Skip to content

GitLab

Open
Created by Andrew Bauer@acbauerDeveloper

partitioned dataset writer doesn't work with Catalyst V2 in parallel

The partitioned dataset writer is mangling the meta-file when outputting from Catalyst V2 in parallel. I'm using the Catalyst2/CxxImageDataExample and running with mpirun -np 4 bin/CxxImageDataExampleV2 --output datasets/dataset-%04ts.vtpd. The dataset-0001.vtpd file looks like:

<VTKFile type="vtkPartitionedDataSet" version="1.0" byte_order="LittleEndian" header_type="UInt64">
  <vtkPartitionedDataSet>
    <DataSet index="0" file="dataset-0001/dataset-0001_0.vti"/>
    <DataSet ind format="binary" RangeMin="0.1" RangeMax="0.1">
      CAAAAAAAAACamZmZmZm5Pw==
    </DataArray>
    <DataArray type="Float64" Name="time" NumberOfTuples="1" format="binary" RangeMin="0.1" RangeMax="0.1">
      CAAAAAAAAACamZmZmZm5Pw==
    </DataArray>
    <DataArray type="Int32" Name="cycle" NumberOfTuples="1" format="binary" RangeMin="1" RangeMax="1">
      BAAAAAAAAAABAAAA
    </DataArray>
    <DataArray type="Int32" Name="timestep" NumberOfTuples="1" format="binary" RangeMin="1" RangeMax="1">
      BAAAAAAAAAABAAAA
    </DataArray>
    <Array type="String" Name="__CatalystChannel__" NumberOfTuples="1" format="binary">
      BQAAAAAAAABncmlkAA==
    </Array>
  </FieldData>
</VTKFile>
e="Int32" Name="timestep" NumberOfTuples="1" format="binary" RangeMin="1" RangeMax="1">
      BAAAAAAAAAABAAAA
    </DataArray>
    <Array type="String" Name="__CatalystChannel__" NumberOfTuples="1" format="binary">
      BQAAAAAAAABncmlkAA==
    </Array>
  </FieldData>
</VTKFile>
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information