
Accelerating FPGA & DSP Design with AMD Vitis™ Model Composer
Free
- Length
- 1 hr
- Platform
- Zoom
- Topic
- More Webinars
About this webinar
Join me for a practical walkthrough of AMD Vitis™ Model Composer, the model-based design environment that sits inside MATLAB® and Simulink® and turns one executable model into verified FPGA and adaptive SoC implementations. Starting from why the model should be the source of truth, the session introduces gateway blocks, fixed-point data types, and standard versus AXI4-Stream interfaces, then works through three flows in turn: classical DSP with an AM modulation and demodulation example in the HDL path (including hardware-in-the-loop on a Kintex™ 7 KC705), algorithm-centric C/C++ kernels and AMD Vitis libraries in the HLS path, and high-throughput FFT acceleration on AI Engines in AMD Versal™ adaptive SoCs, configured graphically without hand-written AIE C++. You will see how simulation, auto testbenches, match reports, co-simulation, and IP packaging keep the same model coherent from algorithm to silicon, and leave with a clear path to try the flows yourself using AMD reference tutorials and Adiuvo’s related white paper.
By the end of this webinar, participants will be able to:
• Explain why a single Simulink model can serve as specification, simulation, verification, and implementation artifact for FPGA and adaptive SoC designs.
• Apply Vitis Model Composer concepts (gateway blocks, fixed-point types, and standard versus AXI4-Stream interfaces) when defining the hardware boundary.
• Compare the HDL, HLS, and AI Engine flows and choose an appropriate abstraction for control logic, algorithm kernels, and high-throughput DSP.
• Describe how hardware-in-the-loop co-simulation and packaged Vivado-compatible IP cores move a verified model into real fabric and production block designs.
• Identify practical next steps using AMD reference tutorials (including the AI Engine FFT example) and Adiuvo’s white paper on accelerating designs with Vitis Model Composer.