In modern PC gaming, framerates/frametimes and latency are more than just numbers — they shape how smooth your game feels and how quickly it responds to your inputs. While raw average FPS often gets all the attention, experienced gamers and performance enthusiasts know that consistent frame pacing and low system latency can make a bigger difference to smoothness and responsiveness than a random FPS figure alone.

That’s where framerate limiters come into play: complex algorithms that cap the number of frames your system renders per second, while balancing many trade-offs. At first glance, limiting FPS might seem counterintuitive: why hold back performance? Unfortunately, uncapped framerates often push GPUs and CPUs harder than necessary, creating excessive heat/fan noise, inconsistent frametimes, and sometimes annoying screen tearing when the GPU produces far more frames than the display can show. Capping FPS intelligently can ease this strain, leading to steadier performance, smoother frame delivery, cleaner presentation, and even reduced latency in many scenarios.

Framerate limiting isn’t a one-size-fits-all setting. It exists in various forms: built into games, embedded in graphics drivers, or implemented via external third-party utilities. Each method has different implications for performance stability, frame time consistency, and system latency, AKA the time between when your input is registered and when the result appears on screen. Understanding these differences is critical for anyone who wants to optimize gameplay beyond simple FPS numbers.

In this article, we will take a deep dive into framerate limiters using CapFrameX , a modern benchmarking tool that measures a plethora of performance metrics, though in this article, we shall only focus on four of those metrics: average FPS to gauge raw performance, 1%/0.1% low average FPS and adaptive standard deviation to gauge frametime consistency, and finally average PC latency to measure system latency. We will also be focusing on a single, modern, and well-optimized PC title: Cyberpunk 2077 . We’ll benchmark a range of popular FPS limiter implementations: the vertical sync (AKA VSync) limiter, the native in-game limiter, the NVIDIA graphics driver limiter, the NVIDIA Reflex Low Latency limiter, the various RivaTuner Statistics Server (RTSS) limiters (async, front-edge sync, back-edge sync), and finally the advanced Special K limiter. By standardizing our testing methodology and performance metrics, we aim to reveal how these limiters affect the smoothness and responsiveness of real-world gameplay scenarios.

Whether you’re trying to maximize frame time consistency or minimize system latency, this guide should hopefully help you understand which framerate limiters deliver the best performance trade-offs for your specific use case in a suitable PC game.

When we talk about framerate limiters , we aren’t just talking about tools that cap framerates — we’re talking about different mechanisms that interact with a game's rendering pipeline and your PC hardware in unique ways, each with different trade-offs for smoothness, input latency, and visual integrity. The following is a breakdown of each limiter we’ll be benchmarking in Cyberpunk 2077 .

What it is: VSync (short for vertical synchronization ) is a classic synchronization method that forces a game's frame presentation rate to match your display’s refresh rate . Its primary purpose is to eliminate screen tearing , which is a phenomenon that occurs when a new frame is sent to your display mid-refresh, causing annoying visual screens to appear on your screen. VSync effectively makes your GPU “wait” for the next screen refresh before presenting a new frame to your screen .

How it affects game performance and presentation:

• No screen tearing: Keeps GPU frame presentation and display refresh rates in lockstep.
• Increased input latency: Because the GPU may have to wait for the next screen refresh, it adds delay between frame readiness and presentation, increasing system/input latency, or the more colloquially known term "input lag".
• Stuttering at low FPS: If your framerate drops below the monitor’s refresh rate, VSync can cause uneven frame pacing and stutter because it tries to maintain sync even when the GPU struggles.

When it’s useful: VSync is the simplest and often acceptable in single-player or cinematic contexts where tearing can be quite distracting and input latency isn’t critical. In variable refresh rate (VRR) setups (like AMD FreeSync or NVIDIA G-SYNC ), VSync can act as a backup to prevent screen tearing when framerate exceeds the VRR refresh rate range on your display , though its latency penalty can be greatly mitigated when used with certain FPS limiters — such as NVIDIA Reflex Low Latency , which we shall see later.

