Uploaded November 2025 | Updated September 2026, 2 weeks ago
Learn how MATLAB® and Simulink® streamline the development and deployment of AI and signal processing applications on Qualcomm® Hexagon® DSPs and NPUs. Adam Cook, application engineering manager at MathWorks, and Talha Gorsi, AI product manager at Qualcomm, introduce the Hexagon Hardware Support Package—a new capability that connects the MATLAB and Simulink environment with Qualcomm’s Hexagon processor, part of the Snapdragon®SoC.
The Hexagon Support Package enables you to generate production-quality C code from MATLAB algorithms and Simulink models and deploy it directly to Hexagon hardware or simulators. The generated code is optimized for performance using scalar and vector optimizations and integrates AI inference engine APIs to access the Embedded Neural Processing Unit (eNPU). This workflow eliminates the need to learn Qualcomm-specific tools, allowing you to stay within the familiar MATLAB and Simulink environment.
Key topics include the value of the support package, steps to get started, and a practical example using a smart speaker model. You’ll see how to integrate signal processing, control logic, and AI components into a complete system and automatically deploy it to Hexagon targets. Features such as processor-in-the-loop (PIL) enable real-time communication between the simulator and hardware, while performance analysis tools help you evaluate metrics like CPU utilization.
A major advantage of this workflow is seamless AI integration. For example, in the smart speaker model, the voice command recognition engine—a trained convolutional neural network—runs on the eNPU, while other components execute on the Hexagon ADSP. This automated integration saves significant development time compared to manual approaches.
Talha Gorsi provides an overview of Qualcomm’s AI Engine architecture, which distributes AI workloads across CPU, GPU, Hexagon NPU, and Sensing Hub. Each compute block is optimized for specific tasks: CPUs for latency-sensitive applications, GPUs for graphics-heavy workloads, NPUs for high-performance AI acceleration, and the Sensing Hub for ultra-low-power, always-on use cases. This flexibility supports real-time applications such as computer vision and generative AI at the edge.
Learn more:
- Qualcomm Hexagon - Hardware Support - - Qualcomm Hexagon - Hardware Support - MATLAB & Simulink: bit.ly/47zAqtG
Chapters:
00:00 Introduction & Welcome
01:00 Hexagon Hardware Support Package Overview
02:30 Key Concepts & Workflow
04:00 Processor-in-the-Loop (PIL) Concept
05:00 AI Integration & System Design
06:30 Qualcomm AI Engine & Compute Architecture
08:00 Qualcomm AI Stack & Development Approach
09:30 Real-Time Applications & Edge AI
11:00 Demo: Installing & Setting Up Support Package
13:00 Demo: Smart Speaker Model Walkthrough
15:00 Demo: Hardware Target Configuration & Code Generation
17:00 Demo: Processor-in-the-Loop Simulation & Profiling
20:00 Performance Comparison & Optimization
22:00 Getting Started & Resources
23:00 Conclusion
--------------------------------------------------------------------------------------------------------
Get a free product trial: goo.gl/ZHFb5u
Learn more about MATLAB: goo.gl/8QV7ZZ
Learn more about Simulink: goo.gl/nqnbLe
See what's new in MATLAB and Simulink: goo.gl/pgGtod
© 2025 The MathWorks, Inc. MATLAB and Simulink are registered trademarks of The MathWorks, Inc.
See mathworks.com/trademarks for a list of additional trademarks. Other product or brand names may be trademarks or registered trademarks of their respective holders.
Learn how MATLAB® and Simulink® streamline the development and deployment of AI and signal processing applications on Qualcomm® Hexagon® DSPs and NPUs. Adam Cook, application engineering manager at MathWorks, and Talha Gorsi, AI product manager at Qualcomm, introduce the Hexagon Hardware Support Package—a new capability that connects the MATLAB and Simulink environment with Qualcomm’s Hexagon processor, part of the Snapdragon®SoC.
The Hexagon Support Package enables you to generate production-quality C code from MATLAB algorithms and Simulink models and deploy it directly to Hexagon hardware or simulators. The generated code is optimized for performance using scalar and vector optimizations and integrates AI inference engine APIs to access the Embedded Neural Processing Unit (eNPU). This workflow eliminates the need to learn Qualcomm-specific tools, allowing you to stay within the familiar MATLAB and Simulink environment.
Key topics include the value of the support package, steps to get started, and a practical example using a smart speaker model. You’ll see how to integrate signal processing, control logic, and AI components into a complete system and automatically deploy it to Hexagon targets. Features such as processor-in-the-loop (PIL) enable real-time communication between the simulator and hardware, while performance analysis tools help you evaluate metrics like CPU utilization.
A major advantage of this workflow is seamless AI integration. For example, in the smart speaker model, the voice command recognition engine—a trained convolutional neural network—runs on the eNPU, while other components execute on the Hexagon ADSP. This automated integration saves significant development time compared to manual approaches.
Talha Gorsi provides an overview of Qualcomm’s AI Engine architecture, which distributes AI workloads across CPU, GPU, Hexagon NPU, and Sensing Hub. Each compute block is optimized for specific tasks: CPUs for latency-sensitive applications, GPUs for graphics-heavy workloads, NPUs for high-performance AI acceleration, and the Sensing Hub for ultra-low-power, always-on use cases. This flexibility supports real-time applications such as computer vision and generative AI at the edge.
Learn more:
- Qualcomm Hexagon - Hardware Support - - Qualcomm Hexagon - Hardware Support - MATLAB & Simulink: bit.ly/47zAqtG
Chapters:
00:00 Introduction & Welcome
01:00 Hexagon Hardware Support Package Overview
02:30 Key Concepts & Workflow
04:00 Processor-in-the-Loop (PIL) Concept
05:00 AI Integration & System Design
06:30 Qualcomm AI Engine & Compute Architecture
08:00 Qualcomm AI Stack & Development Approach
09:30 Real-Time Applications & Edge AI
11:00 Demo: Installing & Setting Up Support Package
13:00 Demo: Smart Speaker Model Walkthrough
15:00 Demo: Hardware Target Configuration & Code Generation
17:00 Demo: Processor-in-the-Loop Simulation & Profiling
20:00 Performance Comparison & Optimization
22:00 Getting Started & Resources
23:00 Conclusion
--------------------------------------------------------------------------------------------------------
Get a free product trial: goo.gl/ZHFb5u
Learn more about MATLAB: goo.gl/8QV7ZZ
Learn more about Simulink: goo.gl/nqnbLe
See what's new in MATLAB and Simulink: goo.gl/pgGtod
© 2025 The MathWorks, Inc. MATLAB and Simulink are registered trademarks of The MathWorks, Inc.
See mathworks.com/trademarks for a list of additional trademarks. Other product or brand names may be trademarks or registered trademarks of their respective holders.
