One thing that's lovely about Linux is this kind of analysis is not only possible, but meaningful. These results will get reported back to the graphics software authors and the distribution packagers and the ecosystem will improve. There's no sense with Microsoft that kind of improvement is possible.

I recently switched to Linux after years on Windows desktop, mostly because the KDE Plasma desktop feels snappier than Windows 11. Also the feeling that if something isn't working right I can probably tinker and improve it. It's been really nice. If you haven't tried Linux desktops in awhile give Bazzite a whirl: it's a Fedora customized for gaming. Even if you don't game it's an easy way to get a very functional Linux desktop in no time at all.

Awesome article.

I switched my daily driver / gaming rig to Fedora a few months back.

Everything seems snappier compared to Windows, but not sure if it’s in my head, and I’ve been very curious about gaming input latency. This helps answer some questions.

I recently switched to hyprland and I’m very interested how that fits in these results. hyprland uses Wayland so I hope the author might revisit now that hyprland is gaining in popularity.

I’ve considered using gamescope to hopefully get in front of some of these concerns, but I’m on nvidia and there is some discussion about it not working well there.

Now the author's got me thinking about gaming-optimized kernels, which I did not realize was a thing.

I play competitive fighting games so input latency is a huge concern. Would love to hear from anyone else who’s been down this path.

He seems confused at the end why people think wayland is so slow, but don't you think it's because of his xwayland result? People were probably running x11 games on wayland and noticed that significant lag. Just a wild guess. Very nice article, wish people did actual measurements like this more often, of all sorts of things.

This used a 500Hz display which hides a lot of the problems that would show up on slower displays.

The XWayland result is 3ms slower, which at refresh rates this high makes me wonder if it was one frame behind.

Running the tests at 120Hz or even 60Hz might be more interesting because we could start to separate out very small differences in timing from the much larger effects of being a full frame behind.

Addressing all the people here talking about latency of under 10ms, I direct you towards these:

https://www.youtube.com/watch?v=vOvQCPLkPt4

https://www.youtube.com/watch?v=fE-P_7-YiVM

https://news.ycombinator.com/item?id=42832155

There's no such thing as "Wayland input latency". It's just a word salad, akin to "HTTP animation smoothness". The post is measuring Xorg vs. KWin (and also XWayland), other implementations of either X11 or Wayland will have different characteristics.

I wonder where the XWayland's added latency comes from though, it seems suspiciously high to just be easily hand-waved as overhead.

Been dealing with this a bit at Breaka Club. We teach kids to code with a modded version of Overcooked 2!.

We stream OC2[1] with our mod preinstalled over WebRTC. This ensures that kids/schools don't have to try and install the mod. This is particularly important since we support running on school provided hardware. Installing a game without a mod would be hard enough. Added advantage though is kids play with a virtual (on screen) gamepad on iPads in Mobile Safari.

Game instances run in Docker containers in Kubernetes/k3s atop very outdated nVidia hardware. Given we're already going across the Internet into school networks, we've tried very hard to optimize latency across the board. Using NVidia NVEnc with DMABuf (zero copy) etc. We're unfortunately using XWayland at present so experience the documented input overhead. Although our inputs are virtual devices at this point, so the overhead may be a bit different. Trying to optimize this whole thing end to end has been a challenge. I would say that performance is currently "acceptable".

OC2 coding: https://www.youtube.com/watch?v=ITWSL5lTLig (not streamed in this case)

[1] We've bought a limited number of copies of OC2 and pods claim a license on startup. If we're at capacity, kids play something else.

I am not super familiar with Wayland, but basically how composited rendering under X11 works, is if you have a fullscreen window, you can give an 'unredirect' hint to the compositor, in which case when nothing else is drawing your screen, it will stop compositing, and pass your app's swapchain directly to the screen.

This is pretty much optimal, and you can't really do much better than this.

Once a stray window appears on top, or something makes the compositor think it can't do this, it'll do the intermediate step of compositing your app window with others into a temp buffer, and render that.

Sometimes the unredirect breaks for some reason (I remember a case where for some inexplicable reason my app kept creating a window 1px smaller than the screen height), or you use XWayland, you get bad latency.

