Best FPS Cap for Low Latency (and Whether to Cap at All)

On this page

Yes — you should almost always cap your FPS for the lowest input lag, and the best value is a few frames below your monitor’s refresh rate (for example 138 on a 144 Hz panel). Capping keeps a G-Sync or FreeSync display inside its variable-refresh window and stops your GPU from saturating at 100% load, which is where the render queue fills and latency spikes. The one caveat: prefer an in-engine or NVIDIA Reflex cap over an external tool like RTSS, because it limits frames earlier in the pipeline.

Best FPS Cap for Low Latency (and Whether to Cap at All)

An uncapped frame rate isn’t faster input — a saturated GPU queues frames and your mouse waits in line.

Why a cap lowers latency at all

It feels backwards that limiting frames could make a game feel more responsive, but the mechanism is straightforward. When your GPU runs flat-out at 100% utilization, it can finish frames faster than the rest of the pipeline can use them, so finished frames pile up in a render queue. Your input gets sampled, then waits behind those queued frames before it ever reaches the screen.

A cap that keeps the GPU just below full load empties that queue. Frames are rendered closer to “just in time,” so the gap between your click and the photons that show it shrinks. This is the same problem NVIDIA Reflex and AMD Anti-Lag attack from the driver side — a cap is the manual version of the same idea.

What value to cap at

The cap interacts directly with your monitor’s refresh rate and whether you run variable refresh (VRR):

SetupRecommended capWhy
G-Sync / FreeSync (VRR)3–5 FPS below refreshStays in the VRR window, avoids V-Sync ceiling latency
144 Hz with VRR138–141Just under refresh keeps sync engaged
240 Hz with VRR234–237Same rule, scaled up
Fixed refresh, no VRRA few below refresh, or uncapped competitiveAvoids GPU saturation; tearing remains

The widely used guidance for VRR is to cap a few frames under your refresh rate so the display never hits the top of its sync range, where V-Sync would otherwise kick in and add a frame or more of delay. The precise number is not magic — anything 3 to 5 below refresh works.

In-engine vs Reflex vs RTSS

Not all caps are equal, because where in the pipeline the frame is held determines how much latency it adds:

  • NVIDIA Reflex cap (best): When you enable Reflex in a supported game, it sets its own optimal cap and manages the render queue directly. This is the lowest-latency option and you should prefer it whenever it exists.
  • In-engine limiter: A frame cap built into the game’s own settings limits early, before the render queue, so it is typically lower latency than an external overlay.
  • RTSS (RivaTuner): RTSS has excellent, very consistent frame pacing but limits later in the chain, adding roughly one frame of delay versus an in-engine cap. Great for smoothness, marginally worse for raw latency.

For pure competitive responsiveness: Reflex if available, otherwise the in-engine cap. For the smoothest frame pacing in a single-player game, RTSS is a fine trade.

When uncapped actually wins

There is a narrow case where no cap is the competitive choice: a fixed-refresh setup with V-Sync fully off, where a player wants the absolute highest raw frame rate and accepts screen tearing. Higher real FPS does shorten the input-to-photon gap, so in titles like CS2 or Valorant some pros run uncapped on purpose.

But this only helps if your GPU is not the bottleneck — if you are GPU-bound, uncapped just means a saturated GPU and a full queue, which is worse. For the vast majority of players on a VRR monitor, a cap just below refresh is both smoother and lower latency.

The settings that matter more than the cap

A cap is one link in the chain. The bigger system-level wins are getting your Windows timer resolution right with Tier1Timer for consistent frame delivery, enabling NVIDIA Reflex or AMD Anti-Lag, and tuning your display pipeline. Set the cap, then verify the whole stack by measuring your input lag before and after.

Cap your FPS a few frames below your refresh rate, let NVIDIA Reflex or an in-engine limiter do the capping rather than an external overlay, and you get a smoother picture and lower input lag at the same time — the rare tweak with no real downside.

Frequently asked questions

Should I cap my FPS for the lowest input lag?

In most cases, yes. Capping a few frames below your monitor's refresh rate keeps a variable-refresh display (G-Sync or FreeSync) inside its sync window and stops the GPU from running at 100% load, which is where the render queue fills and latency balloons. The exception is an uncapped competitive setup with V-Sync fully off, where some players prefer maximum raw frame rate at the cost of tearing.

What FPS cap should I use for G-Sync or FreeSync?

Cap roughly 3 to 5 FPS below your monitor's refresh rate — for example, 138 or 141 on a 144 Hz panel, or 234 on a 240 Hz panel. This keeps the variable-refresh range engaged so you avoid both tearing and the latency penalty of V-Sync hitting its ceiling. The exact number matters less than staying just under the refresh rate.

Is an in-engine FPS cap or RTSS better for latency?

An in-engine limiter, and especially the cap that NVIDIA Reflex applies automatically, generally produces lower latency than an external tool like RTSS because it limits frames earlier in the pipeline, before the render queue builds. RTSS uses a very consistent frame-pacing method but adds about one frame of delay. If a game supports Reflex, let Reflex set the cap.

Does capping FPS reduce input lag or add it?

Capping below the point where your GPU saturates reduces input lag, because a maxed-out GPU queues finished frames and delays your input. Capping too low — well under what your system could render — leaves frame rate on the table and slightly raises latency. The goal is a cap just below your refresh rate, not an aggressively low one.

Should I still cap FPS without a variable-refresh monitor?

Yes, capping still helps on a fixed-refresh display by keeping the GPU off its 100% load ceiling, which is the main source of queue latency. You will still see tearing without V-Sync or VRR, but the responsiveness benefit of avoiding GPU saturation is real either way.