CreateIDESystemProj.md

AI Engine Development See Vitis™ Development Environment on xilinx.com See Vitis™ AI Development Environment on xilinx.com

Porting a Command Line Project to the Vitis IDE Project

This section walks you through the flow of porting your command line project to the Vitis IDE project. One advantage of porting the design from command line to the IDE project is to use the debug capabilities of the GUI-based project both at an AI Engine level and at a system level.

• Before beginning this section of tutorial, it is expected that you have run the steps to set the environment variables as described in Introduction.
• Clone the GIT repository, and navigate to the 09-debug-walkthrough tutorial directory.

Creating the Application Project

1. Open the Vitis IDE tool by typing vitis in the console, set the workspace, and the IDE opens with Welcome page.

2. Select the Create Application Project and click Next.

3. Select the xilinx_vck190_base_20xxx0_1 platform from the list of platforms, and click Next.

4. Enter the Application project as PeakDetect, and click Next.

5. Leave the domain as aiengine, and click Next.

6. Select the Empty Application and Finish.

7. The Application project gets created as follows.

   

Preparing the AI Engine Domain Project

1. Expand the PeakDetect AI Engine domain project, right-click the src directory, and choose Import Source.
2. Browse to the cmd_src/aie/ directory, and tick the /aie on left side, that enables graph.h,graph.cpp files along with the kernel sources in kernels/ directory.
3. Click Finish, and observe the sources added to the PeakDetect AI Engine domain project.
4. Now add the input stimulus. Right-click the data/ directory -> Import sources -> browse to cmd_src/data directory, and choose the inx.txt file.
5. Right-click the AI Engine domain project, and select the C++ Build settings. Navigate to the Input Spec, and add ../src/kernels in the include paths. >
6. You can click No in the dialog box that opens. Double-click the PeakDetect.prj file under the AI Engine domain project, and click Click to select Top-level File. Expand the PeakDetect/src/, and select the graph.cpp file. Click OK.

The AI Engine domain project is ready to build.

NOTE: When you change the build target from Emulation-AIE(default) to Emulation-SW, make sure to repeat the step 5.

Adding PL Kernels

In this step, you will add the PL kernels, s2mm and mm2s.

1. To add the PL kernels, right-click the PeakDetect_system project, and select Add Hw kernel Project.
2. Name the Hw kernel project name as PeakDetect_PL, and click Finish.
3. Observe the PeakDetect_PL gets added to the system project.
4. Right-click the PeakDetect_PL -> src, and select Import sources. Browse to cmd_src/pl_kernels/, and select Open.
5. Select both mm2s.cpp and s2mm.cpp. Click the Finish button.
6. Double click the PeakDetect_PL -> PeakDetect_PL.prj, and in the project settings, select the Add Hardware Function.
7. Choose the mm2s and s2mm functions, and click OK. Observe the hardware function gets added as follows.

Prepare the HW-Link Project

1. Double click the PeakDetect_system_hw_link -> PeakDetect_system_hw_link.prj, and change the number of compute units for s2mm as 2 by double-clicking and editing the value. This is because for the design, there is one input PLIO port and two output PLIO ports.

2. Specify the system.cfg file that contains the connections between the PL and AI Engine kernels. To do this, right-click the binary_container_1, and select Edit V++ Options.

   

   Add the v++ command line option --config ../system.cfg as shown below and click Apply and Close.

   

3. Right-click the PeakDetect_system_hw_link, and select Import sources.

4. Browse to the cmd_src directory, and choose the system.cfg file. In the Into folder field, browse to the hw_link project.

5. Click Finish.

6. While creating a HW-link project, by default, the Vitis IDE tool creates the binary_container_1-link.cfg file under {$PROJECT}/{BUILD_TARGET}/ that contains the connectivity.

   [connectivity]
   nk=mm2s:1:mm2s_1
   nk=s2mm:2:s2mm_1.s2mm_2
   

   To port a command line project to a Vitis IDE environment, make sure to remove the above connectivity statements that starts with nk in your system.cfg file. Also make sure the compute unit names in system.cfg file matches with the names mentioned in connectivity. To do this, use the following steps.

7. Double-click the system.cfg file, and modify the sc=mm2s.s:ai_engine_0.inx to sc=mm2s_1.s:ai_engine_0.inx.

Host Application Project

1. To add the host application project, right-click the PeakDetect_system project, and select Add Application Project.

2. Enter the Application project name as PeakDetect_host, and leave the target processor selected as psv-cortexa72 SMP.

3. In the Application settings, add sysroot path,Root FS and Kernel Image by clicking Browse in each section, and click Next.

4. Select Empty Application in the Template selection window, and click Finish.

5. Now import sources by right-clicking the PeakDetect_host -> src, browse to the cmd_src/sw/ directory, and choose the files host.cpp and data.h.

6. Click Finish.

7. Now, right-click the PeakDetect_host, select C++ build settings, and add linker option

8. Click Apply and Close.

   NOTE: When you change the build target from Emulation-SW(default) to Emulation-HW or Hardware, make sure to repeat step 7.

System Project Settings

1. Double-click PeakDetect_system -> PeakDetect_system.sprj, and set the sysroot,Root FS, and Kernel Image. Also set the package options.

   

   As the AI Engine graph is being loaded by the host PS application, we can defer the running of the graph after the graph has been loaded using the xrt::graph API. By default, the AMD platform management controller (PMC) loads and runs the graph. However the v++ --package.defer_aie_run option will let you defer the graph run until after the graph has been loaded using the xrt::graph API.

   NOTE: When you change the build target from Emulation-SW(default) to Emulation-HW or Hardware, make sure to repeat the step-35.

   This creates the system project which can be used to build and run for all the targets.

Support

GitHub issues will be used for tracking requests and bugs. For questions, go to support.xilinx.com.

_{Copyright © 2020–2023 Advanced Micro Devices, Inc}

^{Terms and Conditions}