Is my mouse's DPI accurate?
Not exactly, but close, and the error is a fixed percentage you can correct once. Across 16 current mice TechPowerUp measured, the median miss was 2.5 percent, 14 of 17 samples landed within 3.8 percent, and the worst read 11.3 percent high. Only the two Logitech mice hit their numbers exactly.
Not exactly, but usually within a few percent. TechPowerUp measured CPI deviation on 16 current gaming mice. Only the two Logitech mice were exact; the median miss across all 17 samples was 2.5 percent, and the Akko Dash Ultra read 11.3 percent high. On every mouse the error was the same percentage at 400, 800, 1600 and 3200, so it is a fixed scale factor you can correct once.
Measured deviation, 16 mice
TechPowerUp's reviews test the four most common steps and chart the deviation of actual from nominal CPI. The reviewer lists the usual causes as "firmware, mounting height of the sensor not meeting specifications, mouse feet thickness, mousing surface, among others", and warns that the variance "will still differ from unit to unit".
| Mouse | Sensor | Deviation | 800 is really | cm/360 |
|---|---|---|---|---|
| Akko Dash Ultra | PAW3950 | +11.3% | 890 | 46.70 |
| Razer Viper V4 Pro | Focus Pro 50K Gen-3 | +6.3% | 850 | 48.90 |
| Pulsar X2N Crazylight (Medium) | XS-1 (PAW3950) | +6.3% | 850 | 48.90 |
| Pulsar X2N Crazylight (Mini) | XS-1 (PAW3950) | +3.8% | 830 | 50.08 |
| Razer Viper V3 Pro | Focus Pro 35K Gen-2 | +3.8% | 830 | 50.08 |
| Vaxee NP-01S V3 Wireless | Core1 (PAW3955) | +3.8% | 830 | 50.08 |
| Turtle Beach Burst II Pro | PAW3950 | +3.1% | 825 | 50.38 |
| Razer DeathAdder V4 Pro | Focus Pro 45K Gen-2 | +2.5% | 820 | 50.69 |
| SCYROX V8 | PAW3950 | +2.5% | 820 | 50.69 |
| Glorious Model D3 Wireless | BAMF 3.0 (PAW3950) | +1.9% | 815 | 51.00 |
| ASUS ROG Harpe II Ace | AimPoint Pro | +1.3% | 810 | 51.31 |
| Logitech Pro X Superlight 2 | HERO 2 | 0 | 800 | 51.95 |
| Logitech Pro X2 Superstrike | HERO 2 | 0 | 800 | 51.95 |
| Rapoo VT2 Max | PAW3950 Ultimate | -0.6% | 795 | 52.28 |
| Endgame Gear OP1 8K V2 | PAW3950 | -1.9% | 785 | 52.95 |
| LAMZU Maya X | PAW3950 | -2.5% | 780 | 53.29 |
| G-Wolves HTX Ultra | PAW3950 | -3.8% | 770 | 53.98 |
Eleven samples read high, four read low and two were exact. Six landed within 2 percent, 14 within 3.8 percent, and three passed 6 percent. The spread from the slowest to the fastest is wide: set to the same 800, the HTX Ultra reports 770 counts per inch and the Dash Ultra 890, a 15.6 percent difference in real sensitivity.
One percentage at every step
Two ideas circulate about this. One is that the number in the software is the number you get. The other, repeated on some brand blogs, is that deviation varies by setting, perhaps none at 400 and a few percent at 3200. Neither fits these 16 mice. Each missed, and each missed by the same percentage at all four steps: the Viper V4 Pro was +6.3 percent at 400 (+25 CPI) and +6.3 percent at 3200 (+200 CPI).
So the error is a scale factor, not noise. It does not make your aim less consistent, which is why it hides: your hand adapts to whatever the real figure is. It surfaces the day you change mouse, copy a professional's settings, or convert a sensitivity and find the result feels faster than the arithmetic promised.
