Once again Socionext is showcasing Akida at CES 2024.
Come see Socionext’s latest Custom SoC offerings and solutions at CES 2024! Socionext will showcase our leading-edge technologies, custom and standard products and solutions in the automotive, data center, smart devices, sensing, and IoT to help our customers develop feature-rich custom SoCs...
socionextus.com
View attachment 51266
View attachment 51267
Learning
Hi Learning,
I'm a little confused by the
"Socionext;s partner - Brainchip's Videos" followed by 2 purely Socionext 60GHz radar videos.
On the Socionext automotive page:
https://socionextus.com/products/automotive-custom-soc/ ,that heading is followed by links to Brainchip's MetaTF, Youtube and Enablement information.
The Socionext radar videos are further up the page and the block diagram is at
https://www.socionext.com/en/products/assp/radar-sensor/SC1260/
This does not include an NNA.
Advanced AI Solutions for Automotive
S
ocionext has partnered with artificial intelligence provider BrainChip to develop optimized, intelligent sensor data solutions based on Brainchip’s Akida® processor IP.
BrainChip’s flexible AI processing fabric IP delivers neuromorphic, event-based computation, enabling ultimate performance while minimizing silicon footprint and power consumption. Sensor data can be analyzed in real-time with distributed, high-performance and low-power edge inferencing, resulting in improved response time and reduced energy consumption.
However, the following does appear nearer the top of the page in relation to the Socionext multi-processor SoC:
Automotive Custom SoC Technologies and Solutions (socionextus.com)
Socionext’s advanced automotive custom SoC solutions are designed to help OEMs and tier-one automakers achieve differentiating technologies
socionextus.com
A
utomotive Custom SoC Technologies and Solutions
Featuring advanced technologies in automotive display, AI, and smart sensing
Socionext’s advanced custom solutions are designed to help OEMs and tier-one automakers achieve complete ownership of key differentiating technologies with an added competitive edge.
These custom SoCs enable a wide range of applications, including ADAS sensors, central computing, networking, in-cabin monitoring, satellite connectivity, and infotainment.
D
esign and Development:
High core count CPU subsystem development
Safety Island, System Management, and Secure Boot
High Performance interconnect (NIC/NOC)
ISP, GPU & NNA acceleration for power/performance optimization
The NNA accelerator is shown top right of the diagram, but this is not the Radar block diagram which does not include the multi-processor SoC.
Given the references to Brainchip on the page, it is possible that the NNA is or will be Akida in the next generation.
EDIT: No - that's wrong:
"Sensor data can be analyzed in real-time with
distributed, high-performance and low-power edge inferencing, resulting in improved response time and reduced energy consumption."
If they are using Akida in a distributed way (at the sensor), then they will not be using the full scale processor (CPU/GPU/NNS/ISP) SoC at the sensor.
But that does not stop them from using Akida in the full processor at some future date.
This is the Socionext gesture recognition patent
US2021026458A1 GESTURE RECOGNITION METHOD AND GESTURE RECOGNITION DEVICE
The recognition engine 28 is trained by "machine learning" by training it on various signal patterns generated by different gestures, but it is not described as a neural network. So the recognition engine is performing a task that Akida could do, but we don't know how efficiently it performs.
Given that the patent was filed before they knew about Akida, and they learned about Akida before the ink was dry on their patent application, perhaps they have an updated design in the pipeline. Possibly even one where Akida is incorporated into the radar receiver?