Computing Fundamentals
A track of P3 · Computer Science & Computer Architecture.
What a computer is, why GPUs matter for AI, and what happens in a data centre — three plain-language walks through the machinery under every model.
The laptop you are reading this on has perhaps eight processor cores. A graphics card built for AI has well over ten thousand. Eight brilliant workers against ten thousand simple ones — and for AI, the ten thousand win every time. Why? And why, even with those ten thousand, does a large language model still refuse to run on anything you could plug in at home? Those two questions are the whole of this track.
Three sessions, one story: how the box on your desk becomes the building full of servers behind every AI assistant. The first session introduces a computer's three working parts — the CPU as the chef, RAM as the countertop, storage as the pantry — and shows that everything a computer does is moving information between them. The second returns to the eight-versus-ten-thousand puzzle and resolves it: AI is millions of tiny, identical multiplications, a job that wants many hands rather than one genius, and the GPU is exactly that — thousands of simple hands doing the same small step at once.
The third session follows a large language model into the data centre and lets you do the arithmetic yourself. A model like GPT-3 stores 175 billion numbers — picture 8.75 million printed dictionaries — and every single one is consulted to produce each word of an answer. Stack that against what a home machine can hold and move per second, and "the cloud" stops being a vague place and becomes an obvious engineering necessity.
Written for readers with no technical background: each session is a plain-language walk with quizzes and quick recaps, and the analogies carry all the weight. If you have ever nodded through the words "GPU", "VRAM" or "the cloud", this track replaces the nod with a picture — and with numbers small enough to check on the back of an envelope.