On Nov 30, 2011, at 6:30 PM, Stefano Zampini wrote: > What version of cmake are you using? >
I am using an older version. I need to install it myself. Sean: don't worry about Franklin, I'll get to it next week. > 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/20111130/35280eab/attachment.html>