Cadence has announced the launch of the Tensilica NeuroEdge 130 AI Co-Processor (AICP), aimed at enhancing the performance and efficiency of AI tasks within advanced SoCs across automotive, consumer, industrial, and mobile markets. Built on the architecture of the Tensilica Vision DSPs, the new processor offers more than 30% area savings and over 20% in dynamic power and energy reduction, without affecting performance. Designed to complement any neural processing unit, the NeuroEdge 130 AICP handles pre- and post-processing tasks commonly unsuited for NPUs, improving execution of modern AI workloads. It also provides flexibility for integration with in-house and third-party NPUs and includes programmability features to support future AI needs.
The NeuroEdge 130 AICP is supported by Cadence’s NeuroWeave SDK, allowing streamlined AI model deployment through TVM-based tools. It includes a standalone AI library for direct programming, reducing dependency on heavier frameworks. Industry support has been strong, with companies like Indie, MulticoreWare, and Neuchips noting the processor’s value in boosting AI performance while meeting design constraints in power and area. Cadence continues to build on its expertise in AI-enabled IP solutions, addressing growing demands for efficient co-processing in AI-focused SoCs. Multiple customers have already engaged with the new processor, reflecting strong market interest.
