---
title: "Math for Memory, Throughput & Speedups"
description: "Napkin arithmetic for LLM inference: count the bytes a model needs in GPU memory, turn bandwidth into a speed ceiling, and check what a speedup is worth."
canonical_url: "https://fanout.sh/inference-eng/curriculum/00-01"
md_url: "https://fanout.sh/inference-eng/curriculum/00-01.md"
access: "public"
---

# Math for Memory, Throughput & Speedups

Napkin arithmetic for LLM inference: count the bytes a model needs in GPU memory, turn bandwidth into a speed ceiling, and check what a speedup is worth.

---
This representation contains public Fanout content only. Protected Pro lessons, account data, billing, checkout, and pricing are not included.

Browse the public content map: https://fanout.sh/sitemap.md