Since this is a fundamental constraint, other compositors on different OSes must work like this, and you can run into issues like this as well.

Another thing - Wayland afaik started exporing 'display planes' - which are a HW feature of GPUs, that allow it to composite multiple layers together - which means the game can render at full FPS and all the windows on top will be drawn into a different plane and get composited with no ill effects - not sure if this is actually used in production yet.

It looks like consoles and PCs have settled on somewhat different gaming configurations. Consoles usually try to target a fixed output frame rate, while the resolution is often dynamic. On PCs, by contrast, the resolution stays static, while the frame rate and frame-time pacing are dynamic. How does this fit into the latency discussion?

Especially in competitive gaming, I often see people targeting frame rates way beyond their display’s refresh rate. I’m not sure whether this actually provides a real benefit or whether they’re chasing a placebo effect.

Am I out of touch, or is it the children with colored LEDs on their DRAM sticks who are wrong?

I appreciate the article, it's cool to see such small differences across all those settings! It says they learned to solder, if you see solder joins that look like those in the first picture, keep learning! Those are some dry joins, which can easily lead to failure, or intermittant signal loss. Soldering can be a touchy sport!

These results are way better than what Dan Luu measured almost a decade ago:

https://danluu.com/input-lag/

The Rock Band guitars have a photoresistor for precisely this purpose: the screen flashes and the guitar responds when the light hits it. It helps make the otherwise very painful calibration process transparent.

It would be so cool to get that to work in Linux. I know the instrument code is in hid-sony. Here are some open tabs I've got in case anyone's curious:

- https://pascal.giard.info/techreports/nguyen-daniel-autocali...

- https://www.niangames.com/articles/reverse-engineering-rockb...

- https://github.com/torvalds/linux/blob/master/drivers/hid/hi...

The input latency that has been driving me crazy lately is the screen unblank latency to allow input to a program.

E.g. I have an old laptop running a browser playing some internet radio stream. Eventually the screen blanker (without locking) activates.

Some real life event makes me want to hit the space bar to pause music. But the modern screen blank has decided that it should eat/ignore key presses while blank. So hitting the key doesn't pause music. I have to wait for the screen to light up before it will be possible to trigger the pause, and this delay feels interminable!

I seem to recall that in the old days the input remained active to the focused window even if the screen was in a power saving state. This power saving was not conflated with screen-lock security etc. I much prefer that. I think this was because DPMS power saving was an underlying X server behavior, not delegated to a screensaver/lock application?

I'd also be partially satisfied with the async behavior of old terminal programs. My inputs should be buffered and processed even if the effects haven't returned to the screen yet. Then I could at least hit keys twice and be trained to know that one would unblank, the other would pause, and all would be well (eventually).

The current behavior is like having a temporarily numb hand, and being frustrated waiting for sensation to return before I can operate anything!

Very interesting!

Latency numbers are written with three significant digits (4.21 ms). I'm curious about the accuracy of the measurement device. If it can measure tens of microseconds, I'm impressed. If it can't, the conclusions in this article should be taken more coarsely.

Great article! Thank you. Also in case others walked away with the same question I had, I'll save you the googling: use the utility vrrtest to help validate if VRR is properly configured on your machine.

Really hope to see a Windows result. What if Windows has a latency of 10 ms+. xD

SteamOS and Bazzite both use Xwayland through gamescope which is the worst option as it seems https://github.com/ValveSoftware/gamescope

>Avoid XWayland. It added 3.13 ms of latency, more than all other effects combined.

I have a vague memory of (X?)Wayland being much worse than X11 before, and some patch or protocol making it out to all the relevant implementations, but I might have imagined that, since these result show virtually no difference, and only XWayland shows a marginal difference.

Or maybe it just came out of nowhere and was never true.

Using Wayland (hyprland) for daily driver and it's the happiest I've been with an OS in a long time on feel. Feels crisp in that spartan way that windows & macos just don't - no animations, taskbar, popups or god knows what else.

Yes, we know wayland is not only slower but also with much less features.

