[MOAB-dev] Using MOAB linked to Nek5000

[Vijay S. Mahadevan]

> Vijay: As you recommended I have added explicitly hdf5-ldflags and every
> thing went well. However, I have got a couple of errors during "make check"
> regarding imesh as follows, which I think are not important at this stage.

Can you show the full error message here (make -j1 | tee make.log).
The itaps/imesh interface is critical for NEK to work correctly with
MOAB and if the configuration/build is not clean, your linkage with
nek examples will have problems.

Vijay

On Mon, Feb 3, 2014 at 3:17 PM, Mohsen Behzad
<m.behzad.jazi at mail.utoronto.ca> wrote:
> Thanks for your replies.
> Vijay: As you recommended I have added explicitly hdf5-ldflags and every
> thing went well. However, I have got a couple of errors during "make check"
> regarding imesh as follows, which I think are not important at this stage.
>
> make[4]: *** [ScdMeshF90] Error 1
> make[4]: Leaving directory
> `/home/a/ashgriz/mbehzad/nek5_svn/moab-4.6.2/itaps/imesh'
> make[3]: *** [check-am] Error 2
> make[3]: Leaving directory
> `/home/a/ashgriz/mbehzad/nek5_svn/moab-4.6.2/itaps/imesh'
> make[2]: *** [check] Error 2
> make[2]: Leaving directory
> `/home/a/ashgriz/mbehzad/nek5_svn/moab-4.6.2/itaps/imesh'
> make[1]: *** [check-recursive] Error 1
> make[1]: Leaving directory
> `/home/a/ashgriz/mbehzad/nek5_svn/moab-4.6.2/itaps'
> make: *** [check-recursive] Error 1
>
> tim: You mentioned that the code should check for the parallel support of
> HDF5. I couldn't find such thing in the log file, thought, the HDF5 that is
> linked to MOAB is a parallel one. I uploaded the log file for you.
>
>
> Thanks
> Mohsen
>
> On 3-Feb-14, at 2:20 PM, Vijay S. Mahadevan wrote:
>
> Mohsen,
>
> Please attach the config.log file so that we can find out the exact cause of
> errors. On first look the configure options do look right but it is possible
> that you may need to provide the library list explicitly with hdf5-ldflags.
> But we cant suggest any other helpful pointers at the moment without more
> information on why the configuration failed.
>
> Vijay
>
> Dear MOAB team,
>
> I am writing to let you know that I intend to use MOAB with "Nek5000" for my
> simulations which will be performed on the SciNet cluster
> (http://wiki.scinethpc.ca/wiki/) in collaboration with the University of
> Toronto, in which I am a post-doc fellow.
> Right now, I tried to build MOAB with previously built packages on this
> system including "hdf5" (v. 1.8.7 compiled with openmpi) and "netcdf"
> (parallel version: 1.2.0 compiled with openmpi). As it is stated both of
> these packages were compiled with "openmpi" (v. 1.4.4 compiled with intel v.
> 12.1) and therefore there should be no issue of compilers compatibility.
> However, when I attempted to build MOAB on this system with the following
> configuration options and flags:
>
> ./configure FC=mpif90 F77=mpif77 CC=mpicc CXX=mpicxx \
> --prefix=`pwd` \
> --with-mpi=${SCINET_MPI_BASE} \
> --with-netcdf=${SCINET_PNETCDF_BASE} \
> --with-hdf5=${SCINET_HDF5_BASE} \
> --with-zlib="/scinet/gpc/Libraries/HDF5-1.8.7/zlib-intel/" \
> --with-szip=${SCINET_SZIP_BASE} \
> --enable-mbzoltan \
> --with-zoltan="/home/a/ashgriz/mbehzad/nek5_svn/Zoltan_v3.8/BUILD_DIR/"
>
> I have encountered an error in the configuration as:
> "configure:  error: HDF5 library not found or not usable."
> It should be noted that "zlib" and "szip" libraries were compiled with intel
> too.
>
> I am wondering if you can help me on this issue.
>
> Regards,
>
> Mohsen Behzad, Ph.D.
> Department of Civil Eng.
> University of Toronto
> Room 117, 170 College st. M5S 3E3
> Toronto, Ontario, Canada
> m.behzad.jazi at mail.utoronto.ca
> http://mussl.mie.utoronto.ca/
>
>