The DATAFLOW optimization tries to create task-level parallelism between the various functions in the code on top of the loop-level parallelism where possible. The optimization is done in two parts — first, the compiler infers the parallel tasks and sets up the communication channels and the associated control; second, after the generated RTL design is simulated, you can analyze the performance of the DATAFLOW optimization. Due to this fact, the Dataflow viewer lets you see the dataflow structure inferred by the tool, inspect the channels (FIFO/PIPO), and examine the effect of channel depth on performance. Performance data is then back-annotated to the Dataflow viewer from the co-simulation results. The Dataflow viewer also displays a representation of the dataflow graph structure, showing the different processes and the underlying producer-consumer connections.
TIP: You must apply the DATAFLOW pragma or directive to your design for the Dataflow viewer to be populated. You can apply dataflow to the top-level function, or specify regions of a function, or loops. For more information, refer to set_directive_dataflow in the Vitis HLS flow of the Vitis Unified Software Platform documentation (UG1416).