But BRN doesn't work on the same functionality as a conventional...

But BRN doesn't work on the same functionality as a conventional CPU/GPU - it doesn't perform the same raw calculation function. It doesn't operate on GHz cycles or process things in framerates. You're comparing apples and oranges mate.

The important metrics are, how fast it can learn, how accurately it can identify a condition that requires a decision eg. object/person, smell, vibration; and how much power it uses.

It helps if you understand what the chip is for before asking for something it is not supposed to do.

The Akida chip is meant to identify conditions that require an action from analogue data and then push that info, when required, to a decision making chip - this could be another Akida chip that is pulling in data from other sensors, or a CPU/GPU that fires up and start crunching numbers.

Think of it a bit like this - when you first drive a car as a Learner at 60km/h - I remember the experience - your senses almost get overwhelmed because you have so much information coming at you (vibration from the car, the sound of the engine, wind noise over the windscreen, the way the road/objects comes up on you at that speed) - but as you practice, you brain learns to ignores the info that isn't relevant, until you only deal with data that requires you to make a decision and appropriate action. The decision you make also becomes learned behaviour so most of the time you don't even consciously think about what course of action you are going to take. That's how Akida works.