Latency breakdown
Not testedHow to test
About 30 sTipRelease each button before the next press. Holding a button counts as one press.
Where a press spends its time
6 stagesThis controller input lag test splits a press into six stages, from button to screen. The page measures the middle two, estimates the two before them from their longest possible time, and can't see inside the controller or the display. The bars share one 0–20 ms scale.
Refresh rate vs polling rate
Average waitTo reduce controller input lag, raise the screen's refresh rate before the polling rate. On average a press waits half a report interval for the controller but about 1.5 frames for the screen, so the screen costs more.
When the result looks wrong
5 casesMeasure full input lag
Slow motionA page can't see inside the controller or display, so the full delay takes slow motion. At 240 fps a frame is 4.17 ms: if the button goes down in frame 1 and the screen changes in frame 4, that's 3 frames, or about 12.5 ms.
Film the controller and screen together at 240 fps
Where the button goes down and where the screen changes
Multiply the frames by 4.17 ms. Repeat and take the median
Frequently asked questions
11 questionsWhat is normal controller input lag?
Depends on refresh rateThis controller latency test measures the browser side, from report to screen, and most of that is waiting for the next frame and showing it. It usually comes to about 25–30 ms at 60 Hz and 12–15 ms at 144 Hz. If yours is higher, start with the highlighted stage in the breakdown.
How accurate is a web gamepad latency test?
Partly measuredPage read and To screen are measured directly. For Report wait and Browser read, the page knows the maximum and uses the average. Time inside the controller and the display isn't included; for the full delay, film in slow motion.
Why is the report wait missing for an Xbox controller?
No WebHIDOn Windows, Microsoft's Xbox controllers use their own driver and aren't HID devices, so they don't appear in the WebHID list. The page can't time their reports, so the report wait is hatched and left out of the result.
Is Bluetooth slower than wired?
Compare the intervalThe connection matters mostly through the report interval: on average a press waits half of it. With WebHID, run one round wired and one over Bluetooth, then compare the Interval cell. WebHID works with PS5 and PS4 controllers. Compare connections in the polling rate test →
Does a higher refresh rate reduce input lag?
Yes, more than polling rateGoing from 60 Hz to 144 Hz cuts a frame from 16.7 ms to 6.9 ms. At about 1.5 frames on average, that saves about 14.7 ms. Raising the polling rate from 125 Hz to 1000 Hz saves about 3.5 ms.
What is the difference between polling rate and latency?
How often vs how longPolling rate is how often the controller sends a report, in Hz. Latency is how long a press takes to reach the screen, in ms. Polling rate sets only one part of it: a press waits up to 1 ms at 1000 Hz and up to 8 ms at 125 Hz.
Is a higher polling rate always better?
Gains shrinkOn average a press waits half a report interval: 4 ms at 125 Hz, 0.5 ms at 1000 Hz. Above 1000 Hz each step saves a fraction of a millisecond, which is small next to one frame.
Is this the same as input delay in games?
Games add moreGames add their own render queue, V-Sync and frame generation, which can add several frames. This page draws one simple frame, so it shows the lowest input delay on the browser side. Use it to compare controllers, connections and screen settings.
Why 20 presses?
For a steady medianEach press lands at a different point between two frames, so a single reading jumps between zero and one frame. The median of 20 is steadier. A press during a page stall over 100 ms still gets a row, marked Stall, but isn't counted.
How do I measure the full input lag?
Film in slow motionFilm the controller and screen at 240 fps. Count the frames from the button going down to the screen changing, and multiply by 4.17 ms. A new row appearing in the test is an easy screen change to spot.
Is any data uploaded?
NoPress times are worked out in the browser. Each round is kept on this page only and cleared when you refresh or close it. Privacy policy →