I need to create a number of vtkPartitionedDataSet object which is completely independent of each other.
Then, they will be added to vtkPartitionedDataSetCollection. How do I parallelize this process? Are there any out of box class I can use?
I was looking into vtkThreadedCompositeDataPipeline but it seems I still need to implement ThreadedExecute() method?
Shared data between threads has to be considered very carefully. Seems like AppendFilter is shared between threads - that is certain to cause problems. Advanced Capabilities and Future Work section in the linked articles gives you suggestions on how to use shared data and how to use VTK objects inside your threads. I would also browse the VTK source to see examples on how SMPTools is used.