The AI Engine provides the ability to output a summary of AI Engine status, which includes error events that have occurred. When a deadlock is detected in the hardware, warnings are also issued. The status and warnings can be further analyzed in the AMD Vitis™ IDE for debug purposes.
There are two ways to output AI Engine status when running designs in hardware:
- Automated and periodic AI Engine status output: After initial setup in xrt.ini, this method requires minimal user intervention, because the tool outputs periodic status at specified time intervals.
- Manually output the AI Engine status: Run a command each time you want a status output report.
You can then open the status output in the IDE for further analysis.
This section covers the steps needed to obtain status output, and analysis of that output. For more details on understanding stalls, see AI Engine Stall Analysis in the Vitis IDE.