Description
Specifies that dataflow optimization be performed on the functions or loops, improving the concurrency of the RTL implementation.
All operations are performed sequentially in a C/C++ description. In
the absence of any directives that limit resources (such as set_directive_allocation), Vitis HLS seeks to minimize latency and improve concurrency. Data dependencies can limit
this. For example, functions or loops that access arrays must finish all read/write
accesses to the arrays before they complete. This prevents the next function or loop
that consumes the data from starting operation.
It is possible for the operations in a function or loop to start operation before the previous function or loop completes all its operations. When the DATAFLOW optimization is specified, the HLS tool analyzes the dataflow between sequential functions or loops and creates channels (based on ping-pong RAMs or FIFOs) that allow consumer functions or loops to start operation before the producer functions or loops have completed. This allows functions or loops to operate in parallel, which decreases latency and improves the throughput of the RTL.
config_dataflow command specifies the default memory
channel and FIFO depth used in DATAFLOW optimization.For the DATAFLOW optimization to work, the data must flow through the design from task to task. One should pay special attention to the following situations. Refer to Limitations of Control-Driven Task-Level Parallelism and Dataflow Region Coding Style for more details.
- Single-producer-consumer violations
- Feedback between tasks
- Conditional execution of tasks
- Loops with multiple exit conditions
Finally, the DATAFLOW optimization is not hierarchical by default. If a sub-function or loop contains additional tasks that might benefit from the DATAFLOW optimization, you must apply the optimization to the loop, the sub-function, or inline the sub-function.
Syntax
set_directive_dataflow <location> -disable_start_propagation-
<location>is the location (in the formatfunction[/label]) at which dataflow optimization is to be performed. -
-disable_start_propagationdisables the creation of a start FIFO used to propagate a start token to an internal process. Such FIFOs can sometimes be a bottleneck for performance.
Examples
Specifies dataflow optimization within function foo.
set_directive_dataflow foo