GreenWaves GAP9 Processor Designed for Ultra Low Power AI Devices

It is designed for release on 22-nanometer FD-SOI technology.
At a recent RISC-V Summit, Greenwaves demonstrated an application processor for IoT devices. It is called GAP9 and has five times less power consumption than the previous GAP8 model.

GAP9 combines architectural enhancements with the advanced 22nm Global Foundries FDX (FD-SOI) process technology. The processor is characterized by a bandwidth of the memory subsystem of 41.6 GB / s and a performance of 50 GOPS with a total power consumption of 50 mW.

Compared to GAP8, GAP9 adds support for floating point calculations by all cores with 8-, 16- and 32-bit precision and vectorization support. Integer calculations are also complemented by support for vectorized 4-bit and 2-bit operations. GAP9 can run complex neural networks, such as MobileNet V1, processing images with minimal energy. The advantages of GAP9 also include the presence of bidirectional multi-channel synchronized digital audio interfaces, making the processor ideal for complex wearable audio devices. There are also interfaces for connecting cameras.

GAP9 includes additional security features that protect the firmware and devices from unauthorized access, including hardware support for AES128 / 256 encryption and the Physical Unclonable Function (PUF) module, which allows device identification.

Allegedly, GAP9 allows you to embed machine learning and signal processing capabilities into battery-powered devices such as consumer wearable devices and medical devices, and even devices powered by energy collected from the environment.

Leave a Reply

Your email address will not be published. Required fields are marked *

12 − seven =