The settings checklist
Everything on this page is free and takes about twenty minutes in total. Between them these settings routinely remove more input latency than a display upgrade, and they make every calculation on this site actually apply to your machine.
1. Input path (five minutes)
This section makes every other number on this site true for your machine. If any of these is wrong, your sensitivity is not what you think it is.
| Setting | Set to | Why |
|---|---|---|
| Windows pointer speed | The sixth notch of eleven | The only position that passes your counts through unchanged. Detail. |
| Enhance pointer precision | Off | Windows mouse acceleration under a friendlier name. |
| Mouse software sensitivity | 1.0, or absent | Some drivers stack a multiplier on top of DPI. |
| Mouse acceleration and angle snapping in firmware | Off | Angle snapping decides you meant a straight line when you did not. |
| In-game raw input | On | Bypasses the OS pointer pipeline entirely. |
| In-game mouse acceleration and smoothing | Off | Named differently in every engine. Find it. Detail. |
| Mouse DPI | 800, or 1600 if you want a faster desktop | Above the granularity floor, below where engine rounding matters. Detail. |
| Mouse polling rate | 1000 Hz | Removes 3.5 ms versus 125 Hz. Detail. |
2. Latency (five minutes, biggest win)
| Setting | Set to | Worth |
|---|---|---|
| Nvidia Reflex / AMD Anti-Lag | On, in every game that offers it | Typically 10 to 40 ms when GPU-bound. Detail. |
| Framerate cap (with VRR) | 3 to 5 fps below refresh rate, set in game | Avoids the vsync ceiling penalty entirely. |
| Framerate cap (no VRR) | Uncapped, or 5 to 10% below your sustained GPU limit | Keeps the render queue from filling. |
| Driver-level vsync | On, when using VRR | Handles the ceiling case gracefully. |
| In-game vsync | Off | Redundant with the driver setting and usually higher latency. |
| Fullscreen mode | Exclusive fullscreen if the game offers it | Bypasses desktop composition on some configurations. |
| Network connection | Ethernet, not Wi-Fi | Lower latency and, more importantly, far lower variance. Detail. |
3. Display (five minutes)
| Setting | Set to | Why |
|---|---|---|
| Refresh rate in the OS | The panel's maximum | Windows frequently defaults to 60 Hz on a new connection. Check it. |
| Variable refresh | On, in BOTH the monitor menu and the GPU driver | The monitor-side setting is off by default on many panels. Detail. |
| Overdrive | The highest setting with no visible bright trail | Retune if you change refresh rate. Detail. |
| Monitor top edge | At or just below eye level | Neutral neck position. Detail. |
| Viewing distance | Far enough that the screen covers no more than 45 degrees | Keeps the whole screen inside your scanning arc. |
4. Audio (three minutes)
| Setting | Set to | Why |
|---|---|---|
| Game output mode | Stereo headphones | Headphones reproduce the two strong localisation cues perfectly. |
| Engine HRTF | On, if offered as a separate option | The game knows true 3D positions. Post-mix processing does not. Detail. |
| Windows spatial audio and headset surround | Off | Adds artefacts without adding position information. |
| Music volume | Zero | Occupies the frequency range movement cues live in, carries no information. |
| Dynamic range / night mode | On | Lets you raise overall level without gunfire being painful. Detail. |
| Driver equaliser and enhancements | Off | Apply EQ deliberately, in one place, or not at all. |
| Connection | Wired or 2.4 GHz dongle. Never Bluetooth. | Bluetooth adds tens to hundreds of milliseconds. Detail. |
5. Physical (two minutes, free)
- Move your keyboard left and rotate it 15 to 30 degrees. Free desk space for your mouse arm. Detail.
- Set your chair so your forearm is level with the desk and your shoulder is relaxed. Add a footrest if your feet no longer reach. Detail.
- Place your mouse's home position left of centre on the pad, since you have more room to the right.
- Lift your mouse cable clear of the desk if you use a wired mouse. Detail.
6. Verify (ten minutes)
- Measure your cm/360 by hand and check it matches the calculator. A mismatch means something is still scaling your input.
- Verify your polling rate with a test utility, checking consistency rather than just the average.
- Measure your latency with a phone camera before and after this checklist. Procedure.
- Check your 1% low framerate in a busy fight, not a menu, and compare it to your refresh rate. Detail.
If you only do three things
- Enable Reflex or Anti-Lag, or cap your framerate below your GPU's limit.
- Set polling to 1000 Hz and Windows pointer speed to 6/11.
- Check your monitor is actually running at its full refresh rate with VRR enabled on both sides.
Those three take five minutes and are worth more than most hardware purchases people are considering when they arrive here.
Related reading
- The upgrade order, which this checklist is items 3, 4 and 7 of.
- Input lag budget to see what each change is worth on your setup.
- Settling your sensitivity, which is item 1 and is not a setting.
Frequently asked questions
What is the single most important setting?
Enabling Reflex or Anti-Lag, or capping your framerate below your GPU's limit. The render queue is the largest variable latency term in most setups, frequently 20 to 50 ms, and it is a checkbox. Nothing else on this list is close.
Why does Windows pointer speed need to be at 6/11?
Because every other notch multiplies your mouse counts before any application sees them, and the notches below 6 discard counts entirely through rounding. The sixth notch applies a multiplier of exactly 1. See Windows pointer speed.
Should I cap my framerate?
If you use variable refresh, yes, 3 to 5 fps below your refresh rate, which keeps you inside the VRR window and avoids the vsync ceiling penalty. If you do not use variable refresh and accept tearing, an uncapped framerate well above your refresh rate gives you the freshest possible image.
Do these settings actually make a measurable difference?
The latency ones do, and you can verify them with a phone camera in ten minutes. See measuring latency. Measure once before and once after, and the difference will be obvious.
Is there a downside to any of these?
Two small ones. Framerate caps mean you are not seeing your highest possible number in a benchmark, which costs you nothing real. And Reflex can slightly reduce average framerate when heavily GPU-bound, which is a good trade for the latency it removes.