The microcontrollers bring generative AI to devices, enabling real-time, on-device processing without cloud dependence. Read more!
Alif Semiconductor, a supplier of secure, connected, power-efficient AI and ML microcontrollers (MCUs) and fusion processors, has announced the launch of the second generation of its Ensemble product family.
The E4, E6, and E8 Ensemble MCUs and fusion processors integrate Arm’s Ethos-U85 neural processing unit (NPU) with Alif’s memory and power management. This architecture supports the NPU operators required by transformer networks to handle generative AI workloads. Transformer networks improve endpoint use cases by focusing on key parts of older input sequences while processing new sequences in parallel for contextual understanding. This results in a faster, more accurate, and more satisfying experience for multimodal AI in endpoint products without cloud support, such as those used for vision, voice, text, and sensor fusion.
The Ensemble devices use the system architecture from the first-generation E1, E3, E5, and E7 products. These included significant on-chip memories to support pairs of Arm Cortex-M55 CPUs and Ethos-U55 NPUs, Alif’s aiPM power management, a secure enclave, and multiple ways to stream image, sound, and sensor data. Ensemble devices can perform neural network inferencing using only internal or external memory.
The second generation of Ensemble adds features for improved AI/ML performance at lower power consumption and support for embedded vision and audio applications. New features include a wider internal memory bus for increased data bandwidth for the NPU, dual MIPI-CSI camera interfaces, a hardware image signal processor running up to 200 frames per second, and faster interfaces to external memory with 800MB/s each. The updated SRAM data retention options provide more flexibility for faster wake-up from sleep.
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 chose Arm’s standardized 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.”
