Model Composer enables the rapid simulation, exploration, and code generation of algorithms targeted for AI Engines from within the Simulink® environment. You can achieve this by importing AI Engines kernels and data-flow graphs into Model Composer as blocks and controlling the behavior of the kernels and graphs by configuring the block GUI parameters. Simulation results can be visualized by seamlessly connecting Simulink source and sink blocks with the Model Composer AI Engines blocks. Furthermore, the simulation results can be sent to the MATLAB® workspace for further analysis.
Refer to Creating an AI Engine Design using Model Composer for more information on importing AI Engine kernels and graphs as blocks.
- Blocks to import kernels and graphs which can be targeted to the AI Engine portion of Versal devices.
- Block to import HLS kernels which can be targeted to the PL portion of Versal devices.
- Blocks that support connection between the AI Engine and the Xilinx HDL blockset.
- Configurable AI Engine functions such as FIR filters.
Connecting HLS kernel blocks, HDL library blocks, and AI Engine blocks, allows modeling and simulation of a heterogeneous platform which can be targeted to both programmable logic and AI Engines in Versal™ ACAP devices.
In addition to simulation, you can also use the Model Composer Hub to generate dataflow graphs. For more details on the Model Composer Hub block, specific to AI Engine code generation, refer to Code Generation.
Model composer allows you to verify the generated dataflow graph code using the AI Engine simulator. Based on verification requirements, you can choose to verify your algorithm from the Model Composer Hub block.
A typical AI Engine design flow is shown in the following diagram.
To learn more about the AI Engine flow in Model Composer, refer to the Vitis Model Composer Tutorial (UG1498).