I wonder though, updated pixels might not have the same latency on the whole screen, and it might even be affected by some updates mechanisms, like panel replay. I.e. it would have been interesting to also measure the screen position as a dimension.

Also, both the input latency (usb controller, and its driver), and screen latency (input latency + processing + update delay) are supposedly also affecting all measurements, but hopefully somewhat consistent or at least filtered out.

A lot of people conflate Wayland being worse than X11 with Gnome on Wayland being worse than Gnome on X11.

Wayland has been great for me for a few years now. I don't use Gnome or nvidia though.

This is awesome. I would like to see tests like this done at 60 Hz as well, and also with non-3D apps. I suspect the results might look different in those conditions. A 500 Hz monitor is not the common case. 2ms is a whole frame!

I saw a very similar post a month or two ago, is this the same author?

edit: no, this is the one I was remembering: https://farnoy.dev/posts/linux-latency

The main issue is with mouse movement rather than clicks tho...

With the exception of XWayland, all of the tests had input latencies within a very small range. No human could tell them apart by those latency differences alone. I would be amazed if someone could notice the 3ms difference jump to XWayland.

The difference could be much larger on a slower monitor. However the differences between Wayland and X11 as protocols is negligible. XWayland as an implementation looks to have a limitation.

As far as I'm aware, you get marginal FPS gains switching kernels. It helps some. It's just a matter of whether the effort is worth it to you. The main change is the scheduler: rather than trying to evenly distribute CPU time to processes, it will prioritize bursty processes.

I've been a fan of Hyprland for gaming so far. Much more configurable for things like VRR/tearing and other precise tweaks via Gamescope than when I was on AwesomeWM with X11. Been especially nice having Lua for configuration, which finally feels very familiar with my AwesomeWM roots.

I moved to OpenSuse and have the same experience/opinion. The only caveat I had in Wayland is Game Streaming. Sunshine/Moonlight work but the input lag is noticeable and there are artifacts in the game. I go back and forth between X11 and Wayland just because X11 better for game streaming but in time I'm sure I'll go full-time into Wayland.

I've been using Linux since the mid 1990's. I'm no newbie to any of this. I literally can't tell the different between X11 and Wayland when using either of them and I don't care about all the arguing. This is just Vim vs Emacs and Gnome vs KDE all over again. At this point when I see people complaining about it I just click off the page. It's all stupid and pointless.

> and noticed that significant lag

Only xwayland showed that result. The difference was only a couple milliseconds. That’s in the range where I start to doubt that people are feeling the latency difference. If it was 10-20ms I could believe it, but not when it’s a couple milliseconds.

The author of this post did a good job of getting all of the other confounding settings out of the way. It’s possible that the people complaining that Wayland was slow were starting from an unoptimized situation and as part of switching to some low latency variant they set all the correct settings.

You can't just test one wayland compositor and talk about the performance of all wayland compositors. They're vastly different, especially when it comes to the extensions to wayland needed to handle input devices (ref: https://wayland.app/protocols/). It's not like how xorg is the standard strong reference implementation for X11 everywhere that works the same everywhere.

What's probably happening is that other wayland compositors are slower than KDE Plasma wayland which he tested. And people report that experience. Some other wayland compositors might even be faster than plasma. But what is for sure is that every wayland is very different from every other wayland.

Isn't Wayland always one frame delayed compared to Xorg to avoid tearing or has that been changed? If so, his very high refresh rate would minimize that effect

I have found that input latency is a LOT more acceptable with a controller than it is with a mouse.

Playing a cinematic game with a controller (especially with auto-aim) at 30FPS with vsync is fine. Playing a first person shooter with a mouse, or a game where you control your camera with a mouse, at 30FPS with vsync feels very bad.

That's my theory on why the priorities are different, at least.

As someone who is in the rendering space for work. Having a higher framerate does help, but in a weird way. Basically the start of the frame rendering is what mostly dictates where objects are rendered. By getting a higher framerate the position of objects that you see in game are much closer to their "real" position. So it's less about seeing more frames at that point and more about seeing the most up to date information possible. Technically it could be possible to render the frame in sync with the framerate and just offset the rendering so it finishes right before it's pushed to the screen, but if you're slightly wrong you'll get really bad stuttering and the execution time of gpus and the cpu submitting the work isn't really deterministic.