Pros/cons :

• Pros: effectively eliminates screen tearing and can provide visually calm, tear-free output; can help reduce minor visual artifacts when frametimes are stable.
• Cons: can introduce noticeable input lag and added latency; can cause stuttering/judder if framerate drops below refresh rate; caps framerate at refresh rate even when higher framerates would be possible, which is especially undesirable in competitive games.

What it is: Many modern games include a native option to cap the framerate directly within the game engine. This cap stops the game from rendering frames above your target, reducing system workload and preventing excessive production of unneeded extra frames .

How it works: Unlike VSync, an in-game engine FPS limiter simply waits after a frame finishes rendering before starting the next one , keeping frame pacing relatively consistent (if the implementation is good enough) without adding the same level of latency that synchronization to the display can introduce.

Pros/cons:

• Pros: Typically lower latency than driver or external caps; directly integrated in game engine logic and is thus more convenient to use (when properly implemented).
• Cons: Quality and accuracy vary by game — built-in limiters can be coarse or inaccurate in some game engines.

What it is: Also known as “ Max Frame Rate ” in the NVIDIA Control Panel or the NVIDIA App, this limiter is part of the NVIDIA GPU driver and applies globally or per-application/game .

How it works: The GPU driver intercepts game engine rendering calls and enforces a ceiling on the framerate . It tends to be low latency and accurate enough for most use cases, though historically some older driver implementations didn’t pace frames as smoothly as third-party tools — an issue that’s largely been addressed in recent NVIDIA graphics driver versions.

Pros/cons:

• Pros: Easy to apply without external tools; often low overhead.
• Cons: Can sometimes produce less consistent frametimes compared to tools like RTSS in some titles.

What it is: NVIDIA Reflex Low Latency is a technology that is designed by NVIDIA that works exclusively on its GPUs, starting from the " Maxwell " GTX 900 Series and up, to significantly reduce system latency/input lag in games, making controls feel more responsive, especially in competitive titles, by synchronizing CPU and GPU work to eliminate render queue buildup. Reflex Low Latency has its own framerate limiting algorithm, which works only when combined with VSync and G-Sync .

How it works: Reflex Low Latency integrates with the game engine (when supported) to complete rendering tasks just-in-time , minimizing the time frames spent queued to be rendered by the GPU before being displayed. It accomplishes this by aligning game engine work to complete just-in-time for rendering, eliminating the GPU render queue, and reducing CPU back pressure in GPU-intensive scenes . This reduces the delay between user input and visible result on a screen, which is key in reducing what NVIDIA calls PC Latency — a metric that combines game latency, GPU rendering latency, in addition to latency that's caused by other miscellaneous operations performed by your hardware during the process of game rendering.

The Reflex Low Latency FPS cap — which is triggered when Reflex is used alongside VSync and G-Sync — has a certain formula that has been determined heuristically by the community, which is the following:

Reflex LL FPS cap = Refresh Rate - (Refresh Rate * Refresh Rate / 4096 OR 3600)

So, for the most common display refresh rates, we obtain the following framerate cap values:

Pros/cons:

• Pros: Significantly lowers system latency/input lag compared to classic caps; designed specifically for maximizing gaming responsiveness.
• Cons: Only available in games that support it; effectiveness varies with game engine and GPU load; can also sometimes cause additional frametime instability, which will hamper a game's visual smoothness.

is a third-party tool that's made by renowned Russian developer Alexey Nicolaychuk , AKA Unwinder , in collaboration with MSI. It has multiple useful functionalities — especially when used alongside (which Unwinder also made in collaboration with MSI) that include the ability to cap game framerates externally to the game engine, not to mention setting up some fancy performance overlays . It offers several framerate-limiting modes that affect how frames are paced :

Async (asynchronous): The default mode buffers one frame to produce extremely flat frametimes, which can feel very smooth visually . However, this buffer can add some extra latency, especially when having VSync enabled alongside it.

