Ceva, a global leader in AI and semiconductor technology, has unveiled a significant advancement in its NeuPro-M NPU family of IP cores. These enhanced cores now seamlessly support sparse transformer frameworks for generative AI applications, a critical milestone in AI development. The NeuPro-M NPU architecture has undergone extensive redesign to accommodate transformer networks alongside CNNs and other neural networks, ensuring compatibility with future machine learning inferencing models as well.

This breakthrough enables highly-optimized generative AI applications to run on NeuPro-M NPU cores embedded within diverse devices including communication gateways, connected networks, vehicles, notebooks, tablets, AR/VR headsets, smartphones, and more, both at the edge and in the cloud. The improved NeuPro-M architecture leverages integrated Vector Processing Units (VPUs) to facilitate the processing of future network layers with true sparsity for data and weights. This innovation results in up to a remarkable 4X acceleration in performance, a game-changer in transforming AI capabilities across various sectors.

To address scalability demands across diverse AI markets, Ceva has introduced new NPM12 and NPM14 NPU cores, each housing two and four NeuPro-M engines, respectively. These enhancements, coupled with a revamped comprehensive development toolchain based on the acclaimed CDNN neural network AI compiler, position Ceva's NeuPro-M NPU IP family as a versatile and powerful solution. Ran Snir, Vice President and General Manager of the Vision Business Unit at Ceva, highlighted, "The leap in performance we have achieved with this architecture brings the incredible promise of Generative AI to any use case, from cost-sensitive edge devices all the way up to highly-efficient cloud computing and everything in between."

As the AI landscape continues to evolve, Ceva's advancement in NeuPro-M NPU cores offers a pivotal step towards efficient and powerful AI inferencing, particularly benefiting edge devices where performance, power, cost, latency, and memory considerations are paramount. ABI Research predicts strong growth in Edge AI shipments, reinforcing the role of Generative AI in driving innovation and transformation across industries.

Source: eenewseurope.com