[MOAB-dev] question about mbzoltan
Iulian Grindeanu
iulian at mcs.anl.gov
Thu Sep 1 12:22:16 CDT 2011
Hello,
I have a small file that I am partitioning in 2, with mbpart
you can get mbpart if you configure moab with --enable-mbzoltan --with-zoltan=/home/fathom/3rdparty/Zoltan_v3.3 --with-parmetis=/home/fathom/3rdparty/ParMETIS3_1/gcc (at least on mcs network)
So the command I am using is
mbpart 2 MeshFiles/unittest/iGeom/shell.h5m shell_p2.h5m
It seems to be working fine, the mesh is partitioned in 2, there are some tags defined in shell_p2.h5m
But one existing variable length tag, that should have not been modified at all, has one wrong value.
Do you have any idea what am I doing wrong? Or is this a bug in mbpart?
Thanks,
Iulian
So, shell.h5m has this tag:
GROUP "GEOM_SENSE_N_ENTS" {
COMMENT "GEOM_SENSE_N_ENTS"
ATTRIBUTE "class" {
DATATYPE H5T_STD_I32LE
DATASPACE SCALAR
DATA {
(0): 1
}
}
ATTRIBUTE "is_handle" {
DATATYPE H5T_STD_I32LE
DATASPACE SCALAR
DATA {
(0): 1
}
}
ATTRIBUTE "variable_length" {
DATATYPE H5T_STD_I32LE
DATASPACE SCALAR
DATA {
(0): 1
}
}
DATASET "id_list" {
DATATYPE H5T_STD_U64LE
DATASPACE SIMPLE { ( 7 ) / ( 7 ) }
DATA {
(0): 64, 65, 66, 67, 68, 69, 70
}
}
DATATYPE "type" H5T_STD_U64LE;
DATASET "values" {
DATATYPE "/tstt/tags/GEOM_SENSE_N_ENTS/type"
DATASPACE SIMPLE { ( 8 ) / ( 8 ) }
DATA {
(0): 62, 63, 62, 62, 62, 63, 63, 63
}
}
the output file has a wrong value for the first set, a zero instead of original 62.
GROUP "GEOM_SENSE_N_ENTS" {
COMMENT "GEOM_SENSE_N_ENTS"
ATTRIBUTE "class" {
DATATYPE H5T_STD_I32LE
DATASPACE SCALAR
DATA {
(0): 1
}
}
ATTRIBUTE "is_handle" {
DATATYPE H5T_STD_I32LE
DATASPACE SCALAR
DATA {
(0): 1
}
}
ATTRIBUTE "variable_length" {
DATATYPE H5T_STD_I32LE
DATASPACE SCALAR
DATA {
(0): 1
}
}
DATASET "id_list" {
DATATYPE H5T_STD_U64LE
DATASPACE SIMPLE { ( 7 ) / ( 7 ) }
DATA {
(0): 64, 65, 66, 67, 68, 69, 70
}
}
DATATYPE "type" H5T_STD_U64LE;
DATASET "values" {
DATATYPE "/tstt/tags/GEOM_SENSE_N_ENTS/type"
DATASPACE SIMPLE { ( 8 ) / ( 8 ) }
DATA {
(0): 0, 63, 62, 62, 62, 63, 63, 63
}
}
More information about the moab-dev
mailing list