I am considering using oneTBB’s flow_graph as a framework for a dependency graph I am building.
The grunt work done in the oneTBB nodes are performed via VTK code (mostly vtkAlogrithm derived classes)
Will there be architectural/design conflicts with the above thinking ? Will VTK’s vtkExecutive (or it’s execution, data-flow management infrastructure) have contentious issue with oneTBB’s flow_graph buffering/caching of information between nodes ?
Has anyone done something similar (not necessarily with oneTBB) but some external dependency graph driving the data flow but using VTK within a node to do mainly the compute work. Was there any issue when a VTK node reads it’s input e.g. delays, unnecessary data buffering/duplication.
I am trying to wrap my head around which system (oneTBB/VTK) should be responsible for managing upstream node’s data changes propagation ?
Kind regards