Front Edge Sync / Back Edge Sync: These modes sync frame presentation more carefully with the vertical blanking (AKA VBlank) interval at either the front or back edge of the refresh period . They aim to improve timing precision and reduce screen tearing without a heavy latency penalty — effectively sitting between async’s stable frame pacing and a minimalist limiter’s responsiveness.

RTSS + NVIDIA Reflex: Newer versions of RTSS can also use an internal Reflex Low Latency-style FPS limiter, which avoids the classic one-frame buffer and produces lower latency while still capping framerate.

Pros/cons :

• Pros: very accurate and precise framerate caps; generally produces flatter, more consistent frame pacing; highly configurable limiter options.
• Cons: async mode can sometimes introduce higher latency due to internal buffering; multiple modes can be confusing for users; real-world smoothness improvements may vary by game and engine.

What it is: is a powerful utility that's developed and maintained by renowned PC gaming modder Kaldaien . Special K isn’t just an FPS limiter — it’s a comprehensive performance and graphics enhancement framework that allows for many useful game tweaks. Its advanced frame pacing limiter targets smooth frametimes and timing consistency, while powerful graphics tech tweaks (including HDR tooling and NVIDIA DLSS/Microsoft DirectStorage tweaks), not to mention real-time latency/VRR analysis, make it a versatile tool well beyond simple framerate capping.

How it works: Rather than simply delaying between frames, Special K can intercept both before and after the frame's ‘Present’ call. It predicts rendering time and distributes idle CPU/GPU cycles to maintain smoother frame pacing, similar in effect to scanline-sync solutions but integrated deeply with the game engine. Special K also has the highly useful ability to inject Reflex Low Latency markers in certain games that are using certain modern graphics APIs (in order to show PC Latency numbers), and even enable Reflex Low Latency itself in games that don't have official implementations of the latency reduction technology.

Special K actually has four types of framerate limiting modes, which differ in how they handle latency vs frame pacing trade-offs. They are the following:

• Normal mode : This mode is geared towards providing an experience with minimal stuttering.
• Low-Latency : This mode is ideal for displays that support Variable Refresh Rates, and trades some frame pacing stability for even lower latency.
• Latent Sync : This mode is ideal for fixed-refresh-rate displays that don't support VRR.
• NVIDIA Reflex : This mode basically mirrors the behavior of NVIDIA Reflex Low Latency, and is most suited to be used with VRR displays and/or with DLSS Frame Generation/Multi-Frame Generation.

Pros/cons:

• Pros: Can deliver extremely consistent frametimes and very smooth output; has many advanced options for tuning a game's frame presentation model.
• Cons: Requires proper configuration per-game and sometimes troubleshooting — not as plug-and-play as in-game or graphics driver framerate limiters.

All the aforementioned mechanisms aim to put a ceiling on rendering rates, but they interact with the game engine and display differently , which is why their effects on our chosen performance metrics (average FPS, 1%/0.1% low average FPS, adaptive standard deviation and average PC latency) vary — and why benchmarking them with a tool like CapFrameX yields insights that go beyond simple framerate numbers.

To ensure our benchmarking results are meaningful and reproducible, we standardize both our hardware/software setup and the performance metrics we collect. All benchmarks in this article were run on a system with the following relevant specifications:

• CPU: Intel Core i7-14700K;
• RAM: 32 GB DDR5-7000 CL34;
• Storage: 2 TB PCIe 4.0 NVMe SSD;
• GPU : NVIDIA GeForce RTX 4090;
• Operating System: Windows 11 25H2;
• All system firmware, drivers, BIOS, and OS updates were fully applied before testing.

We used CapFrameX’s to capture and analyze raw frametime data during controlled runs of Cyberpunk 2077. For consistency, each test scene follows the same predetermined path while riding a bike through the same area so that rendering load and in-game events are as close to identical as possible. In each case, we enabled CapFrameX’s run history and aggregation feature to record three independent runs of the same scene and aggregate the data — minimizing run-to-run variance and smoothing out transient noise.

