NVIDIA Reflex sits in the settings menu of most competitive games now, usually near V-Sync and frame rate limits, offering Off, On, and On + Boost. Plenty of players toggle it because a video told them to and then check their frame counter, see no change, and turn it back off. That is a reasonable reaction to a badly explained feature, because Reflex is not measured in frames per second and was never designed to be.
What Reflex targets is system latency: the delay between your hand moving the mouse and the corresponding change appearing on your monitor. That delay is made up of input processing, CPU simulation, the render queue, GPU rendering, and display scanout. Reflex attacks one specific part of that chain, and it does so well enough that it has become a default expectation in shooters. Here is what each option does, when Boost helps, and why the frame counter is the wrong place to look for the result.
Quick Verdict
Turn Reflex on in every supported competitive title. Use On + Boost only when you are CPU-bound and are not worried about power draw or heat, and test it rather than assuming. Expect the benefit to show up as a more immediate-feeling mouse, not as a higher number in the corner of the screen. In GPU-bound scenarios Reflex can even trim a few frames off your peak while cutting latency substantially, which is a trade worth making.
| Setting | What it does | Effect on latency | Effect on frame rate | When to use it |
|---|---|---|---|---|
| Reflex: Off | Game submits work normally, frames can queue ahead | Baseline, highest queue delay when GPU-bound | Highest uncapped peak | Benchmarking only, or single-player titles where latency is irrelevant |
| Reflex: On | Paces CPU submission so the render queue stays near empty | Largest single improvement, most visible when GPU-bound | Unchanged or slightly lower at the peak | Default for all competitive play |
| Reflex: On + Boost | As above, plus keeps GPU clocks elevated instead of dropping them | Small further reduction, mainly in CPU-bound scenes | Unchanged | CPU-bound titles, desktops with thermal headroom |
| Driver Low Latency Mode | Limits queued frames at driver level | Similar goal, older mechanism | Minor | DirectX 9 and 11 games only, no effect in DX12 or Vulkan |
| Reflex Latency Analyzer | Measures end-to-end latency in hardware | Measures rather than reduces | None | Requires a compatible G-SYNC monitor and supported mouse |
| Reflex with Frame Generation | Enabled automatically alongside DLSS frame generation | Offsets the latency frame generation adds | Frame generation raises displayed frame rate | Whenever frame generation is used |
How to Do It
1. Find Reflex in the game, not the driver
Reflex is an integrated feature, not a global toggle. Developers add it through NVIDIA’s SDK, which is why you set it inside each game’s graphics or display options rather than in the NVIDIA Control Panel. If a game does not list it, it has not implemented it, and no amount of driver configuration will add it. The driver’s own Low Latency Mode is the older, weaker cousin, and it only applies to DirectX 9 and DirectX 11 titles.
2. Understand what the render queue actually is
Without Reflex, the CPU races ahead of the GPU and stacks up frames waiting to be rendered. Each queued frame is input that was sampled earlier and will be displayed later, which is latency you feel but cannot see on a frame counter. Reflex changes the timing so the CPU submits work just in time, keeping that queue near empty. The effect grows with how GPU-bound you are, which is why players on lower-end cards at high settings often notice the biggest change.
3. Decide between On and On + Boost by testing
Boost keeps the GPU from dropping clocks during CPU-bound stretches, since a downclocked GPU takes longer to render each frame even when it has spare capacity. That helps in titles where the CPU is the bottleneck, typically large-scale multiplayer games and older engines at low settings. In GPU-bound scenarios Boost does close to nothing and simply raises power draw and fan noise. The only reliable way to choose is to run both in your own game on your own hardware.
4. Pair it with a frame cap and variable refresh
Reflex is not a substitute for the rest of a low-latency setup. On a G-SYNC display the standard arrangement is variable refresh enabled, V-Sync on at the driver level, and a frame cap a few frames below your refresh rate so frames stay inside the VRR window. Reflex applies its own internal cap logic in some implementations, but an explicit cap remains the cleanest way to avoid the latency penalty that comes from running against the top of the refresh range.
5. Measure instead of guessing
NVIDIA’s overlay includes a performance monitoring view that can report latency metrics on supported systems, and the newer NVIDIA app exposes the same statistics that GeForce Experience used to. Which metrics you see depends on your driver version, the game’s Reflex integration, and whether your peripherals support the hardware Latency Analyzer path. The full end-to-end measurement, including mouse click to photon on screen, requires a G-SYNC monitor with the Analyzer module and a mouse plugged into that monitor.
6. Leave it enabled when using frame generation
DLSS Frame Generation increases the number of frames displayed but does so by presenting generated frames, which inherently adds latency. NVIDIA ties Reflex to frame generation precisely to offset that cost, so Reflex is enabled as part of the package. If you are running frame generation and wondering whether to disable Reflex to test something, the answer is no.
Latency Versus Frame Rate, Precisely
These two numbers are related but not interchangeable, and conflating them is the single biggest source of confusion around Reflex. Frame rate describes how often a new image is produced. Latency describes how old the information in that image is by the time you see it. You can have a high frame rate with poor latency if frames are queued several deep, and you can have a modest frame rate with excellent latency if each frame is rendered and displayed with minimal buffering.
This is why a player can switch Reflex on, see an unchanged or fractionally lower average frame rate, and still find that tracking targets feels noticeably more direct. The mouse cursor is responding to more recent input. Nothing about the frame counter captures that, which is why NVIDIA reports Reflex results in milliseconds of system latency rather than in frames.
It also explains why Reflex sometimes costs a little peak frame rate. Keeping the queue empty means the GPU occasionally waits rather than always having work pre-staged. In a benchmark that averages frame rate over a run, that shows up as a small deficit. In actual play, where what matters is how quickly the game reacts to you, it is a clear win. NVIDIA has also been developing further latency work under the Reflex 2 banner, including techniques that adjust a rendered frame to account for the most recent mouse input, though availability depends on the title and driver, so check what your games actually support rather than assuming.
Related guides: best nvidia control panel settings for gaming, digital vibrance nvidia how to adjust, how to fix nvidia overlay not working, how to open the nvidia overlay shortcut, how to sign up for an nvidia account
FAQ
Does NVIDIA Reflex increase FPS?
No. Reflex reduces system latency. In some GPU-bound situations it slightly lowers peak frame rate as a side effect of keeping the render queue empty. If your goal is higher frame rate, lower your in-game settings or use an upscaler instead.
Should I use On or On + Boost?
Start with On. Boost helps mainly in CPU-bound scenarios by preventing the GPU from downclocking, and it costs extra power and heat. On laptops or in thermally constrained builds the trade is often not worth it. Test both in the game you play most.
Why is Reflex greyed out or missing in my game?
Either the game has no Reflex integration, or your GPU or driver does not support it. NVIDIA lists support for the low latency mode going back several GeForce generations, but the game-side implementation is what determines whether the option appears at all.
Is Reflex the same as the driver’s Low Latency Mode?
No. They pursue the same goal through different routes. The driver setting works only in DirectX 9 and 11 titles and operates without game cooperation. Reflex is integrated by the developer, works in modern DirectX 12 and Vulkan titles, and is generally more effective.
Do I need a special monitor for Reflex?
Not for the latency reduction itself, which works on any supported display. You only need specific hardware for the Reflex Latency Analyzer, which measures full end-to-end latency and requires a G-SYNC monitor with that module plus a mouse connected through it.
Write Your Review
No reviews yet. Be the first to share your experience!