SamurAI: A 1.7MOPS-36GOPS Adaptive Versatile IoT Node with 15,000× Peak-to-Idle Power Reduction, 207ns Wake-Up Time and 1.3TOPS/W ML Efficiency

• Main Authors: Miro-Panades, Ivan, Tain, Benoit, Christmann, Jean-Frederic, Coriat, David, Lemaire, Romain, Jany, Clement, Martineau, Baudouin, Chaix, Fabrice, Quelen, Anthony, Pluchart, Emmanuel, Noel, Jean-Philippe, Boumchedda, Reda, Makosiej, Adam, Montoya, Maxime, Bacles-Min, Simone, Briand, David, Philippe, Jean-Marc, Valentian, Alexandre, Heitzmann, Frederic, Beigne, Edith, Clermidy, Fabien
• Format: Conference Proceeding
• Language: English
• Subjects: Cryptography; Energy efficiency; Nonvolatile memory; Power demand; Power measurement; Random access memory; Voltage measurement
• ISSN: 2158-5636
• DOI: 10.1109/VLSICircuits18222.2020.9163000

IoT node application requirements are torn between sporadic data-logging and energy-hungry data processing (e.g. image classification). This paper presents a versatile IoT node covering this gap in processing and energy by leveraging two on-chip sub-systems: a low power, clock-less, event-driven Always-Responsive (AR) part and an energy-efficient On-Demand (OD) part. The AR contains a 1.7MOPS event-driven, asynchronous Wake-up Controller (WuC) with 207ns wake-up time optimized for short sporadic computing. OD combines a deep-sleep RISC-V CPU and 1.3TOPS/W Machine Learning (ML) and crypto accelerators for more complex tasks. The node can perform up to 36GOPS while achieving 15,000× reduction from peak-to-idle power consumption. The interest of this versatile architecture is demonstrated with 105μW daily average power on an applicative classification scenario.