Finally, we have decided to run all of our framerate limiter benchmarks while targeting a 120 FPS limit , and with the display refresh rate set to 120 Hz .

Average Frames Per Second (FPS) is the headline performance number most users are familiar with. It represents the total number of rendered frames divided by the capture time, giving a single figure that approximates the overall speed of the game rendering process during the benchmark session. While useful for broad comparisons, average FPS alone can be misleading because it doesn’t account for performance dips or the consistency of frame delivery.

The 1% low average FPS is the average framerate of the slowest 1% of all frames captured. This metric gives insight into the “worst sustained performance,” i.e., how low the framerate tends to dip during demanding moments of the scene. Higher values here (relative to average FPS) typically correlate with smoother overall gameplay and fewer perceptible stutters.

Even more extreme than the 1% low, the 0.1% low average FPS represents the average of the slowest 0.1% of frames. This metric highlights the most severe, albeit rare, performance drops that players are most likely to feel as stutters or hitches. Capturing and reporting this helps ensure we’re not missing tiny but disruptive dips that might occur during heavy simulation or streaming spikes.

Average PC Latency measures the time (in milliseconds) between when a frame starts being rendered and when the completed frame is queued for display, approximating the system’s internal latency . It focuses on how long it takes for your PC to process frames once input is received, excluding peripheral or display latency. This metric is particularly important in latency-sensitive contexts like competitive gaming, where input responsiveness matters as much as framerate.

Adaptive STDDEV is an innovative statistical metric that CapFrameX uses to quantify how much the instantaneous performance values (e.g., frametimes that are converted to FPS) deviate from a moving average over time. Lower Adaptive STDDEV values indicate more consistent frame delivery with fewer sudden swings in frame timing — which usually translates to smoother perceived motion. In layman’s terms, Adaptive STDDEV tells you how “jumpy” the performance feels over time: lower values mean your game runs more evenly, whereas higher values mean more noticeable bumps and irregularities in how frames are delivered. This makes it a valuable complement to percentile lows and averages when comparing FPS limiters’ effectiveness in stabilizing performance.

For each limiter configuration, we run CapFrameX captures over the exact same in-game sequence within Cyberpunk 2077 . Instead of using a synthetic benchmark or a static scene, our choice of a bike-path run ensures repeated dynamic streaming of game assets and NPC interactions, thus better simulating real-world gameplay conditions. All graphics settings, resolution, and in-game options remain identical across these runs and limiter types to ensure fair comparisons.

By using CapFrameX’s run history average across three runs, we reduce anomalies caused by background Windows tasks, momentary hiccups, and other environmental noise. The aggregated data represent a more statistically robust picture of each limiter’s performance profile than a single capture would.

This testing methodology ensures that the results we present are not only precise but also representative of real-world gameplay behavior under varying FPS capping strategies — providing clear insights into how each limiter affects performance in terms of speed , smoothness , and responsiveness .

Another important detail in how we capture and performance analyze Cyberpunk 2077 is that all of our percentile-based FPS metrics — including 1% lows, 0.1% lows, and Adaptive STDDEV — are based on actual display times , enabled via CapFrameX’s msBetweenDisplayChange option. This means the metrics are calculated from the intervals between frames as they actually appear on your monitor , not simply from when the game engine issues a frame to the graphics API. Using display-based timing better reflects what the player actually sees on screen , since it accounts for presentation timing and display refresh synchronization rather than just internal rendering calls. This approach can produce more accurate and perceptually relevant measurements of smoothness and stutter compared to traditional frametime-only methods, especially when evaluating how consistent and fluid a limiter makes the experience feel with the naked eye. This approach also allows for more accurate measurements of smoothness with frame generation technologies enabled, but that is beyond the scope of this article.

