[petsc-users] Compiling with Intel MPI options

Barry Smith bsmith at mcs.anl.gov
Thu Jul 20 17:47:04 CDT 2017


  Steven,

   We are linking against a non-existant library as a test of the linker, the linker is returning without an error code. This means that we cannot trust the error codes returned from the linker which means that our configure tests will not work. 
  
    First thing to do is check if you have any later versions of Intel MPI compilers installed on your systems or if there are any updates to the compilers. Check with your systems support if you don't install the compilers etc yourself.

   Barry



> On Jul 20, 2017, at 5:37 PM, Steven Wong <steven_wildman_wong at hotmail.com> wrote:
> 
> If I use on a Centos 7.2 system
> ./configure --with-cc=mpiicc --with-cxx=mpiicpc --with-fc=mpiifort
> --with-blas-lapack-dir=/opt/intel/mkl/lib/intel64
> 
> I get 
> ===============================================================================
>             Configuring PETSc to compile on your
> system                       
> ===============================================================================
> TESTING: checkCCompiler from
> config.setCompilers(config/BuildSystem/config/setCompilers.py:553)
> *******************************************************************************
>         UNABLE to CONFIGURE with GIVEN OPTIONS    (see configure.log
> for details):
> -------------------------------------------------------------------------------
> C compiler you provided with -with-cc=mpiicc does not work.
> C compiler mpiicc is broken! It is returning a zero error when the
> linking failed! Either
> 1) switch to another compiler suite or
> 2) report this entire error message to your compiler/linker suite
> vendor and ask for fix for this issue.
> *******************************************************************************
> (configure.log attached)
> 
> However, if I use 
> ./configure --with-cc=icc --with-cxx=icpc --with-fc=ifort
> --with-blas-lapack-dir=/opt/intel/mkl/lib/intel64 --with-mpi=0
> All is fine (see configure.log.nompi)
> 
> We don't have issues using mpiifort or mpiicc if I compile my own code.
> What are we missing?  Thanks in advance.
> <configure.log><configure.log.nompi>



More information about the petsc-users mailing list