Skip to content

GitLab

Closed
Created by Utkarsh Ayachit@utkarsh.ayachit⛰Owner

XML partitioned-dataset writer does not work in parallel correctly.

  • Connect to pvserver (2 ranks) using ParaView
  • Open can.ex2, apply
  • Create Adaptive Resample to Image filter, Apply.
  • Show as Surface and color by vtkProcessId to confirm there are as many blocks as ranks (2 in this case)
  • Save data as foo.vtpd

Now restart ParaView (as builtin) and load the file that was saved. Either it will fail to load correctly or will only have 1 block. This is a bug.

This is happening because the writer for vtkPartitionedDataSet is not parallel aware and each rank ends up overwriting what an other rank wrote.

Same is probably true for vtkPartitionedDataSetCollection.

Edited by Utkarsh Ayachit
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information