Many PC games have dynamic resolution, too. The reason consoles target 60fps is because that’s the frame rate you get with most TVs and everyone playing the game has the same hardware (or couple variants).

> Especially in competitive gaming, I often see people targeting frame rates way beyond their display’s refresh rate. I’m not sure whether this actually provides a real benefit or whether they’re chasing a placebo effect.

A newly rendered frame can cut-in during scan out. This shows up as tearing artifacts where the frame is changed while being sent to the display, but it allows fresher pixels to hit the screen below that tearing line. So each frame on the monitor can be a mix of multiple rendered frames.

It’s not as good as having variable refresh rate display with high refresh rate, but it does reduce latency.

For less action based games it’s common to turn vsync on and pace the frames to the refresh rate to eliminate this tearing.

> I often see people targeting frame rates way beyond their display’s refresh rate. I’m not sure whether this actually provides a real benefit or whether they’re chasing a placebo effect.

If they are chasing a placebo effect, it's a really powerful one, since all the actual competitive people are often willing to sacrifice all detail and quite a lot of resolution to get those stupid high frame rates.

I can see the difference too, but the diminishing returns usually make it not worth it, since I prefer the eye candy better details and higher resolutions give me.

Also, some games can adjust the resolutions on the fly to keep a consistent frame rate. It's only become a feature on modern games, but I believe that's mostly a historical accident. PC games could often run on much worse hardware than they were actually designed for (with minimum requirements often being absolute minimums, and not 'this is what we developed for'), so people played them on low frame rates, so that kind of jank was often more culturally accepted on PC, and if you didn't want that experience, you could always upgrade. While on console, there was no upgrade path, and games were optimised for that one config, and thus never allowed to drop too far into the red (and dropping resolution is often a better option in those cases).

Framerates beyond your display's refresh rate are not completely pointless, though a bit wasteful: they do mean that each frame as it is displayed shows a more up-to-date representation of the game state than if your framerate is matched to your refresh rate. In principle you don't need to render the excess frames: ideally your frame time is predictable enough you can kick off the render just before the display refresh, but the penalty is that if you miss the deadline you get some pretty jarring jankiness.

Consoles are more and more supporting VRR, as are televisions.

But yes, given the limitations of the hardware, they often offer two modes - a high framerate but lower quality mode and a high quality but lower framerate mode.

>I often see people targeting frame rates way beyond their display’s refresh rate. I’m not sure whether this actually provides a real benefit or whether they’re chasing a placebo effect.

Pixel refresh is only one part of latency. A higher framerate will lower several other parts of the overall latency. Monitors Unboxed has charts that visualize the amount of latency for each step.

> Especially in competitive gaming, I often see people targeting frame rates way beyond their display’s refresh rate. I’m not sure whether this actually provides a real benefit or whether they’re chasing a placebo effect.

In video games you essentially have one giant loop that runs every frame (today it's more than that, but at its core it's still that). Producing frames faster than the display’s refresh rate can still reduce input latency because the next display refresh is more likely to use a recently generated frame. It does not necessarily mean the game receives more input events, but it can process and reflect those inputs sooner.

Not placebo, but diminishing returns become significant, and the benefit depends on frame queues, VSync, VRR, whether the game is CPU- or GPU-bound, and how its input and simulation loops are designed.

Running above the display refresh rate is only decent when you are wayyyy above it, at least double. If you have a 120hz display and run something variable around 150fps, its much worse. But 500fps does help, you're basically seeing more recent frames as it scans the screen.

> I often see people targeting frame rates way beyond their display’s refresh rate

Could this be to reduce input lag?

"Especially in competitive gaming, I often see people targeting frame rates way beyond their display’s refresh rate. I’m not sure whether this actually provides a real benefit or whether they’re chasing a placebo effect."

Depends on the engine. Anyone remember Quake 3's multiple of 3 frame rate speed hack?