Keyboard latency and polling rate explained

Many sites offer a keyboard latency test or a polling rate checker, but a web page only sees key events after the keyboard, the USB or wireless link and the operating system have done their work. This guide explains each part of the delay between pressing a key and seeing the result, what browser timing can honestly show, and what you need to measure true latency.

Where keyboard delay comes from

Between your finger and the screen there is a chain of steps, each adding some time:

Travel and actuation: the key moves until the switch registers. On mechanical switches that point is fixed; on analog (Hall effect or similar) keyboards it can be adjustable.
Scanning: the keyboard's controller checks the key matrix repeatedly. How often it does so is the scan rate.
Debounce: the controller waits or filters so that a bouncing contact is not reported as several presses.
Polling and transport: over USB the computer asks the keyboard for new data at its polling rate; wireless and Bluetooth links have their own timing.
Operating system and app: the input is processed and delivered to the program.
Rendering and display: the program draws a new frame and the monitor shows it.

End-to-end latency is the sum of all of these. Marketing figures usually describe only one part.

Polling rate in plain numbers

Polling rate is how often the computer collects reports from a USB keyboard. At 125 Hz the interval is 8 ms, at 1000 Hz it is 1 ms and at 8000 Hz it is 0.125 ms. A higher rate shortens the wait for the next report, but only that wait: scan rate, debounce, the system and the display still add their share. Polling rate also says nothing about how quickly the keyboard's own controller notices the key.

What a browser can see

A web page receives keydown and keyup events with timestamps. From those it can show:

the time between a press and its release, and between consecutive presses, which is how the chatter test spots switch bounce;
which keys are held together, as in the ghosting test;
the exact key value and code, as in the keycode tester.

It cannot see when you physically pressed the key, only when the event arrived. Timestamps are also rounded by browsers for privacy, sometimes coarsely, and the system may group several input events together. A key event timestamp is therefore not a measurement of keyboard latency.

Why a website cannot certify keyboard polling rate

Mouse movement produces a continuous stream of reports, so the gaps between events can give a rough estimate of a mouse's rate, which is what the mouse polling rate test does. A keyboard sends a report when keys change state, and you cannot press keys anywhere near 1000 times per second. With no steady stream to time, the browser has nothing to estimate a keyboard polling rate from. A site that prints "your keyboard runs at 1000 Hz" from typing alone is guessing. The setting in your keyboard's software, or a USB analysis tool, is the reliable source.

Actuation point and rapid trigger

On analog keyboards the actuation point (how far the key travels before it registers) and rapid trigger (registering a release as soon as the key starts to rise) are measured by sensors inside the keyboard. The browser only receives the resulting press and release, not the key's position, so a web page cannot measure actuation distance. Use the keyboard's own software, which reads the sensor values directly. What a browser test can confirm is that presses and releases arrive when you expect them, for example by tapping quickly in the keyboard test.

How to measure real latency

True end-to-end latency needs a reference outside the computer: a high-speed camera filming both the key and the screen, or a dedicated latency tester that detects the press and the change on screen. Film or record many presses and average them, because individual results vary with the timing of frames and polls. If you use a camera:

use the same scene, display, refresh rate and settings for every keyboard;
press firmly and consistently;
count frames from the moment the key bottoms out or the switch is known to actuate to the first visible change.

Comparing keyboards fairly

Change one thing at a time: the same port, cable, computer, display and app for every keyboard.
Wired results do not tell you about the same keyboard in wireless or Bluetooth mode; test each mode separately.
Record many samples and compare ranges as well as averages.
Treat any single number from a website, including a browser timing, as a hint rather than a measurement.

Questions

Can a browser measure keyboard latency?

Not true latency. A page only sees when the key event arrived, not when you pressed the key, and its timestamps are rounded. Measuring latency needs an external reference such as a high-speed camera or a dedicated tester.

How do I check my keyboard's polling rate?

Look in the keyboard's software or settings, which usually show and change it. A website cannot reliably estimate keyboard polling rate from typing, because keys do not produce a steady stream of reports.

Is 8000 Hz better than 1000 Hz?

It shortens one part of the delay, from up to 1 ms to up to 0.125 ms between polls. Whether you notice depends on the other parts of the chain, which often add much more.

Does this site have a keyboard latency test?

No, deliberately. TestDock's keyboard tools measure what a browser can see reliably: which keys arrive, how many at once, and press timing for chatter. This guide explains how to measure latency properly.