Latency: Between Touch and Response

Define the clock boundaries

A player feels a physical press, but a test might start when a sensor crosses a threshold and stop when a screen changes. Between those points, the controller scans or samples the input, firmware processes it, a wired or wireless link carries data, the host receives it, software handles it, the game updates and the display presents a frame. Different methods include different parts of this chain. Their values can therefore disagree without either measurement being inherently wrong.

Distinguish controller contribution from end-to-end response. A controller-only measurement needs a clear input event and output event close to the device’s reporting boundary. An end-to-end test includes more variables and may better resemble play, but it cannot isolate the controller by itself. Write down the chosen boundaries before collecting data, and label the result accordingly.

Sampling

The moment or interval at which a system reads the control state.

Transport

The wired or wireless path used to deliver a report to the host.

End-to-end response

Elapsed time from a defined physical action to a defined visible or measured outcome.

Jitter

Variation in timing between repeated events, distinct from their average delay.

Why a single result can mislead

Timing is affected by sampling phase: a press just before a read may be reported sooner than one just after it. Wireless traffic, host load, application scheduling, frame timing and display refresh can add variation. Measurement tools have their own response time, threshold choices and uncertainty. An average alone hides the spread; a best-case number can make a system seem more consistent than it is. Report the distribution or at least a measure of variation alongside the central value.

Polling rate is not the same as latency. A faster reporting interval can reduce one portion of a signal path under particular conditions, but it does not remove game processing or display time. Nor does a nominal interval guarantee that every input arrives at that exact cadence. Avoid converting a published rate directly into a total-response claim without measuring the actual chain.

Testing methodology: compare like with like

For a high-speed camera or photodiode method, define how an input event is recognized and how the response threshold is selected. For software timestamps, account for clock resolution, buffering and the possibility that the timestamp is generated after the event. Use a control stimulus where possible and verify the whole measurement chain against a known timing reference. Keep the raw observations so readers can see how the summary was formed.

Confounders worth naming

Battery level, radio environment, connection quality, operating-system version, background tasks and the active game can all affect an end-to-end test. Display processing and refresh mode influence visible-response measurements. Button force and travel affect when a physical actuator crosses the chosen event threshold. If two devices are tested with different fixtures, alignment may create a false difference. Note these conditions rather than hiding them in a footnote.

Change one variable at a time when investigating a suspected effect. Compare wired and wireless modes on the same host and under the same measurement procedure, then repeat. If the difference vanishes on a second run, preserve that result as variation rather than selecting the more convenient measurement. Where samples are few, say so explicitly.

Interpreting latency in play

Lower measured delay may matter in timing-sensitive play, but a small numerical difference is not automatically perceptible or decisive for every player. Human response, game design, animation, network timing and display behaviour shape experience. A user report can identify a useful question—such as inconsistent response after waking—but cannot establish a controlled latency value on its own.

A credible timing statement names the event boundaries, test path, sample count and variation. Without those, a millisecond figure can suggest precision it has not earned.

Our aim is not to rank controllers by one headline number. It is to explain what the number includes, what it leaves out and whether the comparison was repeatable. A carefully bounded measurement is more informative than a universal claim built from an ambiguous stopwatch test.