2021 BRN Discussion, page-13914

Hi Luke,
Synaptics Katana appears to be an accelerator. It doesn't use a spiking neural processor, they do MAC's so it's just a math chip. They can't do all the things Akida can such as chip learning. I had a quick look and no patents indicating any use of spiking NN's. They don't disclose much information about their product.

IMO they aren't a true competitor.

https://www.electronicproducts.com/synaptics-a-natural-progression-to-edge-ai-processors/

Last year, Synaptics introduced the Smart Edge AI platform, consisting of the VideoSmart VS600 family of edge-computing video SoCs with a secure AI framework. The SoCs combine aCPU, NPU, and GPU, designed specifically for smart displays, smart cameras, video soundcards, set-top boxes, voice-enabled devices, computer-vision IoT products.

The platform uses the company’s Synaptics Neural Network Acceleration and Processing (SyNAP) technology, a full-stack solution for on-device deep-learning models for advanced features. With the inferencing done on the device, it targets privacy, security, and latency issues.

https://www.zdnet.com/article/get-ready-for-really-low-power-ai-synaptics-and-eta-compute-envision-neural-nets-that-will-observe-every-sound-every-motion/

A neural net is composed of predominantly linear algebra operations, multiplications of matrices. But the total problem to be solved in embedded devices is not just mathematical operations. It's also things such as power management, signal processing issues when interfacing with a sensor, and a host of other systems engineering issues.

"It's not just the MAC [multiply-accumulate operation], it's memory management and the compiler that maps TensorFlow efficiently onto the processor," Haddad explained.

"You need to efficiently map your algorithms to these various processors," including the neural net functional block but also the embedded digital signal processor and the host CPU that runs system software routines. "That is a very complex problem, and we solve that."