At the Consumer Electronics Show (CES) in Las Vegas on January 10, 2025, Ceva, Inc. made a significant announcement regarding its embedded AI solutions, particularly its Ceva-NeuPro-Nano Embedded AI NPUs (Neural Processing Units). The company highlighted the rapid adoption of these processors in the Artificial Intelligence of Things (AIoT) and microcontroller unit (MCU) markets, pointing to multiple customer contracts that underscore their growing popularity.

The Ceva-NeuPro-Nano family includes the NPN32 and NPN64 models, which feature 32 and 64 multiply-accumulate (MAC) units, respectively. These NPUs are designed to provide a blend of power, performance, and cost efficiency, crucial factors for semiconductor companies and original equipment manufacturers (OEMs) looking to integrate embedded AI models into their System-on-Chip (SoC) solutions. They aim to facilitate demanding tasks such as voice recognition, vision processes, and sensory input management.

Ceva has further strengthened its development ecosystem by releasing the Ceva-NeuPro Studio, an integrated development environment (IDE) that simplifies the entire software design cycle for AI applications. Features of this IDE include industry-standard Eclipse support, compatibility with various open-source AI frameworks, and tools for simulation, emulation, and debugging. Additionally, it provides access to Ceva’s Model Zoo, which aids rapid benchmarking for developers.

In collaboration with Edge Impulse Studio, Ceva-NeuPro Studio enables clients to evaluate AI models prior to silicon availability and facilitates the deployment and retraining of models using the NVIDIA TAO Toolkit. This integration improves workflow efficiency, potentially accelerating the time to market for new AI-driven products.

The product line’s impact was further acknowledged when it achieved two notable accolades: the Best IP/Processor of the Year 2024 at the EE Awards Asia event and the IoT Edge Computing Excellence Award. Chad Lucien, vice president and general manager of the sensor and audio business unit at Ceva, remarked on the significance of these advancements. He stated, “The rapid adoption of our Ceva-NeuPro-Nano Embedded AI NPUs is a testament to our team’s dedication to pushing the boundaries of embedded AI. MCU and AIoT semiconductor companies have praised the Ceva-NeuPro-Nano’s efficiency as an NPU and its ability to simultaneously handle the demands of neural network compute, feature extraction, and processing complex DSP workloads all in a self-contained architecture."

Ceva's continued emphasis on enhancing its offerings reflects its commitment to fostering innovation within the intelligent edge AI applications sector.

Source: Noah Wire Services