Keyboard Polling Rate: Does 1000Hz (or 8000Hz) Matter?
1000Hz is the sensible modern default, and going higher is marketing for almost everyone. Polling rate is how often the computer asks the keyboard for its key state: 125Hz = a report every 8ms, 1000Hz = every 1ms, 8000Hz = every 0.125ms. Moving from 1000Hz to 8000Hz shaves roughly 0.875ms of worst-case delay — less than one-seventh of a single frame on a 144Hz monitor (~6.9ms) and about 0.5% of a typical 150–250ms human reaction. It is imperceptible for typing and near-imperceptible even in esports. The spec that actually dwarfs polling — internal scan and debounce time, commonly ~5–20ms on budget boards — is the one nobody publishes.
The one thing to know: Polling rate is how often the keyboard reports to the PC: 125Hz = every 8ms, 500Hz = 2ms, 1000Hz = 1ms, and marketing '8000Hz' = 0.125ms. Going from 125Hz to 1000Hz cuts worst-case input delay from ~8ms to ~1ms — a real but tiny change that is completely imperceptible for typing and only marginally relevant to competitive gaming. Above 1000Hz the returns are effectively nil for humans and can raise CPU usage. It is one of the most over-marketed keyboard specs.
What is polling rate, exactly?
Polling rate is a property of how the host computer talks to the keyboard over USB (or a wireless dongle). The computer asks the keyboard for its current state at a fixed interval — the keyboard doesn't push a keypress the instant it happens; it answers the next poll. The polling interval is therefore the worst-case time a registered keypress can sit waiting before the computer hears about it. A higher polling rate does not make a key actuate sooner or register earlier inside the keyboard — it only shortens that waiting-room time between the keyboard and the PC.
The honest math: what each polling rate means in milliseconds
| Polling rate | Report interval (worst-case report delay) | Where you see it |
|---|---|---|
| 125Hz | 8ms | Older boards, some budget Bluetooth keyboards |
| 500Hz | 2ms | Some wireless and mid-range boards |
| 1000Hz | 1ms | The standard on modern wired and 2.4GHz boards |
| 8000Hz | 0.125ms | Flagship "gaming" boards — a marketing headline spec |
So the whole range of the argument — from the oldest 125Hz board to the flashiest 8000Hz flagship — spans just under 8 milliseconds of worst-case delay. That's the entire battlefield the marketing is fought over.
Put 1ms in perspective: frames and human reaction time
Numbers in milliseconds only mean something next to the rest of the chain (figures are approximate):
- One frame on a 144Hz monitor ≈ 6.9ms. Going 1000Hz → 8000Hz saves at most ~0.875ms — under one-seventh of one frame. Your display quantizes away most of the difference before you could ever see it.
- Human reaction time ≈ 150–250ms. The 8000Hz upgrade shaves roughly 0.5% of a single reaction. The variance in your own reactions from moment to moment is orders of magnitude larger than the entire polling difference.
- The full 125Hz → 1000Hz jump saves ~7ms — real, about one 144Hz frame, and the one polling upgrade that ever mattered. Everything above 1000Hz is subdividing a millisecond.
So does polling rate matter? The verdict by use case
- Typing: no, at any rate above 125Hz — and barely even then. Keystroke-to-keystroke timing in human typing varies by tens of milliseconds; a 1–8ms report interval is invisible inside that noise. 8000Hz is purely a marketing number for a typist.
- Everyday and casual gaming: 1000Hz is fully sufficient. It puts polling below every other latency source in the chain.
- Esports: 8000Hz is near-imperceptible, with an honest asterisk. In lab measurements, higher polling produces a small, real reduction in average input latency, and for the top fraction of a percent of competitive players stacking every marginal edge, that measurable-in-instruments difference is a defensible (if tiny) buy. For everyone else it changes nothing they can feel.
- 125Hz matters only as a floor. It's worth avoiding on a gaming board — mostly a concern on old or budget Bluetooth keyboards — but it is not a typing problem.
