<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <font face="Droid Serif">Thanks Barry and Satish for your answers.<br>
      <br>
      I installed the correct version of hypre. Also, I changed the
      paths to short dos paths as Satish suggested. Now PETSc compiles
      without problems and the tests are ok.<br>
      <br>
      Regarding Satish notes:<br>
      <br>
      I had to include -LIBS because otherwise the configure script
      stops while testing blas-lapack with the following error:<br>
      <br>
===============================================================================<br>
      TESTING: checkLib from
config.packages.BlasLapack(config/BuildSystem/config/packages/BlasLapack.py:114)*******************************************************************************<br>
               UNABLE to CONFIGURE with GIVEN OPTIONS    (see
      configure.log for details):<br>
-------------------------------------------------------------------------------<br>
      You set a value for --with-blaslapack-lib=<lib>, but
      ['-L/cygdrive/c/PROGRA~2/INTELS~1/COMPIL~1/windows/mkl/lib/intel64',
      'mkl_intel_lp64.lib', 'mkl_core.lib', 'mkl_intel_thread.lib']
      cannot be used<br>
*******************************************************************************<br>
      <br>
      Do I need to turn on the MKL sparse functionality? what is the
      difference between MKL_SPARSE, MKL_SPARSE_OPTIMIZE and
      MKL_SPARSE_SP2M in the configure options?<br>
      <br>
      Why is better to use sequential MKL? Is it possible to use a
      hybrid MPI - OpenMP approach? for example MPI for internode
      communication and OpenMP for intranode computation? In that case
      would it be good to use threaded MKL + MPI?<br>
      <br>
      Thanks,<br>
      <br>
      Hector<br>
      <br>
      <br>
      <br>
    </font>
    <div class="moz-cite-prefix">On 07/05/2018 08:54 PM, Satish Balay
      wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:alpine.LFD.2.21.1807052046000.29624@asterix">
      <pre wrap="">A few additional notes:

On Thu, 5 Jul 2018, Satish Balay wrote:

</pre>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite">
              <blockquote type="cite">
                <blockquote type="cite">
                  <blockquote type="cite">
                    <blockquote type="cite">
                      <blockquote type="cite">
                        <blockquote type="cite">
                          <blockquote type="cite">
                            <blockquote type="cite">
                              <blockquote type="cite">
                                <blockquote type="cite">
                                  <pre wrap="">
</pre>
                                </blockquote>
                              </blockquote>
                            </blockquote>
                          </blockquote>
                        </blockquote>
                      </blockquote>
                    </blockquote>
                  </blockquote>
                </blockquote>
              </blockquote>
            </blockquote>
          </blockquote>
        </blockquote>
        <pre wrap="">Using configure Options: --prefix=/cygdrive/c/installed/petsc_git-intel-debug/ --PETSC_DIR=/cygdrive/c/sources/petsc --PETSC_ARCH=windows-intel-debug --with-cc="win32fe cl" --with-fc="win32fe ifort" --with-mpi-include="[/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mpi/intel64/include]" --with-mpi-lib="[/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mpi/intel64/lib/release_mt/impi.lib]"
</pre>
      </blockquote>
      <pre wrap="">
</pre>
      <blockquote type="cite">
        <pre wrap=""> --with-openmp=1
</pre>
      </blockquote>
      <pre wrap="">
I don't think this option does anything [on windows]

</pre>
      <blockquote type="cite">
        <pre wrap="">--with-parmetis-lib=/cygdrive/c/installed/parmetis/lib/parmetis.lib --with-parmetis-include=/cygdrive/c/installed/parmetis/include --with-metis-lib=/cygdrive/c/installed/parmetis/lib/metis.lib --with-metis-include=/cygdrive/c/installed/parmetis/include --with-zoltan-include=/cygdrive/c/installed/zoltan/include --with-zoltan-lib=/cygdrive/c/installed/zoltan/lib/zoltan.lib --with-hypre-include=/cygdrive/c/installed/hypre/include --with-hypre-lib=/cygdrive/c/installed/hypre/lib/HYPRE.lib --with-blaslapack-lib="[/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64/mkl_intel_lp64.lib,/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64/mkl_core.lib,/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64/mkl_intel_thread.lib]"
</pre>
      </blockquote>
      <pre wrap="">
</pre>
      <blockquote type="cite">
        <pre wrap=""> --with-scalapack-include="/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/include" --with-scalapack-lib="[/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64/mkl_scalapack_lp64.lib,/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64/mkl_blacs_intelmpi_lp64.lib]"
</pre>
      </blockquote>
      <pre wrap="">
PETSc does not use scalapack or blacs. They are dependencies for MUMPS [which petsc has an interface to]

</pre>
      <blockquote type="cite">
        <pre wrap="">--with-shared-libraries=0 
</pre>
      </blockquote>
      <pre wrap="">
</pre>
      <blockquote type="cite">
        <pre wrap="">-LIBS=""/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64/mkl_intel_lp64.lib" "/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64/mkl_intel_thread.lib" "/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64//mkl_core.lib" "/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64/mkl_scalapack_lp64.lib" "/cygdrive/c/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64/mkl_blacs_intelmpi_lp64.lib" libiomp5md.lib"
</pre>
      </blockquote>
      <pre wrap="">
I'm not sure why you are having to respecify MKL via LIBS option.

Also with PETSc MPI usage - its best to use sequential MKL - and not threaded MKL - for eg:

</pre>
      <blockquote type="cite">
        <pre wrap="">--with-blaslapack-lib='-L/cygdrive/c/PROGRA~2/INTELS~1/COMPIL~2/windows/mkl/lib/intel64 mkl_intel_lp64_dll.lib mkl_sequential_dll.lib mkl_core_dll.lib'
</pre>
      </blockquote>
      <pre wrap="">
</pre>
      <blockquote type="cite">
        <blockquote type="cite">
          <pre wrap="">mhypre.c
C:\sources\petsc\src\mat\impls\hypre\mhypre.c(1453): warning C4002: too many actual parameters for macro 'hypre_TFree'
</pre>
        </blockquote>
      </blockquote>
      <pre wrap="">
You can switch to the compatible version of hypre - or fix revert the code change..

<a class="moz-txt-link-freetext" href="https://bitbucket.org/petsc/petsc/commits/e6de09342ce9c4562cc062ff2c1bac4bd956bda0">https://bitbucket.org/petsc/petsc/commits/e6de09342ce9c4562cc062ff2c1bac4bd956bda0</a>

Satish
</pre>
    </blockquote>
    <br>
  </body>
</html>