will appear in stm32 Neural Processing Unit(NPU)? maybe it's already in stm32? The Internet of Things world cannot do without NPU. It is better to start studying it now. can you suggest something?

[Tesla DeLorean]

Is that for "programmers" ​who can't come up with logical reasoning without machine assistance?

[MTS]

on the growth of the number of timers alone, you will not go into the future. It will require an NPU. it's better to start preparing now. IoT cannot be without NPU. If you don't understand that, then I'm sorry. I'm not a programmer, I'm just a consumer. but perhaps this is what helps me to see what is needed right now.

[MTS]

for example, a robot cat. how can you teach him to recognize a face without a neural network ? robotics should be based on stm microcontrollers. I would be pleased if this happened. If stm gives microcontrollers with NPU to the market now, then programmers will already have experience with this and in the future they will give preference to what is already familiar. concunence can be avoided there.

[MTS]

let the NPU be in stm32 right now. does this interfere? let's say today someone wants to use it. and stm is the first to offer it. so they guarantee themselves a place in the future. it is always necessary to give more than is necessary for now. moreover, 100% of it will be needed in a few years.

[Piranha]

> IoT cannot be without NPU. If you don't understand that, then I'm sorry.

A good example of how marketing brainwashes the simple people. 99% of IoT has nothing to do with ML (machine learning). First, read what an IoT is:

https://en.wikipedia.org/wiki/Internet_of_things

> for example, a robot cat.

In real life (almost) noone needs your robot cat.

> let the NPU be in stm32 right now. does this interfere?

The industry has a semiconductor deficit, but you are suggesting wasting a significant amount of silicon space for some rarely necessary feature with a minor impact on the whole industry. Anyway, there are coming some ML related features also for MCUs with Cortex-M55, Ethos-U55 and other cores:

https://images.anandtech.com/doci/15494/Arm-N55-10.png

> I'm not a programmer, I'm just a consumer. but perhaps this is what helps me to see what is needed right now.

You cannot see even what is necessary now, not even talking about future. For example, for the last decades the software has become the greatest problem. Hardware is pretty nice already and is still evolving, but software is so terribly lagging behind, that most of the hardware capabilities are just unused or wasted because of bloatware.

[JPeac.1]

While an NPU would open up a few applications I believe ST engineering time would be better spent on integrating a quantum entanglement MAC and PHY, creating a wireless communications peripheral that eliminates all external components and not only offers zero latency in transmission time to anywhere on the planet or orbital space, but even does away with all those nasty RF design challenges....

IOT cannot move into the future without quantum level communications. If you don't understand why it's more important than an NPU then I'm sorry.

Jack Peacock.

[MTS]

Piranha, with such documentation as you have and with such an approach to everything and not the ability to listen, you are doomed. please remember this message in the future.