The 8-engine design was also run in hardware on the VCK190 evaluation board. Once again, NPULSE_USE=586 and NFRAME=16 was used to run a full 16 frames with the full compliment of radar pulses for each. The throughput was measured using the xrt::graph::get_timestamp() function as before. A screenshot captured from the VCK190 board run is shown below.
The final throughput using
xrt::graph::get_timestamp()is 18.9 frames per second. This is only slighty lower than 8X times the frame rate achieved with the single engine, indicating the data flow remains efficient even with eight separate engines.