How to run a fair Wi-Fi speed test

A single speed test can mislead you. This page gives a method that lets you compare one test with another. It is based on the test framework in RFC 6349, which is the IETF document for measuring TCP throughput (RFC 6349).

Know the ceiling before you test

Work out the highest number your test can show. RFC 6349 gives the maximum TCP throughput for standard 1500-byte packets: 94.9 Mb/s on a 100 Mb/s link and 949.2 Mb/s on a 1000 Mb/s link (RFC 6349 §4.1.1). The same formula gives about 2373 Mb/s on a 2.5 Gb/s link (computed: 2500 × 1460 / 1538).

Your ceiling is the lowest of:

Step 1: Take a wired baseline

  1. Plug a computer into a router LAN port with a cable.
  2. Close other apps. Pause large downloads and streams on other devices.
  3. Run the test three times. Write down all three results.

If the wired median is close to your ceiling, the line from your provider is doing its job. Any shortfall over Wi-Fi is then inside your home.

Step 2: Test Wi-Fi the same way each time

  1. Stand in the same spot each time. Note the room and the distance.
  2. Note the band (2.4, 5, or 6 GHz) and the channel width your device shows, if it shows them.
  3. Run three tests. Use the same test server each time.
  4. Record the time of day. Repeat at a different time to see whether peak hours matter.

Step 3: Change one thing at a time

Move closer, change band, or change channel width. Change only one thing between runs, so you know which change made the difference.

What RFC 6349 says about window size

RFC 6349 explains that TCP throughput is limited by the window size and the round-trip time: a far server with a long round-trip time can show a lower result even on a fast line (RFC 6349 §3.3). Choose a nearby test server for a fair result.

What we do not claim

We publish no lab test results on this page. Any number you see here is a ceiling from RFC 6349, or computed from it.

Sources

UNSOURCED count: 0.