In this section, we'll show you the raw, objective benchmark captures of the Cyberpunk 2077 test sequence from CapFrameX’s Analysis tab for each framerate limiter that we tested. Each screenshot will include all of the key performance metrics we’re focusing on — Average FPS, 1%, and 0.1% low average FPS, Average PC Latency, and Adaptive STDDEV — so you can visually compare how different limiters behave in the same scene. We will also analyze these objective capture data to determine the characteristics of each framerate limiter in terms of smoothness, latency, and screen tearing behavior.

In the following captures, you will find that the frametimes — which quantify visual smoothness at the PC hardware level — are displayed in blue , and the display times — which determine visual smoothness at the display level — are displayed in green .

Due to how it works, this limiter simply produces the smoothest display times, though it unfortunately comes at a huge cost to average latency and responsiveness. Therefore, we recommend this limiter for use in single-player games (especially slow-paced ones) where latency isn't a critical concern.

This limiter produced relatively smooth display times, and average latency was also quite low. Unfortunately, it also produced a lot of screen tearing (even witht VRR), though this can be mitigated by setting the cap a bit lower than your monitor's max refresh rate. We recommend this limiter to be used by casual gamers who just want a decent set-and-forget kind of FPS limiter .

• NVIDIA graphics driver limiter :

The NVIDIA graphics driver limiter also produced decently smooth display times, though not as smooth as the in-game limiter . However, average latency was a bit lower. Again, screen tearing was a major concern, but that can be mitigated by setting the cap a bit lower than your screen's max refresh rate. We recommend this limiter t o be used by casual gamers who want to be able to easily set a global framerate limiter for their games without installing any third-party utilities .

• NVIDIA Reflex Low Latency + VSync/G-Sync limiter :

NVIDIA Reflex Low Latency — when combined with VSync and G-Sync — simply produced the lowest average latency reading out of all tested limiters , which makes sense given its underlying mechanism. Unfortunately, this did come at a cost of smoothness (indicated by a high adaptive STDDEV value), though VRR (G-Sync) can greatly mitigate this drawback . Reflex Low Latency is, in our opinion, best combined with both VSync and G-Sync (VRR), and is best used in competitive games where input latency is paramount. It is, by the way, also a requirement for NVIDIA's AI frame interpolation technology (DLSS Frame Generation/Multi-Frame Generation) to work. You may have noticed that this limiter — when combined with VSync/G-Sync — results in a slightly lower than average framerate compared to our 120 FPS target. Reflex Low Latency, when used with VSync/G-Sync, automatically paces frame delivery so that the framerate stays just below the display’s maximum refresh rate — preventing VSync backpressure, keeping G-Sync engaged without tearing, and minimizing latency, which effectively produces a cap of a few frames under the monitor’s peak.

Note : You may be aware that NVIDIA Reflex Low Latency has an additional Boost mode, which c an further lower PC latency on the CPU side . We have also tested that mode, and in this case, we've found that i t offered no meaningful differences in terms of both smoothness and average latency.

The RTSS Async limiter produced very respectable smoothness in terms of display time variability , and average latency was also quite low. This limiter is, in our opinion, best used by enthusiasts in games where both smoothness and latency are equally important (such as single-player shooters). Also, since it has no screen-tearing protection, it's also best to be used at a cap that's slightly lower than your monitor's max refresh rate .

• RTSS Front Edge Sync limiter :

This limiter produced relatively smooth display times, though for some reason it also had quite a few lurches in both frametimes and display times . The average logged latency was quite low as well . As such, we r ecommend that enthusiasts experiment with this limiter in various games and scenes to see if it provides any tangible improvements over the default Async RTSS limiter.

• RTSS Back Edge Sync limiter :

Same behavior and recommendation as the RTSS Front Edge Sync limiter.

• Special K Normal limiter :

Despite a few frametime/display time excursions above and beyond the average, this limiter still produced one of the smoothest display times overall , and is only tangibly beaten by this VSync in this regard. Average latency was also decently low , though not as low as the highest-performing limiters in this aspect. As such, we recommend that this limiter be used in single-player by tech-savvy enthusiasts who are willing to spend extra time to get the most out of their hardware. We also strongly recommend that users do not use this limiter (nor any other Special K limiter) in online games, due to the possibility of anti-cheat software flagging Special K's game process hooking mechanism as a potential cheat .

