# How much delay does controller polling rate add?

> Up to one USB interval per press: 8 ms at 125 Hz, 1 ms at 1000 Hz. How bInterval sets the rate, with published DualSense and Xbox Series figures.

- Canonical: https://cronuszenlab.com/reference/controller-polling-rate-explained
- Site: Zen Lab (https://cronuszenlab.com/)
- Author: Dana Rehn (https://cronuszenlab.com/authors/editorial)
- Section: Method
- Published: 2026-10-10
- Tags: polling rate, usb, bInterval, dualsense, xbox series controller, cronus zen

---

Polling adds a wait of between zero and one full interval to every button press, up to 8 ms at 125 Hz and up to 1 ms at 1000 Hz, because a report that is ready between two polls stays in the controller until the host asks again. Polling rate is how often the USB host requests a new input report. The controller's descriptor asks for an interval, and the host decides how to honor it.

## Where does the interval come from?

A USB controller sends its inputs over an interrupt endpoint, and that endpoint's descriptor carries a field called `bInterval`. Microsoft's driver reference for [USB_ENDPOINT_DESCRIPTOR](https://learn.microsoft.com/en-us/windows-hardware/drivers/ddi/usbspec/ns-usbspec-_usb_endpoint_descriptor) says the field holds the polling interval for interrupt and isochronous endpoints, reflects the device's configuration in firmware, and cannot be changed by drivers. It is a relative value, and its meaning depends on the link speed.

At full speed, `bInterval` counts 1 ms frames. Windows maps the requested value to a polling period with this table from the same page:

| `bInterval` requested (full speed) | Period Windows uses | Rate |
| --- | --- | --- |
| 1 | 1 ms | 1000 Hz |
| 2–3 | 2 ms | 500 Hz |
| 4–7 | 4 ms | 250 Hz |
| 8–15 | 8 ms | 125 Hz |
| 16–31 | 16 ms | 62.5 Hz |
| 32–255 | 32 ms | 31.25 Hz |

At high speed the unit is the microframe, eight to a 1 ms frame, and the period is `2^(bInterval-1)` microframes. A value of 1 polls every microframe (125 µs) and a value of 4 polls every eighth one, which is 1 ms. Microsoft's high-speed table reaches its longest period, 32 microframes or 4 ms, at a value of 6 and keeps it for every value up to 255. Low-speed devices get 8 ms at the shortest. Microsoft says these mappings apply to Windows 2000 and later.

Linux rounds in a similar way. The Arch Wiki's [Mouse polling rate](https://wiki.archlinux.org/title/Mouse_polling_rate) page says USB controllers round the interval down to the nearest power of two, so a 10 ms request runs at 8 ms and a 7 ms request at 4 ms.

## How the wait spreads across presses

The controller's chip sees the press, builds a report and holds it until the host polls. A report that is ready just after a poll waits almost a full interval, and one that is ready just before a poll barely waits. When presses land at random points in the cycle, the added delay spreads evenly from zero to one interval: across 8 ms at 125 Hz, across 1 ms at 1000 Hz.

The University of Regensburg group that built LagBox counts the polling interval as one of four contributors to a device's latency on its [project page](https://hci.uni-regensburg.de/projects/input-device-latency), next to switch mechanics, the device's microcontroller and the kernel's process-switching interval. Because the offset between the device's and the host's timing drifts, each device ends up with a latency distribution of its own. Raphael Wimmer, Andreas Schmid and Florian Bockes make the reporting point in their [CHI 2019 paper on USB input latency](https://hci.uni-regensburg.de/publications/on_the_latency_of_usb-connected_input_devices_2019): a mean and a standard deviation alone conceal the distributions that device internals and polling rates produce.

## Can the host poll faster than the controller asks?

On Windows, drivers cannot change `bInterval`, per Microsoft's reference. Linux exposes overrides in its usbhid driver. The Arch Wiki says the `mousepoll` option defaults to 0, which keeps the interval the device requests, and that `jspoll` and `kbpoll` do the same job for gamepads or joysticks and for keyboards. In the kernel source ([drivers/hid/usbhid/hid-core.c](https://github.com/torvalds/linux/blob/master/drivers/hid/usbhid/hid-core.c)), all three parameters start at 0, and a non-zero value replaces the descriptor's interval only when the HID collection declares itself a mouse, a joystick or a keyboard. The switch has no case for a gamepad usage, so `jspoll` reaches a pad only if it reports itself as a joystick. The Arch Wiki adds that the xhci-hcd driver used for USB 3 ports may ignore the override, and that rates above 1 kHz need a USB 2.0 High Speed link.

Forcing a faster rate helps some devices and not others. Wimmer, Schmid and Bockes found that forced polling at 1000 Hz decreased latency for some of their 36 devices but not all. Their project page reports no negative side effects for most devices, with one exception: an Xbox 360 wireless controller showed about 1 ms higher average latency when polled every 1 ms.

