Tech glossary
Plain-English definitions of the terms we use. In any article, tap an underlined term to see its definition in place.
- burn-in
- Permanent ghosting on OLED panels caused by static content displayed for very long periods. Modern panels mitigate it; heavy static UI still accelerates it.
- inference
- Running a trained AI model to get answers (as opposed to training it). "Local inference" means the model runs on your own device.
- NPU
- Neural Processing Unit — a chip block built to run AI models efficiently, using far less power than a CPU or GPU for the same inference work.
- OLED
- A display technology where each pixel emits its own light — perfect blacks, high contrast, and better battery at dark UI, at some risk of burn-in.
- thermal throttling
- When a chip slows itself down because it is running too hot — the reason sustained performance can be far below peak benchmark numbers.
- TOPS
- Tera-Operations Per Second — the headline speed figure for NPUs. Comparable only within the same precision (e.g. INT8).