Skip to the notes

What adds latency between a controller press and the screen

Every stage from a controller press to the screen adds delay. Published figures and bounds for each, from pad firmware and USB polling to the game frame and TV.

Dana RehnTechnical editor

Filed Revised Nov 14, 20266 min read

Latency between a controller press and the screen is the sum of every stage in between, and only some of those stages have a published figure or bound. Seth Schneider's Reflex article for NVIDIA (1 September 2020) sorts the stages into peripheral, PC and display latency. On a console, the console takes the PC's place. A pass-through device such as the Cronus Zen falls inside the peripheral segment, because the console receives the Zen's reports in place of the controller's.

The budget, stage by stage

Stage Segment Published figure or bound Source
Switch and controller firmware Peripheral No general figure. Whole-pad averages on cable, PC host: 6.50 ms for an Xbox Series controller, 7.81 to 11.3 ms for a DualSense across three dates Gamepadla, 2025–2026
USB polling, wired Peripheral Between zero and one full interval for each press (8 ms at 125 Hz, 1 ms at 1000 Hz) Microsoft Learn; interval arithmetic
Wireless link Peripheral Bluetooth LE minimum connection interval 7.5 ms historically, 375 µs with Core 6.2 when both sides support it; no controller named. Microsoft's post gives no interval for Xbox Wireless Bluetooth SIG; Xbox Wire
Pass-through device (Cronus Zen) Peripheral No latency figure in Cronus's guide; its scripting guide asks for VM load comfortably below 80 percent Cronus GPC script guide
Console input handling Console Earlier Xbox controllers sampled and transmitted every 8 ms; Dynamic Latency Input sends the newest input just before the game asks Xbox Wire, 16 March 2020
Game frame Console 16.7 ms per frame at 60 fps, 8.3 ms at 120 fps; going from 60 to 120 fps halves a game's internal latency Frame arithmetic; Xbox Wire
HDMI transmission Display Frame transmission cut from 16.6 ms to 8.3 ms with HDMI 2.1 Xbox Wire
Display processing Display Game Mode named the most important setting; test floor 4.17 ms on a 120 Hz TV, 8.33 ms on a 60 Hz TV RTINGS

The stages that run on cycles of their own, polling and the game frame among them, add a wait that depends on where a press lands in the cycle. That is one reason two presses on the same setup can arrive at different times.

Inside the controller

A device's own share has four parts in the model on the Regensburg LagBox project page: how long the switch takes mechanically, the microcontroller's processing, the USB polling interval, and how often the host kernel switches processes. When the same group timed 36 USB input devices for a CHI 2019 paper, the devices differed in their means and also in the shape of their latency distributions.

Gamepadla publishes whole-controller figures, all taken on PC hosts. Its Prometheus 82 rig, which presses the button with a solenoid, averaged 6.50 ms for the Xbox Series controller on cable in a Windows 11 test of 19 May 2025. Its cable rows for the DualSense without the OC label average 7.81 ms (16 March 2025), 8.27 ms (22 February 2026) and 11.3 ms (6 June 2026), from different testers and Prometheus 82 versions. One pad moving that far between dates is a reason to read any outside figure with its date and host attached.

On a cable, each press waits from zero to one polling interval before the host collects it. The polling note covers where the interval comes from and lists Gamepadla's polling figures for the DualSense and the Xbox Series pad.

A radio link runs on a schedule of its own. The Bluetooth SIG's Core 6.2 feature overview says the minimum Bluetooth LE connection interval was historically 7.5 ms with a 1.25 ms resolution, and that Shorter Connection Intervals reduce it to 375 µs with a 125 µs resolution, aimed at high-performance HID devices and latency-sensitive uses such as gaming. Both peers have to advertise support; otherwise the connection stays on the baseline values. The overview names no controller, and it describes itself as a marketing document that does not override the Core Specification. Microsoft's Xbox Wire post gives no radio interval for Xbox Wireless.

Gamepadla's rows of 19 May 2025 show what the radio did to one pad on a PC. The Xbox Series controller averaged 6.50 ms on cable and 12.23 ms over Bluetooth, and its button jitter rose from 2.09 ms to 3.67 ms.

A pass-through device

Every input bound for the console passes through the Zen first. The device takes the controller's report, applies whatever the active script does, and emits a fresh report over its own USB link to the console, which adds a processing stage and a second polling schedule. Cronus's guide gives no latency figure for the device. What its scripting guide offers is a budget: the performance budget section says to keep VM load comfortably below 80 percent, and to move work such as display rendering and full table transformations behind events or dirty flags.

Cronus's 2024 firmware changelog touches timing without giving a latency figure. Under version 2.2.7 (22 October 2024) it lists PS5 driver CPU usage as improved, and among the fixes in 2.2.8 (4 December 2024) is an item titled "Output protocol cycle 16 ms issue". Elsewhere the guide lists "Ultra-low latency" among Zen Link's features, with no number attached.

Zen Lab does not publish original numerical estimates for the Zen's added latency. How input latency is measured sets out which instruments can isolate a single stage. Troubleshooting a slow or uneven Zen chain sits in a separate guide.

Console, game frame and HDMI

Microsoft's Xbox Wire post of 16 March 2020 covers the console end. It puts the input cycle of earlier Xbox controllers at 8 ms, and says that with Dynamic Latency Input on Xbox Series X the controller samples and transmits the latest input right before the game requests it.

The game frame sets a floor of its own. At 60 fps one frame lasts 1/60 s, about 16.7 ms; at 120 fps it lasts about 8.3 ms. The same post says games can double their frame rate from 60 to 120 fps, which halves their internal latency. It also says HDMI 2.1 let the team cut frame transmission time from 16.6 ms to 8.3 ms, and that the console invokes Auto Low Latency Mode on capable TVs.

The display

RTINGS's input lag test page (updated 3 October 2025) splits a TV's share into acquiring the image, processing it and displaying it. It names Game Mode as the most important setting for the lowest input lag and notes that many TVs offer Auto Low Latency Mode, which switches into Game Mode when a game launches from a compatible device. RTINGS singles out motion interpolation: on most TVs it cannot run inside Game Mode, so turning it on means leaving the low-lag mode.

RTINGS's own figure covers the display alone. A photodiode at the center of the screen stops the clock when the pixels start to change, and pixel response time is left out. Because a TV draws from the top down, the center is reached halfway through a refresh, so the lowest figure the test can report is 4.17 ms on a 120 Hz TV and 8.33 ms on a 60 Hz one.

Reader questions

What causes controller input lag?
Every stage between the button and the screen adds some: the controller's firmware, USB polling or the radio link, any pass-through device, the console's input handling, the game frame, HDMI transmission and the TV's processing.
How long is one frame at 60 fps and 120 fps?
About 16.7 ms at 60 fps and 8.3 ms at 120 fps. Microsoft's Xbox Wire post of 16 March 2020 says doubling a game's frame rate from 60 to 120 fps halves its internal latency.
Does Cronus publish an input lag figure for the Cronus Zen?
Cronus's current guide gives none. The Zen adds a processing stage and a second USB link between controller and console, and the scripting guide sets a budget instead: keep VM load comfortably below 80 percent.
Does a wireless controller add latency?
In Gamepadla's Windows 11 tests of 19 May 2025, the Xbox Series controller averaged 6.50 ms on cable and 12.23 ms over Bluetooth, with button jitter of 2.09 ms and 3.67 ms.

A figure looks wrong? Report an error

Further notes