[gpaw-users] Fwd: can't using gpaw-python

[Marcin Dulak]

Hi,

On 10/24/2013 10:59 AM, 謝其軒 wrote:
> I've compile gpaw without any reference to mkl library.
> I export $PATH and $LD_LIBRARY_PATH myself and I got gpaw well compiled.
> Then I gpaw-test it in the node with 8 cores. All tests passed but 
> with some parallel part acting strange:
> ===================
>
> [z955018 at node05 test]$ mpiexec -np 4 gpaw-python `which gpaw-test` 
> 2>&1 | tee testgpaw.log
>
> --------------------------------------------------------------------------
> An MPI process has executed an operation involving a call to the
> "fork()" system call to create a child process.  Open MPI is currently
> operating in a condition that could result in memory corruption or
> other system errors; your MPI job may hang, crash, or produce silent
> data corruption.  The use of fork() (or system() or other calls that
> create child processes) is strongly discouraged.
>
> The process that invoked fork was:
>
>   Local host:          node05 (PID 7485)
>   MPI_COMM_WORLD rank: 3
>
> If you are *absolutely sure* that your application will successfully
> and correctly survive a call to fork(), you may disable this warning
> by setting the mpi_warn_on_fork MCA parameter to 0.
> --------------------------------------------------------------------------
> python 2.7.5 GCC 4.1.2 20080704 (Red Hat 4.1.2-52) 64bit ELF on Linux 
> x86_64 redhat 5.8 Final
> Running tests in /tmp/gpaw-test-ujm_vy
> Jobs: 1, Cores: 4, debug-mode: False
> =============================================================================
> gemm_complex.py                         0.014  OK
> mpicomm.py                              0.010  OK
> ase3k_version.py                        0.007  OK
>
> ....................
>
>
> integral4.py                            0.025  OK
> parallel/ut_parallel.py            [node05:07477] 3 more processes 
> have sent help message help-mpi-runtime.txt / mpi_init:warn-fork
> [node05:07477] Set MCA parameter "orte_base_help_aggregate" to 0 to 
> see all help / error messages
>      1.074  OK
> transformations.py                      0.017  OK
> parallel/parallel_eigh.py               0.010  OK
> spectrum.py                             0.037  OK
> xc.py                                       0.064  OK
> ......................
>
> parallel/realspace_blacs.py             0.018  OK
> AA_exx_enthalpy.py                    139.244  OK
> cmrtest/cmr_test.py                     0.023  SKIPPED
> cmrtest/cmr_test3.py                    0.012  SKIPPED
> cmrtest/cmr_test4.py                    0.018  SKIPPED
> cmrtest/cmr_append.py                   0.015  SKIPPED
> cmrtest/Li2_atomize.py                  0.015  SKIPPED
> =============================================================================
> Ran 223 tests out of 230 in 2511.9 seconds
> Tests skipped: 7
> All tests passed!
> =============================================================================
> The ground state calculation of parallel version is fine.
>
> But when I run the DF calculations, it seems that something familiar 
> emerges up:
i think this uses scipy, so libmkl_intel_thread.so is coming probably 
from scipy:
python -c "import scipy; print scipy.__config__.show(); print 
scipy.__version__"
If so rebuild scipy. Scipy uses the settings from available numpy (the 
one you get with python -c "import numpy; print numpy") for build.

Marcin
>
> python2.7: symbol lookup error: 
> /opt/intel/composerxe-2011.3.174/mkl/lib/intel64/libmkl_intel_thread.so: 
> undefined symbol: omp_get_num_procs
> (this is run with serial version.....and it pissed me off)
>
> ===============================================================================
> --------------------------------------------------------------------------
> An MPI process has executed an operation involving a call to the
> "fork()" system call to create a child process.  Open MPI is currently
> operating in a condition that could result in memory corruption or
> other system errors; your MPI job may hang, crash, or produce silent
> data corruption.  The use of fork() (or system() or other calls that
> create child processes) is strongly discouraged.
>
> The process that invoked fork was:
>
>   Local host:          node03 (PID 24987)
>   MPI_COMM_WORLD rank: 3
>
> If you are *absolutely sure* that your application will successfully
> and correctly survive a call to fork(), you may disable this warning
> by setting the mpi_warn_on_fork MCA parameter to 0.
> --------------------------------------------------------------------------
> [node03:24983] 7 more processes have sent help message 
> help-mpi-runtime.txt / mpi_init:warn-fork
> [node03:24983] Set MCA parameter "orte_base_help_aggregate" to 0 to 
> see all help / error messages
> gpaw-python: symbol lookup error: 
> /opt/intel/composerxe-2011.3.174/mkl/lib/intel64/libmkl_intel_thread.so: 
> undefined symbol: omp_get_num_procs
> gpaw-python: symbol lookup error: 
> /opt/intel/composerxe-2011.3.174/mkl/lib/intel64/libmkl_intel_thread.so: 
> undefined symbol: omp_get_num_procs
> gpaw-python: symbol lookup error: 
> /opt/intel/composerxe-2011.3.174/mkl/lib/intel64/libmkl_intel_thread.so: 
> undefined symbol: omp_get_num_procs
> gpaw-python: symbol lookup error: 
> /opt/intel/composerxe-2011.3.174/mkl/lib/intel64/libmkl_intel_thread.so: 
> undefined symbol: omp_get_num_procs
> gpaw-python: symbol lookup error: 
> /opt/intel/composerxe-2011.3.174/mkl/lib/intel64/libmkl_intel_thread.so: 
> undefined symbol: omp_get_num_procs
> gpaw-python: symbol lookup error: 
> /opt/intel/composerxe-2011.3.174/mkl/lib/intel64/libmkl_intel_thread.so: 
> undefined symbol: omp_get_num_procs
> gpaw-python: symbol lookup error: 
> /opt/intel/composerxe-2011.3.174/mkl/lib/intel64/libmkl_intel_thread.so: 
> undefined symbol: omp_get_num_procs
> gpaw-python: symbol lookup error: 
> /opt/intel/composerxe-2011.3.174/mkl/lib/intel64/libmkl_intel_thread.so: 
> undefined symbol: omp_get_num_procs
> --------------------------------------------------------------------------
> mpirun has exited due to process rank 6 with PID 24990 on
> node node03 exiting without calling "finalize". This may
> have caused other processes in the application to be
> terminated by signals sent by mpirun (as reported here).
> --------------------------------------------------------------------------
>
> =============================================================================
>
> And this is the parallel version.
>
> I think right now I might try to build with mkl library or I'd like to 
> kill myself....
>
> BR,
>
> chi-hsuan
>
>


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20131024/33d91dca/attachment.html>