» Current | 2021 | 2020 | 2019 | Subscribe

Linley Newsletter

Ambient’s Analog Cuts AI Power

November 3, 2020

Author: Aakash Jani

Startup Ambient Scientific uses analog technology to reduce power for AI inference and training in IoT devices. Its GPX-10 SoC delivers 510 billion operations per second (GOPS) while consuming only 120mW for active inferencing. In standby or always-on mode, the chip requires as little as 10 microwatts. It includes a wide array of sensor interfaces that feed directly to its proprietary sensor-fusion hub.

Ambient was founded in 2017. The startup is led by GP Singh, who helped design the UltraSparc processor at Sun Microsystems and led the engineering team at Wave Computing. It has withheld its funding amount and the identity of investors, but we estimate it has raised enough to cover the design and production cost of its first device. The GPX-10 is currently sampling and is expected to enter production in 2Q21.

The GPX-10 features a single Cortex-M4 CPU with 10 AI cores. It’s manufactured in TSMC’s 40nm process and has 512KB of on-die flash memory, in addition to support for external flash through QSPI and SPI protocols. The chip handles both analog and digital sensors through its wide range of interfaces.

The GPX-10 is the first chip to bring retraining to IoT devices. Today’s neural networks are trained in the cloud using generic data from many users. This approach takes advan­tage of the massive compute power in cloud data centers, but it creates a one-size-fits-all AI model. To provide truly personal services, an IoT device must adapt the neural network to the user. For example, a voice-recognition model could be trained to recognize a specific voice. This approach doesn’t require training the model from scratch, a computationally intensive task; instead, an existing model can be retrained on the device to more accu­rately respond to a particular user or use case.

Subscribers can view the full article in the Microprocessor Report.

Subscribe to the Microprocessor Report and always get the full story!

Purchase the full article

Free Newsletter

Linley Newsletter
Analysis of new developments in microprocessors and other semiconductor products
Subscribe to our Newsletter »


Linley Fall Processor Conference 2021
Held October 20-21, 2021
Proceedings available
Linley Spring Processor Conference 2021
April 19 - 23, 2021
Proceedings Available
More Events »