trouble testing/running with gnu/5.1.0
Jim Edwards
jedwards at ucar.edu
Mon Jul 20 14:39:32 CDT 2015
Hi Wei-keng,
This is an IBM machine with a poe based version of mpich - nether
mpich2version nor mpichversion are available. Since IBM has basically
abandoned poe I think that we can conclude that although it is mpich2 it's
also very old.
Thanks and thanks to Nick for confirming that gnu/5.1.0 is not the problem.
Jim
On Mon, Jul 20, 2015 at 1:33 PM, Wei-keng Liao <wkliao at eecs.northwestern.edu
> wrote:
> Hi, Jim
>
> From your config.log, I can see MPICH is used.
>
> Could you run command "mpich2version", which tells the MPICH version used
> in you MPI compiler, /glade/apps/opt/modulefiles/ys/cmpwrappers/mpicc
>
> If mpich2version is not available, please try mpichversion (this is for
> newer mpich).
>
> Wei-keng
>
> On Jul 20, 2015, at 1:53 PM, Jim Edwards wrote:
>
> > I think that you are correct I switched to pgi and got the same error -
> I can try on another platform.
> >
> > The pgi config.log is attached...
> >
> >
> > On Mon, Jul 20, 2015 at 12:48 PM, Wei-keng Liao <
> wkliao at eecs.northwestern.edu> wrote:
> > Hi, Jim
> >
> > I suspect it is caused by an old MPI.
> > Can you send me your config.log file?
> >
> > Wei-keng
> >
> > On Jul 20, 2015, at 12:36 PM, Jim Edwards wrote:
> >
> > > I am getting an error when trying to build with gnu/5.1.0
> > > the traceback looks like:
> > > #6 0x00000000014fb478 in ncmpii_in_swapn ()
> > > #7 0x00000000014cd6be in ncmpii_getput_vars ()
> > > #8 0x00000000014d9b64 in ncmpi_put_vara_int_all ()
> > > #9 0x00000000014baf57 in nfmpi_put_vara_int_all_ ()
> > > #10 0x00000000013f61e2 in __pionfput_mod_MOD_put_vara_1d_int ()
> > >
> > > If I try to run the tests with this build I get a hang in the
> sequentiol tests:
> > > poe ./nc_test -v -d .
> > >
> > > 2:*** Testing test_ncmpi_strerror ... 33 good
> comparisons. ok
> > >
> > > 0:*** Testing test_ncmpi_strerror ... 33 good
> comparisons. ok
> > >
> > > 1:*** Testing test_ncmpi_strerror ... 33 good
> comparisons. ok
> > >
> > > 3:*** Testing test_ncmpi_strerror ... 33 good
> comparisons. ok
> > >
> > > 0:
> > >
> > > 1:
> > >
> > > 2:
> > >
> > > 2: FAILURE at line 149 of test_read.c: remove of ./scratch.nc
> failed*** Testing test_ncmpi_open ...
> > >
> > > 2: ### 1 FAILURES TESTING test_ncmpi_open! Stop ... ###
> > >
> > > 0:*** TESTING C ./nc_test for format CDF-1
> ------ failed
> > >
> > > 0:
> > >
> > > 0: FAILURE at line 149 of test_read.c: remove of ./scratch.nc
> failed*** Testing test_ncmpi_open ...
> > >
> > > 0: ### 1 FAILURES TESTING test_ncmpi_open! Stop ... ###
> > >
> > > 0:
> > >
> > > 0:./nc_test: expects 0 failures ... Total number of failures: 1
> > >
> > > 1:*** TESTING C ./nc_test for format CDF-1
> ------ failed
> > >
> > > 1:
> > >
> > > 1: FAILURE at line 149 of test_read.c: remove of ./scratch.nc
> failed*** Testing test_ncmpi_open ...
> > >
> > > 1: ### 1 FAILURES TESTING test_ncmpi_open! Stop ... ###
> > >
> > > 1:
> > >
> > > 1:./nc_test: expects 0 failures ... Total number of failures: 1
> > >
> > > 2:*** TESTING C ./nc_test for format CDF-1
> ------ failed
> > >
> > > 2:
> > >
> > > 2:
> > >
> > > 2:./nc_test: expects 0 failures ... Total number of failures: 1
> > >
> > > 3:*** Testing test_ncmpi_open ... 5 good
> comparisons. ok
> > >
> > > and the parallel test fail in the build step with:
> > >
> > > glade/apps/opt/modulefiles/ys/cmpwrappers/mpif90 -c -g -O0
> -I../../src/lib -I../../src/libf -I../../src/libf90 mcoll_testf.f90
> > >
> > > mcoll_testf.f90:257:22:
> > >
> > >
> > >
> > > t1 = MPI_Wtime ( )
> > >
> > > 1
> > >
> > > Error: Name 'mpi_wtime' at (1) is an ambiguous reference to
> 'mpi_wtime' from module 'mpi_constants'
> > >
> > > mcoll_testf.f90:292:22:
> > >
> > >
> > >
> > > t2 = MPI_Wtime ( )
> > >
> > > 1
> > >
> > > Error: Name 'mpi_wtime' at (1) is an ambiguous reference to
> 'mpi_wtime' from module 'mpi_constants'
> > >
> > > mcoll_testf.f90:313:20:
> > >
> > >
> > >
> > > t3 = MPI_Wtime ( )
> > >
> > > 1
> > >
> > > Error: Name 'mpi_wtime' at (1) is an ambiguous reference to
> 'mpi_wtime' from module 'mpi_constants'
> > >
> > > All with svn revision 2056
> > >
> > >
> > >
> > > --
> > > Jim Edwards
> > >
> > > CESM Software Engineer
> > > National Center for Atmospheric Research
> > > Boulder, CO
> >
> >
> >
> >
> > --
> > Jim Edwards
> >
> > CESM Software Engineer
> > National Center for Atmospheric Research
> > Boulder, CO
> > <config.log.gz>
>
>
--
Jim Edwards
CESM Software Engineer
National Center for Atmospheric Research
Boulder, CO
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/parallel-netcdf/attachments/20150720/5ccc2a84/attachment-0001.html>
More information about the parallel-netcdf
mailing list