The spec nobody publishes: debounce and matrix scan often dwarf polling
Here is why polling-rate marketing is misdirection. Before a keypress ever reaches the USB polling stage, the keyboard has to detect it internally: the controller scans the key matrix row by row, and then applies debounce — a deliberate wait to confirm the noisy mechanical contact has settled and isn't going to register one press as several. On well-tuned boards that internal pipeline is a few milliseconds; on budget boards, debounce alone commonly runs ~5–20ms (approximate — almost no manufacturer publishes it). That means a budget board advertising "1000Hz polling!" can spend 15ms deciding a key was pressed and then report it with 1ms polling granularity — the headline spec is the smallest number in its own latency chain. A keyboard's true end-to-end input latency is scan + debounce + polling + wireless overhead, and the advertised polling rate is often the least important term in that sum.
A footnote on the other direction: 8000Hz polling generates up to 64× the USB report traffic of 125Hz, and each report is a CPU interrupt. On a modern desktop the cost is modest, but measurable CPU-usage increases at 8000Hz are documented — a spec that can cost real frames while saving fractions of a millisecond.
What to actually check instead of the polling number
- Connection type: 2.4GHz dongle vs Bluetooth. A 2.4GHz dongle is effectively wired-class (~1ms); Bluetooth adds latency and occasional microstutter you might feel in a fast game (never in typing). This one spec moves more milliseconds than any polling upgrade above 1000Hz. Wireless latency, decoded →
- If you play competitive FPS: rapid-trigger / adjustable actuation. That is the one switch-level feature with a real, mechanism-level competitive benefit — unlike polling above 1000Hz. Do gaming switches matter? →
- For everything else: hot-swap sockets, PBT keycaps, layout size, and QMK/VIA decide how a board feels and lasts. Polling rate above 1000Hz decides nothing you will ever notice.
Frequently Asked Questions
Polling rate is how often the host computer asks the keyboard to report its current key state, measured in hertz. 125Hz means a report every 8ms, 500Hz every 2ms, 1000Hz every 1ms, and 8000Hz every 0.125ms. The polling interval is the worst-case delay between a keypress being ready to report and the computer actually receiving it — it does not make keys register sooner, it only shortens how long a registered press waits to be reported.
Not for typing, and only marginally in lab conditions for elite competitive gaming. Going from 1000Hz to 8000Hz shaves at most ~0.875ms of worst-case report delay — under one-seventh of a single frame on a 144Hz monitor (~6.9ms per frame) and roughly 0.5% of a typical 150-250ms human reaction time. It is measurable with instruments, not perceptible to humans. 8000Hz can also raise CPU usage from the added USB interrupt traffic.
Yes. 1000Hz (a 1ms report interval) is the sensible modern default and has been standard on wired boards for years. It puts polling delay well below every other latency source in the chain — switch debounce, matrix scan, OS processing, and display refresh — so there is nothing meaningful left for a higher polling rate to remove.
For typing, no — an 8ms worst-case delay is imperceptible when human keystroke timing varies by tens of milliseconds. 125Hz mainly matters as a floor: it shows up on older boards and on some budget Bluetooth keyboards, and for fast-paced gaming an 8ms polling interval plus Bluetooth overhead can add up to something borderline noticeable. For a desk worker, even 125Hz is fine.
The keyboard's internal matrix scan rate and switch debounce time, which manufacturers almost never publish. Debounce alone commonly runs ~5-20ms on budget boards — several times larger than the entire 8ms gap between 125Hz and 1000Hz polling. Connection type also matters more: a 2.4GHz dongle is effectively wired-class (~1ms), while Bluetooth adds latency and occasional microstutter. Check connectivity and (if competitive) rapid-trigger support before caring about polling numbers above 1000Hz.
Yes, modestly. An 8000Hz device generates up to 64x the USB report traffic of a 125Hz device, and each report is an interrupt the CPU must service. On modern desktops the cost is small, but measurable CPU-usage increases at 8000Hz have been reported, and on weaker systems it can slightly reduce game performance — a spec that can cost frames while claiming to save fractions of a millisecond.
How we work
We're a research-based site: we analyze manufacturer specifications, switch and keycap engineering, and aggregated owner-feedback themes rather than lab-testing every board. We explain tradeoffs plainly, always name a downside, and report measurable specs — never a promise about comfort, strain, or health. Read our full methodology →