Pioneering generative AI in edge and endpoint devices

Alif Semiconductor, a leading global supplier of secure, connected, power-efficient AI/ML MCUs and fusion processors, has strengthened its position as the pioneer of endpoint AI MCUs with the announcement of the second generation of its Ensemble family of products.

The new E4, E6 and E8 Ensemble MCUs and fusion processors will integrate the latest NPU from Arm, the Ethos-U85, with the company's unique high bandwidth memory and intelligent power management architecture to support NPU operators needed by transformer networks to efficiently execute generative AI workloads. Transformer networks enrich endpoint use cases by concentrating on the most relevant parts of older portions of an acquired input sequence while rapidly processing in parallel the current acquired sequence to enable deeper contextualisation. For localised use, this results in an instant, accurate, and more satisfying experience when using multimodal AI in endpoint products focused on vision, voice, text, and sensor fusion without relying on the cloud.

Reza Kazerounian, president of Alif Semiconductor, said: "Alif's original Ensemble controllers in full production today remain at the forefront of the market for AI/ML performance, on-chip resources, energy efficiency, and small footprint. We specifically chose Arm's standardised Ethos NPU IP because of its widely supported broad ecosystem and superior performance and efficiency. Alif's second-generation controllers will dial up benefits again, enabling developers to create products that they previously could only wish for, unleashing endless new use cases that will enter our everyday lives."

The new devices are based on the proven system architecture, which made the first-generation E1, E3, E5, and E7 products so popular. These included large, fast on-chip tightly coupled memories to feed pairs of Arm Cortex-M55 CPUs coupled to Ethos-U55 NPUs, the company's proprietary aiPM granular power management, a hardware-isolated secure enclave, and multiple ways to stream image, sound, and sensor data into the device. These devices can readily perform inferencing for neural networks employing only their internal memory and external memory if required.

The second generation of Ensemble introduces a set of improvements to give even higher AI/ML performance at lower power consumption, and to supply more integrated support for embedded vision and audio applications. New features incorporate a wider internal memory bus topology, bringing abundant localised data bandwidth for NPU consumption, dual MIPI-CSI camera interfaces, a fast hardware image signal processor operating up to 200fps, and a pair of faster interfaces to external memory, each capable of 800MB/s. Finally, new, broader options for SRAM data retention give designers more flexibility in waking from sleep faster.

Paul Williamson, senior vice president and general manager, IoT Line of Business at ARM, said: "As edge AI deployments continue to grow at pace, we need to deliver smarter, faster, and more secure processing on the devices closest to the data source. With the world's first implementation of the Ethos-U85 in an endpoint AI MCU, Alif is enabling on-device generative AI, which is particularly valuable in voice and vision applications. OEMs can now embed AI into battery-powered applications, illustrating why edge AI is being built on the energy-efficient Arm compute platform."

Pete Bernard, executive director of the EDGE AI FOUNDATION, added his appreciation of the role that Alif has played in enabling AI at the edge. He said: "The use of transformer networks in endpoint devices such as smart glasses or earbuds promises to make interactions with devices astonishingly intuitive and natural. Alif is opening a pathway to this new opportunity by implementing the Ethos-U85 NPU in the high-performance, low-power Ensemble architecture."