Throughput Measurements using Event Trace - 2025.2 English - UG1079

AI Engine Kernel and Graph Programming Guide (UG1079)
Document ID
UG1079
Release Date
2025-11-26
Version
2025.2 English

Using event trace for throughput measurements requires that you compile the AI Engine application using AI Engine compiler option --event-trace=runtime. Then run the complete system in hardware using the XRT or XSDB flow to extract event trace.

You can visualize the trace data after extraction within vitis -a. An example of trace data is as follows:

You can position markers where you want to compute the data throughput. Use icons to precisely position on signal edges. Use icon to add a new marker at the exact position of the mobile marker.

Position the marker at the beginning of the kernel frame run to get the most precise result in throughput computation. The added marker should be positioned at the beginning of the first frame you want to take into account. Position the mobile marker on the first frame after the last one you want to take into account for the throughput computation:
Figure 1. First Frame and Last Frame Marker Position

Click the flag of the fixed marker which is on the first frame edge to define it as the origin time. Relative time is shown on the time scale and at the bottom of all other markers:

Figure 2. Relative Time Displayed on Time Scale and at Bottom of all Other Markers

Using the number of samples per kernel run and the number of frames processed, you can calculate average throughput over the first-to-last frame duration.