• Special K Low-Latency limiter :

Unfortunately, this limiter underperformed quite a bit when it comes to display time smoothness, as indicated by it having the highest adaptive STDDEV value out of all tested limiters . Thankfully, average latency was quite low . Therefore, we recommend this limiter to be used by advanced enthusiasts in single-player games that don't support any latency mitigation technologies . Again, we strongly urge people not use Special K's limiters in online games.

• Special K Latent Sync limiter :

This limiter produced the second-lowest adaptive STDDEV value out of all limiters , as it was only beaten by VSync in this regard. Furthermore, it achieved this with a much lower average latency than VSync , though it had quite a bit of screen tearing as well, which would make it a good candidate to be used in single-player games and onfixed refresh rate displays, especially when the cap is set a bit below the max display refresh rate . And lastly, please don't use this limiter in online games due to the threat of anti-cheat software flagging it as a cheat.

As we saw across our standardized Cyberpunk 2077 test runs, each framerate limiting method produced distinct patterns in smoothness and latency behavior. Taken together, it’s clear that no single limiter is perfect — some favor tight, consistent frametimes or display times at the cost of latency, others favor responsiveness at the cost of microvariability in frame/display times. Which limiter is “best” depends on whether you prioritize smoothness , responsiveness , or a balance of both .

Below is a table that summarizes all of our previous findings:

After running Cyberpunk 2077 through CapFrameX with multiple framerate limiting methods and examining not just average FPS but also 1%/0.1% low average FPS, adaptive standard deviation, and average PC latency, it becomes abundantly clear that every limiter carries its own set of trade-offs . Tools like RTSS (especially with its flat frametime pacing) and Special K tend to deliver very consistent frame delivery , reflected in low adaptive standard deviation values and stable frametime graphs — but this sometimes comes at the cost of slightly higher system latency due to internal buffering behaviors. Conversely, simpler caps such as the in-game limiter or the NVIDIA graphics driver's “Max Frame Rate” limiter typically produce lower latency and respectable stability, though they may not flatten frametimes as perfectly as RTSS or Special K. VSync — when used by itself — often adds noticeable latency and stutter/judder if your GPU can't keep up with your perf targets, which makes it less ideal unless combined smartly with an additional framerate cap and/or an adaptive sync/VRR solution.

NVIDIA Reflex Low Latency stands out as a unique option in this lineup: rather than enforcing a strict cap, it dynamically paces frame submission to minimize the render queue and therefore reduce latency, often keeping effective framerates just below your refresh boundary for a balanced experience. Reflex’s dynamic behavior means that in many cases it lowers system latency more effectively than traditional caps , though this unfortunately comes at the cost of frame-time consistency. This still makes Reflex Low Latency a compelling choice for responsiveness-focused players, especially in competitive/e-sports games that support the technology. By contrast, if your priority is absolute smoothness in frametime delivery — such as for cinematic single-player or capture scenarios — external limiters like RTSS (or Special K’s advanced limiter) can yield flatter frame delivery graphs, even if the raw latency cost is a touch higher.

So which limiter should you use? If you’re chasing the lowest possible input latency and your GPU is powerful enough, Reflex Low Latency — or AMD and Intel's equivalent respective techs for their GPUs: Anti-Lag and Xe Low Latency — is typically the best everyday choice. For players who value visual smoothness and frame pacing consistency over split-millisecond latency differences, RTSS or Special K can be worth the added complexity. And for setups without adaptive sync/VRR support, pairing a well-chosen frame cap slightly below your display’s refresh rate with your limiter of choice — whether in-game, driver, or third-party — gives you a solid balance of smoothness, low latency, and minimal screen tearing. In short, the “best” limiter depends on your priorities: low latency for competitive play, or tight consistency for pleasing motion smoothness and stability .

Follow to get more of our news coverage in your feeds.