Getting Started
To excercise and walk through the different hardware debug methodologies, it is required to clone the git repository, get the design files ready to build, and generate the hardware image file (sd_card.img).
Make sure you set the environment variables as explained in this link.
Introduction
Designs running on AMD Versal™ AI Engine devices can target the AI Engine, programmable logic (PL), and Arm® host. To ensure a design targeting such multi-domain devices is functionally correct and meets the design performance specification, AMD recommends a five stage profile and debug methodology in hardware.
The stages are as follows:
Design Execution and System Metrics.
System Profiling.
PL Kernel Analysis.
AI Engine Event Trace and Analysis.
Host Application Debug.
Design Execution and System Metrics
This stage helps you determine:
If the design and host application can run successfully in hardware.
How to use APIs in your host application to profile the design as it is running on hardware.
Error reporting APIs to retrieve the asynchronous errors into the user-space host code.
If the design meets throughput, latency, and bandwidth goals.
In addition, you can troubleshoot AI Engine stalls and deadlocks using reports generated when running the design in hardware.
System Profiling
In this stage, you can profile the AI Engine Core, Interface, and Memory modules in the XRT or XSDB flows. It is a non-intrusive feature which can be enabled at runtime using the XRT.ini file or running scripts in XSDB. The feature uses performance counters available in the AI Engine array to gather profile data. The amount and type of data gathered is limited by the number of performance counters available.
PL Kernel Analysis
The goal of this stage is to determine the exact PL kernel causing a throughput drop by:
Inserting one or more integrated logic analyzers (ILAs) to monitor specific PL AXI interfaces to help identify exactly where and when a thorughput drop occurs.
Replacing the PL kernels.
Profiling using PL profile monitors.
AI Engine Event Trace and Analysis
This stage helps you determine the AI Engine kernel or graph construct causing the design performance drop or stall or causing a deadlock by:
Running and analyzing the runtime trace data using the AI Engine Event trace flow.
Profiling Intra-kernel performance.
Using the Vitis IDE debugger to debug the kernel source code.
Host Application Debug
The goal of this stage is to debug the host application and address application exceptions or crashes, if any exist using printf() or the Vitis IDE debugger.
Complete Hardware Profile and Debug Methodology
Support
GitHub issues will be used for tracking requests and bugs. For questions, go to support.xilinx.com.
Copyright © 2020–2024 Advanced Micro Devices, Inc