Arm, a leading semiconductor architecture company, is making significant strides in the AI sector with its latest edge AI technology embedded in new chip designs.
Today, the Cambridge, England-based company introduced its Ethos-U85 Neural Processing Unit (NPU) and the Corstone-320 IoT Reference Design Platform, specifically tailored for edge AI applications. This initiative aims to enhance the intelligence of the Internet of Things (IoT).
Ethos-U85 NPU: Unleashing High-Performance Edge AI
The Ethos-U85 stands as Arm’s most powerful and efficient NPU to date, addressing the rising demands for edge AI applications. With a fourfold performance boost and 20% greater power efficiency compared to its predecessor, the Ethos-U85 excels in applications like factory automation and smart home cameras.
Featuring a scalable architecture with 128 to 2048 MAC units—equating to 4 TOPs at 1GHz—the Ethos-U85 achieves outstanding performance while conserving energy. It supports leading AI frameworks, including TensorFlow Lite and PyTorch, ensuring easy integration for developers. Its compatibility with Transformer Networks and Convolutional Neural Networks (CNNs) opens the door for a wide range of AI applications, from vision tasks to generative AI.
The Ethos family of NPUs has already attracted considerable industry attention, with over 20 partners licensing the technology. Early adopters like Alif and Infineon have recognized the Ethos-U85's potential to drive innovation in edge AI.
“Machine learning workloads for next-generation edge AI applications demand high performance in a power-efficient manner,” stated Reza Kazerounian, president of Alif Semiconductor. “Alif was first to market with a solution based on Arm Cortex-M55 and Ethos-U55, and we welcome the Ethos-U85, which will support the compute performance for our next-generation Ensemble family of microcontrollers and fusion processors.”
Corstone-320 IoT Reference Design Platform
Arm also launched the Corstone-320 IoT Reference Design Platform alongside the Ethos-U85. This platform aims to expedite the development of edge AI systems by integrating the Arm Cortex-M85 CPU, Mali-C55 Image Signal Processor, and Ethos-U85 NPU, enabling real-time processing of voice, audio, and vision data.
This comprehensive hardware and software platform supports rapid prototyping and deployment of edge AI applications. It provides essential tools, support, and Arm Virtual Hardware, allowing software development prior to silicon availability, thereby reducing time-to-market for complex edge AI devices.
“Edge AI use cases are becoming increasingly sophisticated and require secure, high-performance compute systems,” remarked Steve Tateosian, senior vice president at Infineon. “We are excited to build on our partnership with Arm and leverage the Ethos-U85 to enable these advanced systems for edge AI deployments.”
Arm's architecture serves as the foundation for chips designed by 15 million developers worldwide. By releasing innovative designs like the Ethos-U85 and Corstone-320, Arm enables its licensed partners to create and manufacture their own chips based on these cutting-edge technologies.