## Published polling rates for a DualSense and an Xbox Series controller

Gamepadla measures polling with a software tool that shows how often the operating system receives reports from the USB stack, separately from its Prometheus 82 latency hardware. Every row comes from a Windows PC, so the figures describe each pad on a PC host and say nothing about how a console or a Cronus Zen polls it. Cable and Bluetooth rows as listed in October 2026, each identified by the button test number Gamepadla prints on the row (dongle rows left out):

| Controller | Connection | Windows build | Median polling | Gamepadla row |
| --- | --- | --- | --- | --- |
| [DualSense](https://gamepadla.com/sony-dualsense.html) | Cable | Windows 11, 10.0.26100 | 249.92 Hz | #4753, 16 March 2025 |
| DualSense | Cable | Windows 11, 10.0.26200 | 250 Hz | #8150, 22 February 2026 |
| DualSense | Cable | Windows 11, 10.0.26200 | 250 Hz | #9164, 6 June 2026 |
| DualSense | Cable, labeled OC | Windows 11, 10.0.26200 | 500 Hz | #8151, 23 February 2026 |
| DualSense | Cable, labeled OC | Windows 11, 10.0.26200 | 1000 Hz | #8154, 23 February 2026 |
| DualSense | Bluetooth | Windows 11, 10.0.26100 | 542.42 Hz | #4752, 16 March 2025 |
| DualSense | Bluetooth | Windows 11, 10.0.26200 | 798.72 Hz | #9167, 6 June 2026 |
| [DualSense Edge](https://gamepadla.com/sony-dualsense-edge.html) | Cable | Windows 11, 10.0.26100 | 986.19 Hz | #4723, 14 March 2025 |
| DualSense Edge | Cable | Windows 10, 10.0.19045 | 1000 Hz | #9184, 6 June 2026 |
| [Xbox Series controller](https://gamepadla.com/xbox-core-controller.html) | Cable, XInput | Windows 11, 10.0.26100 | 125 Hz | #5397, 19 May 2025 |
| Xbox Series controller | Bluetooth | Windows 11, 10.0.26100 | 133.16 Hz | #5399, 19 May 2025 |

On cable, the standard DualSense rows without the OC label sit at about 250 Hz, a quarter of the Edge's rate on the same kind of connection. The Xbox Series controller's 125 Hz on a PC is an 8 ms cycle, and the same 8 ms figure turns up in an [Xbox Wire post](https://news.xbox.com/en-us/2020/03/16/xbox-series-x-latency/) Microsoft published on 16 March 2020 as the interval at which earlier Xbox controllers sampled and transmitted their inputs. That post credits the Series controller's Dynamic Latency Input with sampling and sending the most recent input just ahead of the game's request. It describes the Xbox Series X console, and Gamepadla's row comes from a PC, so the two figures do not share a host.

## Two polling links through a Cronus Zen

A Zen sits between pad and console: it takes in the controller's report, lets the loaded script act on it and passes its own report on. A wired pad's input therefore crosses two USB polling schedules, controller to Zen and Zen to console, and each can add its own wait of up to one interval. A wireless pad adds a radio hop on the controller side, over Bluetooth to the Zen or through a wireless adapter plugged into it.

The [guide Cronus publishes](https://guide.cronuszen.com) was searched in full for this note in October 2026 and again on 14 November, and it states no polling rate for either link. Polling comes up only for third-party controllers: the [controller page](https://guide.cronuszen.com/supported-devices/controller) lists modes such as "XInput mode (Xbox 360® protocol, 250 Hz)", and setup steps include setting a GameSir G7 Pro to 250 Hz for PlayStation use. The one timing entry in its [2024 changelog](https://guide.cronuszen.com/changelogs/archive/changelog-24) belongs to firmware 2.2.8 (4 December 2024), which lists "Output protocol cycle 16 ms issue" among its fixes without saying more.

Marketplace copy fills the gap without a source. An undated "wiki" article on a large online marketplace, which states that its content comes from third-party contributors or is generated by AI, claims the Zen optimizes polling from a default of 1 ms (1000 Hz) to 0.5 ms (2000 Hz). It cites nothing. A 0.5 ms period is shorter than the 1 ms frame Microsoft's tables set as the full-speed minimum, and the Arch Wiki ties rates above 1 kHz to USB 2.0 High Speed, so the claim would need a high-speed link on the console side, and no Cronus figure confirms one.

## Reading a controller's descriptor

On Windows, Microsoft's [USBView](https://learn.microsoft.com/en-us/windows-hardware/drivers/debugger/usbview) (usbview.exe, installed with the Debugging Tools for Windows option of the Windows SDK) shows the device, configuration, interface and endpoint descriptors of a connected USB device. On Linux, `lsusb` lists devices with their vendor and product IDs, and the Arch Wiki's page shows how to filter one device's descriptor down to its interval:

```bash
lsusb
lsusb -vd VVVV:PPPP | grep bInterval
```

The value to read sits on the endpoint whose direction is IN and whose type is interrupt. The Arch Wiki notes that it shows the interval the device requests, which may differ from the one in use. The rate a console or a Zen grants can be established only by measuring on that host. The instruments able to do that are compared in [how input latency is measured](/reference/how-input-latency-is-measured), and [the full latency budget](/reference/what-adds-controller-input-latency) places polling among the other stages.

## FAQ

**What is controller polling rate?**

How often the USB host asks a controller for a new input report. At 1000 Hz the host asks every 1 ms; at 125 Hz, every 8 ms.

**What polling rate does a DualSense use?**

On cable, Gamepadla's Windows 11 rows without the OC label read 249.92 to 250 Hz on three test dates. Its Bluetooth rows read 542.42 Hz and 798.72 Hz, and all of these describe a PC host only.

**Does the Cronus Zen change polling rate?**

Cronus's guide gives no polling figure for either link through the Zen, controller to Zen or Zen to console. A marketplace claim of 2000 Hz cites no source.

**Does a higher polling rate always lower latency?**

No. Wimmer, Schmid and Bockes (CHI 2019) found that forcing 1000 Hz polling lowered latency for some USB input devices and not for others.

**How can a controller's polling interval be checked?**

Read bInterval on the controller's interrupt IN endpoint with USBView on Windows or lsusb -v on Linux. That value is what the controller requests, and the host may round it.

