1. Home
  2. Blogs & News
  3. Latest Blogs
  4. AnalogAI has chosen Microchip’s SST memBrain SAGE intellectual property to develop its first real-world edge AI processor

AnalogAI has chosen Microchip’s SST memBrain SAGE intellectual property to develop its first real-world edge AI processor

0

AnalogAI has selected the memBrain Synaptic Analog Generation Engine (SAGE) neuromorphic hardware intellectual property (IP) from Silicon Storage Technology (SST), a subsidiary of Microchip Technology Inc., for the development of real-world edge AI processors. AnalogAI’s edge AI processors are equipped with proprietary algorithms optimised for hardware, enabling them to perform both AI model training and inference tasks simultaneously in real-world environments.

GR-26-115972-260818-SST-PR

AnalogAI utilises SST’s memBrain SAGE IP to deliver high-performance analogue in-memory computing (aCIM) at power consumption levels of 1 watt or less, making it suitable for ultra-low-power edge applications. Building on the silicon-proven memBrain SAGE IP, which is based on SST’s SuperFlash technology, AnalogAI is committed to developing inference applications for real-world scenarios such as humanoid robots, drones and in-vehicle devices.

Mark Reiten, Senior Vice President of Microchip’s Smart Computing Business Unit, said: “AnalogAI is dedicated to developing innovative, co-optimised edge AI inference solutions that enable edge devices to adapt in real time to sudden changes in their environment. As the core inference engine for AnalogAI’s first products, SST’s memBrain SAGE IP delivers the computational performance and energy efficiency required to meet their application needs. We warmly welcome AnalogAI as the latest licensed partner for SST memBrain IP, joining our ever-expanding ecosystem.”

Key features of the SST memBrain SAGE IP:

  • memBrain Tensor-in-Memory Logic Unit (TILE)
  • The memBrain solution utilises optimised ESF3 memory cells to achieve a storage capacity of up to 8 bits per cell whilst consuming only nanoampere-level power.
  • memBrain customised arrays, decoders and driver circuits
  • Optimised Digital-to-Analogue Converters (DACs) and Analogue-to-Digital Converters (ADCs)
  • Summers and high-voltage bias circuits
  • Na'an-class storage unit control logic
  • memBrain proprietary test circuit
  • SST memBrain Technical Documentation and Simulation Models
  • A comprehensive range of IP integration services and technical support
Jaejun Lee, CEO of AnalogAI, stated: “After screening all available solutions across the industry, we ultimately selected SST’s memBrain SAGE IP. This silicon-proven IP enables us to shorten our development cycle whilst meeting the dual requirements of ultra-low power consumption and high performance in our market segment. Analog-in-memory computing is a rapidly evolving field, and AnalogAI is developing new capabilities for edge AI devices that require real-world adaptability.”

SST’s memBrain SAGE IP has been developed and deployed using production-ready SuperFlash memory on 40 nm and 28 nm foundry processes; the current technology roadmap also includes the development of 22 nm memBrain SAGE IP. SuperFlash memory enables highly reliable, high-performance and low-power non-volatile storage on-chip, making it widely applicable to various applications that require fast access, high endurance and long-term data retention without the need for external memory devices.

Related Article

lazy img
Obsolete & EOL Parts Sourcing
Save Your Cost And Time
lazy img
Strict Quality Assurance
Trusted Global Supply Network
lazy img
Large Inventory Supply
Support Same Day Shipping
lazy img
Customer Support
Quick Quote & 365 Days Warranty
TOP
RFQ List ( 0 items)