It is the mouse, not the sensor
Ten of the 17 samples use a PAW3950 or a customised one, and they span -3.8 to +11.3 percent. The clearest case is the Pulsar X2N Crazylight: the same sensor in two sizes, and the Medium measured +6.3 percent while the Mini measured +3.8. Buying a mouse for its sensor name tells you nothing about its CPI accuracy, which fits our sensor guide's point that the sensor is rarely the variable left that matters.
What it does to your cm/360
Your real cm/360 uses the counts the mouse actually sends:
An exact 800 gives 51.95 cm. Move from a Superlight 2 to a Viper V4 Pro with the same settings and your 360 shrinks by 3.05 cm, which is 5.9 percent faster. See cm/360 for why that number is the one to keep fixed.
How to correct it
Pick one of two fixes.
- Lower the DPI step. Set nominal = target / (1 + deviation). TechPowerUp did exactly this for its own testing, using 375, 750, 1500 and 3000 on the Viper V4 Pro and 350, 700, 1450 and 2900 on the Akko. 750 on the V4 Pro gives about 797 real CPI. The Viper V3 Pro and DeathAdder V4 Pro accept CPI in steps of 1, per TechPowerUp, so on those you can land almost exactly.
- Keep the DPI, scale the sensitivity. Multiply your in-game value by nominal over actual. Viper V4 Pro at 800: 800 / 850 = 0.941, so sensitivity 1.00 becomes 0.94.
The second fix leaves your desktop speed alone; the first fixes every game at once.
Measuring your own mouse
Your unit is not TechPowerUp's unit, so measure it. In CS2, mark a reference point, turn exactly 360 degrees slowly against a ruler, and solve the formula above for CPI:
Repeat three times and average. A ruler read to 0.5 cm on a 52 cm turn is about 1 percent, so this resolves anything above roughly 1 percent, which covers 14 of the 17 samples here. Any game with a known yaw works; ours are in the game sensitivity constants table.
Where this stops applying
Each figure is one review sample, and TechPowerUp says unit-to-unit variance is real. Its listed causes include mouse feet thickness and the mousing surface, so a new pad or replacement skates can move your figure. Firmware updates can too.
And the error is not always fixed. TechPowerUp found the Pro X2 Superstrike, exact by default, would "occasionally, and for reasons unknown" switch to a different set of effective values, which its chart shows at -3.8 percent on every step. It reproduced this on two pads and calls the cause unclear. If a Superstrike suddenly feels slow, re-measure before changing your sensitivity.
What to do
- Expect your mouse to be a few percent off: median 2.5 percent, worst 11.3, on 16 current mice.
- The error is one percentage at every step, so correct it once.
- Correct when you change mouse or copy settings; your consistency is unaffected either way.
- Fix it with a lower DPI step or sensitivity x nominal / actual, then measure your own unit with a ruler.
Frequently asked questions
How far off is a gaming mouse's DPI?
Usually a few percent. Across 17 samples of 16 current mice measured by TechPowerUp, the median deviation was 2.5 percent and 14 of the 17 were within 3.8 percent. Three were above 6 percent: the Razer Viper V4 Pro and the medium Pulsar X2N Crazylight at +6.3 percent, and the Akko Dash Ultra at +11.3 percent.
Is DPI deviation different at 400, 800, 1600 and 3200?
Not on any of the 16 mice here. Each one missed by the same percentage at all four steps, so the error is a scale factor, not something that grows at high DPI. A single correction fixes every step.
Does DPI deviation affect my aim?
Not your consistency, because the error is constant. It changes your real cm/360, which matters when you move to a new mouse or copy someone else's settings. At 800 DPI and CS2 sensitivity 1, a mouse 6.3 percent high turns a 51.95 cm 360 into 48.90 cm.
How do I correct DPI deviation?
Divide by the error. Either set a lower DPI step (TechPowerUp used 375 on the Viper V4 Pro to land near 400) or keep the DPI and multiply your in-game sensitivity by nominal over actual, for example 800 / 850 = 0.94.