The tiny silicon revolution
Onboard AI in the wild
Privacy, security, and the future
007skaty Neural processing units run artificial intelligence tasks directly on modern smartphones
007skaty Since Apple's 2017 A11 Bionic chip, specialized NPUs have used massive matrix multiplication to run AI faster than the cloud.
In what year did Apple first introduce its dedicated 'Neural Engine' inside an iPhone?
007skaty Onboard AI requires shrinking large language models to fit strict mobile memory limits.
What is the primary purpose of the AI optimization technique known as 'quantization'?
Power of TOPS: A modern smartphone NPU can perform 45 trillion operations per second while drawing less power than a single household LED lightbulb.
007skaty Local hardware processes artificial intelligence directly to bypass slow cloud networks.
007skaty Wire the mini processor's connections to discover how onboard AI makes instant decisions offline.
What term describes the delay or lag caused by sending data back and forth to a cloud server?
007skaty TinyML Microcontrollers
On a tiny chip, on a tiny chip, Neural networks running offline.
What type of extremely low-power chip allows simple appliances to run basic AI models?