VeriSilicon Unveils Ultra-Low Energy NPU for Mobile AI Applications

VeriSilicon has announced its new ultra-low energy Neural Network Processing Unit (NPU) IP, which supports on-device inference of large language models (LLMs) with AI computing performance exceeding 40 TOPS, announced in a press release. This NPU is designed to meet the growing demand for generative AI capabilities on mobile platforms, offering powerful computing performance for AI PCs and other end devices while optimizing energy efficiency for AI phones and mobile platforms.

The NPU is built on a configurable and scalable architecture, supporting mixed-precision computation, advanced sparsity optimization, and parallel processing. It incorporates efficient memory management and sparsity-aware acceleration to reduce computational overhead and latency, ensuring smooth AI processing. The NPU supports hundreds of AI algorithms and leading AI models, such as Stable Diffusion and LLaMA-7B, and can be integrated with VeriSilicon's other processing IPs for heterogeneous computing.

Additionally, the NPU IP supports popular AI frameworks like TensorFlow Lite, ONNX, and PyTorch, facilitating easier deployment and integration across various AI use cases. VeriSilicon continues to invest in ultra-low energy NPU development for AI phones and PCs, collaborating with leading SoC partners to realize this technology in silicon for next-generation devices.