The Pixel Pusher
The CPU thinks one thing at a time. The GPU does a million tiny things at once — that's how games and AI run.
What you'll learn
- GPU
- Parallelism
- Pixels & Polygons
In this video
Okay, look at your screen right now. Every single tiny dot you see is called a pixel, and your computer has to pick a color for two million of them. And it has to do that sixty times every second. So how on earth does it keep up?
Watch the CPU try. The CPU is the main brain of your computer, and it's super speedy — but it can only paint one pixel at a time. See that timer? It just blew way past sixteen milliseconds, and the screen is barely filled. Frame missed.
So engineers built a totally different kind of chip. Meet the GPU. Look — see all those tiny dots lighting up? Each one is a little worker. The CPU has one really fast worker. The GPU has thousands of slower ones, all painting side by side.
And here's the magic trick. Instead of painting pixels one after another, the GPU hands one pixel to each worker. They all paint at the same time — that's called working in parallel. Boom, the whole screen is done before the timer hits sixteen.
Your turn — go play with this. Start with one worker, like a CPU, and see how slow it crawls. Then jump to four, sixteen, all the way up to two hundred and forty. Watch what happens to the time. Each new worker just shaves it down.
Pixels, matrix math, AI brains — they're all the exact same shape. Just millions of tiny jobs that don't depend on each other. So when ChatGPT writes you a story or an AI draws a picture, guess what's doing the heavy lifting? Yep, a GPU.
So here's the whole idea. CPU, one super-fast worker, perfect for thinking through tricky decisions. GPU, thousands of tiny workers, perfect for the same job repeated a million times. Your computer needs both — they're a team.
Topics
Ready to try it yourself?
The Pixel Pusher is a hands-on interactive video — watch it, then play with the idea live in your browser. Free, no sign-up.
Open in Kekula →