A Channel is the flow data going into and out of each Processes. Channels are named after the variable defining them in the C source code, they are usually arrays, pointers or hls::streams. The Channel table describes each Channel in a separate row.
The details available on each row are:
Channel name
C/C++ declaration
Data volume expected to be delivered over the channel - measured from C-Simulation run
Estimated throughput: data volume divided by the estimated Transaction Interval of the current region
Channel’s bitwidth
Access mode from producer process to consumer process
Producer and Consumer processes names
The Channel table rows and graph edges are cross-linked: selecting one select will the other and vice-versa. You can highlight multiple channels with CTRL+CLICK. It is also possible to sort the table columns or search in the table.
NOTE: The average throughput is the channel volume divided by the region’s transaction interval, so it is meaningful to compare with other channels throughput values, for one given dataset and implementation. However, changing the C code or C-TB inputs might change transaction intervals or volume and hence change the throughput values, so be careful when comparing throughput of different runs.