sirupsen /napkin-math

Techniques and numbers for estimating system's performance from first-principles

★ 5,842⑂ 241👁 96⑃ 22● 19▤ 0
RustMITCreated 2019-06-01www.youtube.com/watch?v=IxkSlnrRFqc
TA的项目
+0
日增长
5,842
stars
241
forks
96
watchers
22
prs
19
issues
0
releases

README

Napkin Math

The goal of this project is to collect software, numbers, and techniques to
quickly estimate the expected performance of systems from first-principles. For
example, how quickly can you read 1 GB of memory? By composing these resources
you should be able to answer interesting questions like: how much storage cost
should you expect to pay for logging for an application with 100,000 RPS?

The best introduction to this skill is through my talk at
SRECON.

The best way to practise napkin math in the grand domain of computers is to work
on your own problems. The second-best is to subscribe to this
newsletter where you'll get a problem every few
weeks to practise on. It should only take you a few minutes to solve each one as your
facility with these techniques improve.

The archive of problems to practise with are
here. The solution will be in the following
newsletter.

 Numbers

Below are numbers rounded for memorization, not faux precision. Last measured
in October-November 2026, averaged over GCP C4 and AWS m8i (x86) and m8g
(Graviton4), 48 vCPU each, and their local-SSD variants; network disks on
Hyperdisk, EBS and pd-standard, HDDs on an AWS d3. Blob storage was measured in
March 2026, and the GPU rows are NVIDIA's specs. Most cells are one run per
machine, so trust the order of magnitude more than the digit. See
BENCHMARKING.md to run them yourself.

| Operation                                                               | Latency | Throughput |
| ----------------------------------------------------------------------- | ------- | ---------- |
| GPU Memory Sequential Read (B200)                                       |         | 7 TiB/s    |
| GPU ↔ GPU Transfer, NVLink (B200)                                       |         | 800 GiB/s  |
| CPU Decode Bitpacked Integers (512 B), all cores                        | 50 ns   | 500 GiB/s  |
| L1 Cache Read (64 B), 1 thread                                          | 2 ns    | 200 GiB/s  |
| Memory Sequential Read (64 B), all cores                                | 20 ns   | 200 GiB/s  |
| Memory Sequential Write (64 B), all cores                               | 30 ns   | 200 GiB/s  |
| Memory Random Read (64 B), all cores                                    | 150 ns  | 200 GiB/s  |
| Memory Random Write (64 B), all cores                                   | 150 ns  | 200 GiB/s  |
| L2 Cache Read (64 B), 1 thread                                          | 5 ns    | 150 GiB/s  |
| Memory Sequential Read (64 B), other socket, all cores‡                 | 100 ns  | 60 GiB/s   |
| CPU Decode Bitpacked Integers (512 B), 1 thread                         | 10 ns   | 50 GiB/s   |
| GPU ↔ GPU Network, RDMA (400 Gbit/s)                                    | 5 μs    | 50 GiB/s   |
| GPU ↔ CPU Transfer, PCIe (B200)                                         | 10 μs   | 50 GiB/s   |
| L3 Cache Read (64 B), 1 thread                                          | 50 ns   | 40 GiB/s   |
| Memory Sequential Read (64 B)

Top issues

Snapshot: 2026-11-16T09:23:14Z · Data license Apache-2.0 · JSON