[MOAB-dev] running pcomm_unit
de Almeida, Valmor F.
dealmeidav at ornl.gov
Tue Jun 17 13:19:04 CDT 2008
When you have time....
I did this:
->mpirun -n 4 mbparallelcomm_test 0 3 test/h5file/ptest.cub
Because I could not find the partest.cub file. These are the cub files
(moab rev. 1902) I have
./test/brick_cubit10.cub
./test/h5file/ptest.cub
./test/brick_cubit10.1.cub
./test/brick_cubit10.2.cub
The result follows below. It seems the read is fine but the processor
boundaries have problems.
Also, when I build moab I do a make check which, for the most part,
passes the tests. (there are some HDF5-DIAG: Error detected in HDF5
(1.8.1) thread 0: .....; I can send you the output if you like to see
it.)
Thanks,
--
Valmor
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
->mpirun -n 4 mbparallelcomm_test 0 3 test/h5file/ptest.cub
Reading file test/h5file/ptest.cub
Broadcasting mesh.
Broadcasting mesh.
Broadcasting mesh.
Checking global ids.
Getting fileset entities.
Broadcasting mesh.
Bcast done; entities:
Number of entities per type:
Vertex: 153
Bcast done; entities:
Number of entities per type:
Vertex: 153
Edge: 300
Tri: 0
Quad: 164
Polygon: 0
Tet: 0
Pyramid: 0
Prism: 0
Knife: 0
Hex: 64
Polyhedron: 0
EntitySet: 164
Getting fileset entities.
Deleting nonlocal entities.
My partition sets: EntitySet 67109020
Bcast done; entities:
Number of entities per type:
Vertex: 153
Edge: 300
Tri: 0
Quad: 164
Polygon: 0
Tet: 0
Pyramid: 0
Prism: 0
Knife: 0
Hex: 64
Polyhedron: 0
EntitySet 67109024
EntitySet: 164
Getting fileset entities.
Deleting nonlocal entities.
EntitySet 134217885
EntitySet 134217889
Gathering related entities.
Number of entities per type:
Vertex: 153
Edge: 300
Tri: 0
Quad: 164
Polygon: 0
Tet: 0
Bcast done; entities:
EntitySet 67109028
My partition sets: Pyramid: 0
Prism: 0
Knife: 0
Hex: 64
Polyhedron: 0
EntitySet: 164
Getting fileset entities.
Deleting nonlocal entities.
EntitySet 201326750
EntitySet 201326754
Gathering related entities.
Edge: 300
My partition sets: Gathering related entities.
Tri: 0
Quad: 164
Polygon: 0
Tet: 0
Pyramid: 0
Prism: 0
Knife: 0
Hex: 64
Polyhedron: 0
EntitySet: 164
Deleting nonlocal entities.
My partition sets: EntitySet 155
EntitySet 159
EntitySet 163
Gathering related entities.
Removing deletable entities from keepable sets.
Deleting deletable entities.
Exchanging ghost entities.
Couldn't read mesh; error message:
Removing deletable entities from keepable sets.
Deleting deletable entities.
Failed in step PARALLEL EXCHANGE_GHOSTS
Failed to find iface tag.
Failed to get iface sets, procs
Exchanging ghost entities.
Couldn't read mesh; error message:
Failed in step PARALLEL EXCHANGE_GHOSTS
Failed to find iface tag.
Failed to get iface sets, procs
Removing deletable entities from keepable sets.
Deleting deletable entities.
Exchanging ghost entities.
Couldn't read mesh; error message:
Failed in step PARALLEL EXCHANGE_GHOSTS
Failed to find iface tag.
Failed to get iface sets, procs
Removing deletable entities from keepable sets.
Deleting deletable entities.
Exchanging ghost entities.
Couldn't read mesh; error message:
Failed in step PARALLEL EXCHANGE_GHOSTS
Failed to find iface tag.
Failed to get iface sets, procs
Proc 1: Success.
Proc 2: Success.
Proc 3: Success.
Proc 0: Success.
Times: 0.0525429 0.0525398 -1.21373e+09 3.86615e-39 1.21373e+09
(total/read/shared/report/delete)
> -----Original Message-----
> From: Tim Tautges [mailto:tautges at mcs.anl.gov]
> Sent: Tuesday, June 17, 2008 10:56 AM
> To: de Almeida, Valmor F.; moab-dev at mcs.anl.gov
> Subject: Re: [MOAB-dev] running pcomm_unit
>
> You shouldn't be getting these results. Right now I'm working on a
> deadline that'll be challenging, but I'll try to check this out. One
> thing you can try to verify parallel functionality is build & run
> mbparallelcomm_test, in the main source directory:
>
> make mbparallelcomm_test
>
> mpirun -n 4 mbparallelcomm_test 0 3 test/partest.cub
>
> That will pick up test/partest.cub, use the MATERIAL_SET entity sets
as
> the decomposition, read and distribute the mesh, delete the non-owned
> mesh, and establish shared interfaces. I'm pretty sure this doesn't
> work at the moment because of a commit I did last week though.
>
> - tim
>
More information about the moab-dev
mailing list