] Error 1 >>>> >>>> >>>> >>>> On Nov 30, 2011, at 4:00 PM, Tim Gallagher wrote: >>>> >>>> Without knowing how your system is configured, this may not help. But >>>> on the Cray systems we use (they have the modules package installed), if we >>>> are using a compiler other than GNU (Intel, PGI, etc) but we want to use >>>> cmake, we still have to load the gcc module (not the programming >>>> environment, just the module). >>>> >>>> On our machines, it means we may load the PrgEnv-Intel module (so cc >>>> points to icc, ftn points to ifort, etc) but we also have to load >>>> gcc/<version> to get cmake to work. >>>> >>>> Tim >>>> >>>> ------------------------------ >>>> *From: *"Aron Ahmadia" <aron.ahmadia at kaust.edu.sa> >>>> *To: *"For users of the development version of PETSc" < >>>> petsc-dev at mcs.anl.gov> >>>> *Sent: *Wednesday, November 30, 2011 3:49:25 PM >>>> *Subject: *Re: [petsc-dev] metis config error on XT4 >>>> >>>> looks like a broken cmake... >>>> >>>> What happens when you just run: cmake by itself? >>>> >>>> A >>>> >>>> On Wed, Nov 30, 2011 at 11:47 PM, Mark F. Adams < >>>> mark.adams at columbia.edu> wrote: >>>> >>>>> I'm getting this error on Franklin at NERSC: >>>>> >>>>> >>>>> ******************************************************************************* >>>>> UNABLE to CONFIGURE with GIVEN OPTIONS (see configure.log >>>>> for details): >>>>> >>>>> ------------------------------------------------------------------------------- >>>>> Error running configure on METIS: Could not execute "cd >>>>> /global/u2/m/madams/petsc-dev/externalpackages/metis-5.0.2 && make >>>>> distclean && make config >>>>> prefix=/global/homes/m/madams/petsc-dev/arch-xt5-opt cc="cc" >>>>> doubleprecision=1": >>>>> rm -rf build/Linux-x86_64 >>>>> rm -rf build/Linux-x86_64 >>>>> mkdir -p build/Linux-x86_64 >>>>> cd build/Linux-x86_64 && cmake >>>>> /global/u2/m/madams/petsc-dev/externalpackages/metis-5.0.2 >>>>> -DCMAKE_VERBOSE_MAKEFILE=1 -DGKLIB_PATH= >>>>> -DCMAKE_INSTALL_PREFIX=/global/homes/m/madams/petsc-dev/arch-xt5-opt >>>>> -DCMAKE_C_COMPILER=cc -DMETIS_USE_DOUBLEPRECISION=1 >>>>> cmake: /opt/gcc/4.5.3/snos/lib64/libstdc++.so.6: version >>>>> `GLIBCXX_3.4.15' not found (required by cmake) >>>>> make: *** [config] Error 1 >>>>> >>>>> ******************************************************************************* >>>>> >>>>> any ideas? >>>>> >>>>> Mark >>>>> >>>> >>>> >>>> >>>> >>>> >>>> >>> >> >> > -- Stefano -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20111201/839e2f0c/attachment.html>