[petsc-dev] mkl_pardiso test

[Barry Smith]

   That file contains a real matrix, you built PETSc to use complex matrices. PETSc Mat/VecLoad can only load matrices/vectors as the same type as PETSc was built with.

   You can use any example in PETSc that generates matrices to run with mkl-paridiso you don’t need to read from a file.

  Barry

On Jun 26, 2014, at 8:54 PM, José David Bermeo López <jdbermeol at gmail.com> wrote:

> So I was adding a new test in src/mat/examples/tests/ using the file ex125.c, this test required data from ftp.mcs.anl.gov:/pub/petsc/matrices, I download the data, but I get the following error:
> 
> > [0]PETSC ERROR: --------------------- Error Message --------------------------------------------------------------
> > [0]PETSC ERROR: Read from file failed
> > [0]PETSC ERROR: Read past end of file
> > [0]PETSC ERROR: See http://www.mcs.anl.gov/petsc/documentation/faq.html for trouble shooting.
> > [0]PETSC ERROR: Petsc Development GIT revision:   GIT Date: 
> > [0]PETSC ERROR: ./ex125 on a linux-complex named conte-fe00.rcac.purdue.edu by jbermeol Thu Jun 26 21:36:20 2014
> > [0]PETSC ERROR: Configure options CC=mpiicc CXX=mpiicpc FC=mpiifort COPTFLAGS=-O3 CXXOPTFLAGS=-O3 FOPTFLAGS=-O3 --with-scalar-type=complex --with-shared-libraries=0 --with-debugging=1 --with-pic=1 --with-clanguage=C++ --with-fortran=1 --with-fortran-kernels=0 --download-mumps --download-metis --download-parmetis --with-scalapack-lib="-L/apps/rhel6/intel/composer_xe_2013.3.163/mkl/lib/intel64 -lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_lp64" --with-scalapack-include=/apps/rhel6/intel/composer_xe_2013.3.163/mkl/include --with-blas-lapack-dir=/apps/rhel6/intel/composer_xe_2013.3.163/mkl --with-mkl-pardiso-dir=/apps/rhel6/intel/composer_xe_2013.3.163/mkl
> > [0]PETSC ERROR: #1 PetscBinaryRead() line 274 in /home/jbermeol/software/petsc_cplx/src/sys/fileio/sysio.c
> > [0]PETSC ERROR: #2 MatLoad_SeqAIJ() line 4312 in /home/jbermeol/software/petsc_cplx/src/mat/impls/aij/seq/aij.c
> > [0]PETSC ERROR: #3 MatLoad() line 997 in /home/jbermeol/software/petsc_cplx/src/mat/interface/matrix.c
> > [0]PETSC ERROR: #4 main() line 36 in ex125.c
> > [0]PETSC ERROR: ----------------End of Error Message -------send entire error message to petsc-maint at mcs.anl.gov----------
> > application called MPI_Abort(MPI_COMM_WORLD, 66) - process 0
> 
> Not sure if the file is corrupted, could you please help me with that. Thanks 
> 
> 
> On Thu, Jun 26, 2014 at 3:38 PM, Satish Balay <balay at mcs.anl.gov> wrote:
> On Thu, 26 Jun 2014, José David Bermeo López wrote:
> 
> > Hi, I would like to add a test to the repository for mkl_pardiso, looking
> > for mumps test, I sow test ex125.c that uses MUMPS or other solvers
> > depending on the option, so i have some questions:
> >
> > 1. Would be a good idea to add mkl_pardiso in that test(ex125.c)
> 
> I've added one test. But more tests would be good.
> [check runex2_mkl_pardiso in src/ksp/ksp/examples/tutorials/makefile]
> 
> > 2. For that test you load matrix A, is it a distributed matrix or a
> > sequential matrix.
> 
> Hm - this example appears to also test Superlu [which is sequential] -
> so adding mkl_pardiso similarly might work..
> 
> [Hong might have better suggestions]
> 
> 
> > 3. You had html pages, for documentation of each solver, is this
> > automatically generated base on comments, or do I had to added my self.
> 
> Ah - I didn't check that. You can look at the end of
> mat/impls/aij/mpi/superlu_dist/superlu_dist.c for its
> documentation. If you add something similar to mkl_pardiso.c - the
> docs for it will get autogenerated.
> 
> Satish
>