DPI vs polling rate: which mouse number actually matters?

How far the pointer moves when the mouse moves an inch. Dots per inch counts the position readings the sensor produces across that inch, so a higher figure means a bigger pointer movement for the same hand movement. Logitech G's own guide describes it as essentially the sensitivity setting for your mouse and says outright that a higher DPI means you can move the mouse faster.

That is a setting, and the word is doing a lot of work. Sensitivity is a preference tied to your screen resolution, how much desk you have, and whether you would rather move your wrist or your whole forearm. Two people can be equally good with a mouse and sit five times apart on this number.

You will also see CPI, counts per inch, used for the same idea. It is the more literal name for what the sensor does, since nothing here involves dots or printing. Vendors mostly say DPI because that is the term buyers know, and where a product page uses CPI it means the same number.

No, and the spec sheets are the argument. The published maximums in the table below run from 1,000 all the way to 35,000, a spread of thirty-five times. If that number tracked quality you would expect price to track it too, and it does not: a mouse capped at 2,000 or below can cost more than a gaming mouse advertising twenty times the figure, and one trackball here does exactly that.

The more revealing figure is the other end of the range. Where a mouse publishes a minimum DPI as well as a maximum, that minimum is 100, 200, 400, 500 or 1,000. A mouse advertised at 25,600 that starts at 100 is not telling you it works best at the top. It is telling you it has a slider, and the whole slider is the product.

What we are not going to do is tell you where on that slider to sit. It depends on your screen, your desk depth and how you like to move, none of which we can see from here. What is worth saying is what the box figure is: a ceiling rather than a setting, and a ceiling is only worth paying for if you were going to reach it.

One genuine sensor question does hide behind all this, and it is not the DPI figure. Logitech G distinguishes optical sensors, which it calls more accurate, from laser sensors, which it calls more versatile across surfaces but more prone to jitter and acceleration. Where a mouse names its sensor at all, that name tells you more than the DPI ceiling does.

It is how often the mouse tells the computer where it is. Logitech G's wording is that polling rate is how often your mouse relays its position, so a 1,000 Hz mouse relays it a thousand times a second. Razer describes the same thing as the frequency at which data is communicated from the device to the PC.

Convert that and it becomes concrete. A thousand reports a second is one every millisecond, and Razer puts the input delay at 1,000 Hz at approximately one millisecond for exactly that reason. Below that the gaps get wide enough to show up on a fast screen. At 125 Hz a report arrives every eight milliseconds, so on a monitor drawing 120 frames a second the mouse position used for a frame can already be a frame old. For ordinary office work, a stable connection matters more than chasing the highest polling rate.

This is the number that describes the mouse's share of whether a click feels immediate, and it is the one people almost never look at, because it is almost never printed. Where it is printed, 125 Hz, 1,000 Hz and 8,000 Hz are the figures you see.

It is a share rather than the whole thing, and that is worth being precise about, because a mouse is not the only thing standing between your hand and the picture. NVIDIA names five stages in its Reflex documentation: peripheral latency, game latency, render latency, display latency, and the total system latency they add up to. Peripheral latency is defined there as the time an input device takes to process a mechanical input and send the event to the PC, and polling rate governs part of that one stage. RTINGS makes the same point from the other end, noting that every piece of equipment introduces some delay and that you also need a monitor with low input lag.

So a mouse with a high report rate on a slow monitor in a heavy game is still going to feel slow, and no mouse specification will fix that. What polling rate buys is that the mouse is not the part making it worse.

For nearly everybody, no, and the most useful evidence is Razer's own page for the feature. Razer quotes 0.125 ms of input delay for a wireless mouse at 8,000 Hz against roughly 1 ms at 1,000 Hz, and then notes that in some instances your PC may not be powerful enough to support HyperPolling. A vendor telling you its flagship feature has a CPU cost is a vendor being straight with you, and it is the sentence to read twice.

The reason it can cost anything is that each report is work for the computer. Eight thousand of them a second is eight thousand interrupts a second, and that competes with whatever else the machine is doing. On a system with headroom you get a smaller number. On a system without it you can end up worse off than at 1,000 Hz.

There is also a display ceiling. A saved millisecond only shows up if the screen redraws often enough to render it, and a 60 Hz monitor draws a frame every 16 milliseconds. Going from 1 ms to 0.125 ms under a 16 ms frame is a real change to a number and not to anything you can see.

So the position, plainly: 1,000 Hz is where the meaningful gain already is, and 8,000 Hz is for a competitive player with a high-refresh monitor and CPU to spare. Razer DeathAdder V3 is the mouse here offering it, and it is a good mouse for reasons that have nothing to do with that figure.

Because 35,000 is a better number to print than 1,000, and because DPI has no ceiling in the way a report rate does. A sensor maker can keep raising the maximum, each generation gets a bigger figure on the box, and a buyer has no way to tell that the extra range is unreachable. Polling rate has a sensible value that most decent mice already hit, which makes for a poor marketing ladder. That is our reading of the incentive rather than something either company has said.

What gets printed is the shape of that incentive. A DPI ceiling is on nearly every product page and a polling rate on very few. The gap is not an oversight, because the same product pages that omit the report rate find room for a sensor's marketing name and a lighting mode.

Which makes one entry stand out. Razer DeathAdder V3 publishes 8,000 Hz and does not publish a maximum DPI at all, inverting the usual pattern completely. Whatever the reason, it leads with the number describing how the mouse behaves rather than how far the pointer travels.

Polling rate if it is published, and 1,000 Hz is the figure to want. If it is not published, which will be true most of the time, read the sensor name instead and treat the DPI ceiling as noise. A mouse from a company that names its sensor is a mouse whose maker thinks the sensor is worth naming.

Then stop optimising numbers, because the specification that decides whether you like a mouse is its shape, and there is no unit for that. Where your hand sits on the shell and how heavy the thing is will matter to you every day, long after you have forgotten what the DPI ceiling was. That question is worked through in Palm, claw or fingertip: which mouse grip do you use, and does it matter?

The other number worth a moment is how the mouse connects, which changes the same stage this page is about rather than a different one. It has its own piece: Wired, 2.4 GHz or Bluetooth: which mouse connection is actually fastest? And if you would rather stop reading specs and start comparing actual products, the setup builder puts the mice beside the keyboards and headphones.