Debug Foundry

Systems concepts, slowly enough to follow.

Each chapter is a small interactive simulation. Turn it, pause it, drag through the steps.

Free. No signup. Nothing to install.

Part I — Making LLMs fast

  1. The FlashAttention simulation: tiles moving from slow memory to fast on-chip memory Chapter 1 FlashAttention

    Watch attention drop its biggest memory bill, tile by tile. 8 steps, about 30 seconds.