Cadence Tensilica IP Powers ADI’s Next-Generation Automotive DSP

Cadence and Analog Devices (ADI) have collaborated on the next generation of ADI’s SHARC automotive audio processors, integrating Cadence’s Tensilica DSP technology to deliver higher performance for advanced in-vehicle audio, voice and AI applications.

The new architecture is designed to support increasingly demanding automotive workloads, including road and engine noise cancellation, personal sound zones, immersive 3D audio, voice processing, AI assistants and hands-free communication.

A key part of the collaboration is the new SHARC-FX processor core, developed using Cadence Tensilica DSP technology. According to the companies, the new core delivers up to 5X the performance of the previous-generation SHARC+ processor used in the ADSP-SC59x family.

Designed for AI and Next-Generation Vehicles

ADI’s new ADSP-SC84x/2184x platform moves from a 28nm to a 16nm process, bringing improved energy efficiency, higher operating frequencies and greater integration for software-defined vehicles.

The platform also includes LPDDR4 memory support, increased on-chip L2 memory, a dedicated hardware security module and optional connectivity cores supporting enhanced communications and mainline Linux.

The SHARC-FX architecture adds enhanced VLIW SIMD capabilities, broader data-type support, improved cache handling and an expanded instruction set designed for modern signal processing and AI workloads.

These capabilities can support applications such as AI-enhanced noise cancellation, intelligent voice assistants, music source separation, vocal removal and personalized in-cabin audio.

The architecture is also designed to handle neural-network workloads directly on the DSP. Support for optimized AI/ML kernels, including Google’s LiteRT, allows compatible AI models to be offloaded from the host CPU for real-time processing with improved efficiency.

Expanded Development Ecosystem

ADI’s development ecosystem combines Cadence compilers and tools with its EZ-KIT evaluation systems, CrossCore Embedded Studio and SigmaStudio+, alongside DSP Concepts’ Audio Weaver framework.

The collaboration is aimed at giving automotive developers the processing performance and software tools needed to build increasingly intelligent, personalized and responsive audio systems for next-generation vehicles.