MathWorks, a leading developer of technical computing software, is supporting the Qualcomm Hexagon neural processing unit (NPU) in MATLAB and Simulink, and is improving workflow efficiency by automating code generation for architectures.

MathWorks announced today the release of a hardware support package for the Qualcomm Hexagon NPU (neural processing unit), a technology built into the Snapdragon processor family.

The MathWorks Hardware Support Package automates code generation for MATLAB and Simulink models optimized for Qualcomm Technologies’ Hexagon NPU architecture, improving data accuracy and standards compliance while improving workflow productivity.

This enables engineers to deploy production-quality C code for embedded applications without in-depth knowledge of the NPU using MathWorks software and Model-Based Design.

“Qualcomm’s mission is to enable engineers to create innovative products and applications that take full advantage of the potential of Snapdragon processors,” said Vinesh Sukumar, senior director of product management at Qualcomm Technologies.“With the demand for high-performance NPU applications continuously increasing, MathWorks and Qualcomm are collaborating to help developers efficiently develop embedded applications optimized for Hexagon NPUs,” he said.

The MathWorks Hardware Support Package for Hexagon NPUs addresses key challenges in the development process, including compilation errors, reliance on hand-written frameworks, manual numerical verification, and tracing code back into models.

Engineers without in-depth knowledge of Hexagon NPUs and their tool chains can use the package to run Processor-in-the-Loop (PIL) simulations directly from Simulink.

This allows engineers to verify that the optimized target code operates efficiently, reliably, and accurately before actual hardware deployment.

“Combining the power of Model-Based Design with the advanced capabilities of Hexagon NPUs enables engineers to more easily build high-quality NPU applications,” said Jim Tung, MathWorks Fellow. “Our collaboration with Qualcomm Technologies focuses on eliminating manual processes and introducing highly automated workflows, helping engineering organizations accelerate time to market and ensure optimal performance.”

The automatic code generation capabilities of MATLAB and Simulink are trusted by companies across a wide range of industries, including automotive, aerospace, and medical devices.

MathWorks software supports a wide range of processors, including the ARM Cortex-M, ARM Cortex-A, ARM Cortex-R, AURIX TC3x, AURIX TC4x, and GPUs from Arm and NVIDIA. We have already optimized code generation.