Mouse Test

Measure it. Pick a target distance, mark the start and the end of it on your mousepad, lock the pointer to the pad above, then drag the mouse in a straight line from one mark to the other while the page counts the sensor reports. DPI is simply counts divided by inches. Take three runs and compare them: numbers that repeat are numbers you can trust.

Sensor calibration varies. A few percent off is completely normal even on flagship sensors, and firmware DPI steps are often rounded to tidy numbers. Deviations above roughly 10 percent are worth knowing about, because your effective sensitivity in every game is scaled by exactly that amount: a mouse sold as 1600 DPI that really runs at 1800 makes every in-game sensitivity 12.5 percent faster than the number suggests.

For the standard measurement, yes. Enhance pointer precision on Windows makes the pointer travel further when you move faster, so counts stop being proportional to distance and the answer depends on how quickly you swiped. If your browser grants raw pointer input the tool bypasses the OS pointer pipeline entirely and acceleration no longer matters. The panel tells you which mode you are in before you measure.

An ISO/IEC 7810 ID-1 card, meaning any bank card, ID card or driving licence, is exactly 85.60 mm wide, so it is a ruler almost everyone already owns. Matching the on-screen outline to it gives your display true pixels per inch, which makes the on-screen ruler true to size and lets the page work out how far your hand actually has to move to cross your monitor.

No. DPI is sensor counts divided by how far the mouse moved on the desk, so only your physical measurement of that distance affects it: screen pixel density cancels out of that equation. The calibration powers the true-size ruler, your display real pixel density and the cursor-gain figure. It is useful, but it is not part of the DPI calculation, and a tool that claims otherwise is doing algebra that cancels.

The counting is exact; your ruler is not. Over a 4 inch drag, starting or stopping an eighth of an inch early is a 3 percent error, so use the longest distance you can and mark both ends properly. Three takes agreeing within about 1 percent is a good result. A wide spread means acceleration is still on, the drag curved, or the mouse lifted off the pad mid-swipe.

Move your mouse in fast, continuous circles inside the arena. The tool counts real HID events per second using coalesced pointer events, so it is not limited by your monitor refresh rate, and snaps the peak to the nearest standard rate: 125, 250, 500, 1000, 2000, 4000 or 8000 Hz. Give it 3 to 5 seconds of fast movement and read the Peak value. A mouse only reports while it is moving, so slow movement reads low. Chrome or Edge give the most accurate reading.

Yes, up to a point. Polling rate is how often your mouse reports position: 125 Hz is every 8 ms, 1000 Hz is every 1 ms. Going from 125 to 1000 Hz removes up to 7 ms of input delay and makes aim feel noticeably smoother, which is why 1000 Hz is the gaming standard. Against 500 Hz the difference is 1 ms versus 2 ms, so motion is a touch smoother and latency a hair lower; most people cannot reliably feel it, but virtually all gaming mice handle 1000 Hz with no downside, so there is rarely a reason not to use it. Only drop to 500 Hz if you hit stutters in old games or on weak CPUs. Above 1000 Hz the gains are fractions of a millisecond, mostly useful on 240 Hz and faster monitors, and very high rates can cost CPU time in some games.

Through your mouse software: Logitech G HUB, Razer Synapse, SteelSeries GG, Corsair iCUE, Glorious Core and similar apps all have a polling or report rate setting, usually 125, 250, 500 or 1000 Hz, with 2000 to 8000 Hz on supported mice, sometimes needing a special wireless dongle. Some mice also switch rates with an onboard button combo, often holding a DPI button while plugging in. Office mice without software are typically fixed at 125 Hz. Re-run this test afterwards to confirm the new rate.

Ultra-high rates stress the whole chain: the browser, OS scheduling and even the event timestamps. Chromium delivers coalesced events well past 1000 Hz, but under load some reports arrive late and land outside the measuring window. Treat 6000 to 8000 readings as a confirmed 8000 Hz mouse, and make sure 8K mode is actually enabled in your mouse software.

Click the pad 30 or more times at a normal pace with the button you suspect. Every press and release is timed to the millisecond, and any second press that lands within 70 ms of the previous release is flagged. A finger cannot press again that fast after letting go, so a flag during ordinary clicking means the switch fired twice on its own.

The metal contacts inside the click switch physically bounce apart for a fraction of a millisecond every time they close. Mouse firmware hides that with a debounce window, usually 8 to 20 ms. As the switch wears, the bounce lasts longer, and once it outlasts the debounce window the extra edge escapes as a second click. It is a mechanical fault inside the switch, not a Windows or driver problem.

Only as a stopgap. Some mice expose a debounce or click-latency setting in their configuration software, and raising it to 12 or 20 ms hides a mildly worn switch at the cost of that much extra click latency. It buys weeks or months, it does not repair the contacts. The real fix is replacing the switch, or replacing the mouse.

For deliberate repeat clicking most people land between 120 and 250 ms from release to next press, and the fastest sustained normal clicking sits around 60 to 80 ms. Genuine switch chatter looks nothing like that: the second press almost always arrives less than 20 ms after the release, and often under 5 ms. That is why the sub-20 ms count is shown separately.

Yes, and it is usually the same switch. A worn contact often chatters on release rather than on press, so the button briefly reports as let go part-way through a drag and the file, window or text selection is dropped. Test it by pressing inside the pad, moving the mouse, then releasing: any press shorter than 12 ms is flagged as a runt press.

Usually, yes. Support teams want something reproducible, and a click count, a suspected double-fire count and the shortest recorded gap is exactly that. Use the copy button to grab a plain-text summary, add your mouse model, the purchase date and which button is affected, and send it with the claim.

No. Nothing in the Bench posts a score. The old CPS page had a weekly leaderboard and a daily challenge; this one counts your clicks, prints the number and forgets it. Nothing leaves your browser in any of the four modes.