South Korea, Samsung, Hyundai + Neuromorphic Chips, page-277

Yes totally, utterly and completely wrong.

The best way I can as a lay person explain the difference is this:

1. Presently you have a video camera at the entry to a secure location. It is just a video camera so to be useful you have a CPU installed somewhere in the facility to receive the video stream and identify intruders. There is a delay or latency in moving the video from the camera to the CPU processing it and identifying the intruders.

2. Enter the accelerator it speeds up the process by compressing the video and transmitting it from the video camera to the CPU where it is processed . Latency is reduced but real time processing is still not achieved. The reduction in latency is achieved by adding the power consumption of the accelerator to the existing power consumption of the CPU.

3. AKIDA is however the processor and is located at the video camera receiving the video feed in real time and processing it in real-time at source and the only information it transmits is intruder at the secure entry. It achieves power savings by eliminating the need for the external CPU and by its 1,2 or 4 bit activation and the SNN’s capacity to only process relevant events.

If you want a much more intelligent explanation then petition Barrel Sitter for an audience as everything I do not know he does.

My opinion only DYOR.