Migrating Command-line Projects to Vitis Unified IDE - 2023.2 English

In the AI Engine application development flow, it is common to build and simulate the AI Engine graph application using the command-line (Makefile or config file-based approach). However, to debug and analyze the results, it is seamless to use the AMD Vitis™ IDE. The AMD Vitis™ unified IDE provides a user managed flow to directly consume the pre-compiled results from command-line to the IDE for debug and analysis. This section provides a methodology to port your AI Engine graph application from command line interface to the Vitis unified IDE.

Below are the steps to port your command-line project to the Vitis unified IDE.

1. Open the Vitis unified IDE, and select Open Workspace. Browse to the directory where the AI Engine graph application is built.
2. At this stage, the Vitis unified IDE identifies no IDE managed components and provides options either to create components from scratch or switch to user managed mode as shown below.
   
   
3. When you select User Managed Mode, this opens the directory view as shown below. Highlighted are some output directories that are available after AI Engine compilation and simulation.
   
   

   After you enter the user-managed flow in Vitis Unified IDE, you can use different capabilities of the IDE. Say, crossprobing between microcode and the source code, using Analysis view, Debug AI Engine code, re-build the AI Engine graph.

   From the AI Engine compile output Work/ directory which got generated using the command line flow, you can cross probe between the lines in the microcode (marked in blue dot) and the source code, as highlighted below.

   
   
   
4. When you switch to the Analysis view, all the summary files in that workspace are shown up for easy analysis as shown below.
5. You can debug the AI Engine code in the user managed mode by creating a launch configuration.
   1. Right-click in the Explorer view.
   2. Select Edit launch configuration and add New launch configuration.
   3. In the Create Launch Configuration window, specify the Build Directory that contains the AI Engine compile summary files, and click Submit.
   4. Fill the launch configuration name, and enable the trace and profile settings, if necessary.