Arm recently announced the launch of the Ethos-U85 NPU. As the third-generation NPU product aimed at edge AI from Arm, the Ethos-U85 is suitable for scenarios such as industrial automation and video surveillance, boasting a fourfold improvement in performance.
The Ethos-U85 achieves a 20% improvement in energy efficiency over its predecessor and can achieve an 85% utilization rate on commonly used neural networks. It is designed to be compatible with systems based on Arm Cortex-M/A processor cores, accommodating higher memory latency.
The Ethos-U85 NPU supports configurations ranging from 128 to 2048 MAC units, achieving 4TOPs of computing power at a frequency of 1GHz at its maximum scale.
This NPU product supports the same toolchain as the previous Ethos-U55/65, facilitating ease of use for developers. Additionally, it supports AI frameworks such as TensorFlow Lite and PyTorch.
The first companies to adopt the Ethos-U85 NPU include Infineon and Alif Semiconductor (Note: The latter was established in 2019 and was also one of the first companies to launch with the Ethos-U55).
In addition, Arm also launched the Corstone-320 IoT reference design platform, which integrates the Cortex-M85 CPU, Mali-C55 IPS, and Ethos-U85 NPU, providing sufficient performance for edge AI applications in audio and vision.
The Corstone-320 reference design platform includes Arm virtual hardware, allowing developers to commence software development before the physical chip is ready, thereby shortening the edge AI device launch process.
Paul Williamson, Senior Vice President and General Manager of IoT at Arm stated, “As the deployment scale of edge AI continues to expand, IC designers must address increasingly complex systems and software, the rapidly growing demand for AI performance, and the pressure to accelerate the product-to-market process. At the same time, software developers need a more consistent, simplified development experience and the ability to more easily integrate with new AI frameworks and libraries. The new technology introduced by Arm meets the growing demands for accelerating edge AI deployment.”
Disclaimer: This article is created by the original author. The content of the article represents their personal opinions. Our reposting is for sharing and discussion purposes only and does not imply our endorsement or agreement. If you have any objections, please contact us through the provided channels.
