KDE On Wayland: "The Biggest Thing Needed Now Is Adoption By 3rd Party Apps"

[oiaohm]

Meanwhile, in the real world
https://www.youtube.com/watch?v=5eKSQT5mV-c

Someone need to learn to be way more careful on what they refer to. Look up what NOBARA installs hello it does not install X11 bare metal its a wayland only solution and it the recommend one in that video. Yes and that a Nvidia GPU laptop and the reviewer see no major issues with a full wayland desktop install.


The reality here is this is happening because Wayland on Nvidia is count of 4 overhead X11 without compositor 2 overhead and X11 with compositor is a count of 6 overhead with Nvidia. Thinking most people operate with a compositor loaded for majority wayland even on Nvidia is a improvement in performance.

Wayland = 10% performance hit on average vs xorg-server < This is not correct caused by over simplification. Xorg-server without X11 compositor 10% performance hit using wayland. xorg-server with compositor Wayland has a 10% performance improvement. On solutions like AMD and Intel with functional drivers still a 10% improvement using Wayland over using X11 with compositor but but the Xorg-server baremetal vs wayland is no different.

Lot of gaming benchmarks with linux are not really measuring the results if you are not full screen so not bypassing X11 compositor.

If you are going to be browsing the internet doing word processing and so on guess what compositor is going to be on so compositor on performance comes important.

[mSparks]

Someone need to learn to be way more careful on what they refer to. Look up what NOBARA installs hello it does not install X11 bare metal its a wayland only solution and it the recommend one in that video. Yes and that a Nvidia GPU laptop and the reviewer see no major issues with a full wayland desktop install.

Nothing funnier than someone who clearly didnt watch the video making claims of its content and taking the high ground.

Do you want to know how he found that out without watching the video?

Nobara failed catastrophically worse in every way compared to the FC he setup himself. "and then he realised Nobara was using wayland and redid the tests with that garbage uninstalled".

[oiaohm]

Nobara failed catastrophically worse in every way the FC he setup himself. "and then he realised Nobara was using wayland and redid the tests with that garbage uninstalled".

Also the system was not in performance mode. So he changed 2 things. That performance hit comes from the Nvidia drivers and system being in the wrong power mode. If you forced X11 into modesetting mode wrong power mode and run the same tests guess what going to happen. So the performance hit was not 10% for wayland alone.

Also notice that he did not notice there was a problem until he saw the numbers.

[mSparks]

Also the system was not in performance mode. So he changed 2 things. That performance hit comes from the Nvidia drivers and system being in the wrong power mode. If you forced X11 into modesetting mode wrong power mode and run the same tests guess what going to happen. So the performance hit was not 10% for wayland alone.

Also notice that he did not notice there was a problem until he saw the numbers.

it was a lot more than 10%.
because Nobara went from much worse than his setup when it was on wayland to 10% faster after switching it to X11.

I was just being generous.

As for performance mode, Ive not seen that make much if any difference on recent builds, even powersave will dial everything to max when the nvidia driver kicks in to high gear.

But at least we get a nice meme out of it, to let people in on the secret plan of the
Wayland gnomes
Step 1: Fail
Step 2: Undefined behavior
Step 3: Profit!

(southpark underpant gnomes reference for those who dont know it)

[oiaohm]

As for performance mode, Ive not seen that make much if any difference on recent builds, even powersave will dial everything to max when the nvidia driver kicks in to high gear.

No the laptop he is using not being in performance mode in fact lowers the total watts of power to the CPU and GPU. Yes absolutely extend battery life absolutely hurts performance. Not all devices alter power levels to the CPU/GPU. He has a older video with the same laptop with fedora where he shows a 7% between performance mode and the power saving mode in that laptop.

Yes it depends on your device how big of difference powersave and performance mode does because different vendors different things happen when you change those modes.

[mSparks]

No the laptop he is using not being in performance mode in fact lowers the total watts of power to the CPU and GPU.

I know it does, but only in an idle or moderate use state.
The only difference that comes from performance mode is it keeps everything running at full tilt always except when nearly idle.
performance mode can be useful if you have tasks you want to complete quickly when the machine isnt under much load, but otherwise they all behave identically under heavy load (such as he was testing).
Its not like windows where powersave cripples the machine to something from the naughties.

[MrCooper]

xfree86 is not DDX, its not even xorg-server.... And not one of them is actually an issue.

You shouldn't use terms you clearly don't understand...

The Xorg DDX code is still under hw/xfree86/ (where it was created by XFree86) in the xserver tree, that's why the Xorg DDX label is called "xfree86".

Xorg-Server was released in 2004, been on that since it's initial release.

X was created in the mid-80s, it was half a decade older at that point than Wayland is now.

My first linux distro I used in a meaningfull way was mandrake linux in 1998, so I guess that was probably running XFree86

Hey, that's pretty close to when I first started using Linux. So you know that X wasn't very mature at that point.

[MrCooper]

[...] low latency and independant compositor.

You do realize the latter hurts the former? This is one reason why Wayland can achieve lower latency than X with compositing, even for X apps via Xwayland compared to Xorg.

[mSparks]

You shouldn't use terms you clearly don't understand...

The Xorg DDX code is still under hw/xfree86/ (where it was created by XFree86) in the xserver tree, that's why the Xorg DDX label is called "xfree86".



X was created in the mid-80s, it was half a decade older at that point than Wayland is now.



Hey, that's pretty close to when I first started using Linux. So you know that X wasn't very mature at that point.

That's why I listed the implementations I'd used, the design of X actually dates back to the 60s aiui, based on the mainframes they were building and using for the space program, I don't know the history of when got it's official "X" title, and I wasn't going to believe wikipedia, certainly existed well before it's initial public naming in one form or other.

afaik, DDX is the original "final" graphics API used by very early unix systems, what exists now is a very "lite" DDX driver in the form of "xf86-video-modesetting" just keeps that old stuff working, while handing over the actual rendering to Direct Rendering Manager/DRM - used by both wayland and direct rendering xorg-servers..

You do realize the latter hurts the former? This is one reason why Wayland can achieve lower latency than X with compositing, even for X apps via Xwayland compared to Xorg.

except it seems it can't and it doesn't. I don't know why because in principle I completely agree with the theory, but in practice my stuff was taking nanoseconds from being sent to /dev/input/* to its instrumentation, whereas anything on wayland was milliseconds.

If I was having to guess based on my rough understanding of what they are doing, I would say because when using X11, /dev/input/* is the raw data stream, but on wayland it has to run through a process and significant additional logic first.

Then there's other peoples tests

https://zamundaaa.github.io/wayland/2021/12/14/about-gaming-on-wayland.html

So when you say "You do realize the latter hurts the former?" and that "Wayland can achieve lower latency than X with compositing, even for X apps via Xwayland compared to Xorg."

I'm going to have to say, no, I don't realize that. based on what actual evidence and testing? And how much attention has been given to the significant user feedback contradicting it.

[MrCooper]

What you talking to me for, [...]

Excellent question! I thought I could help you understand some basic concepts of X & Wayland. It's clearly hopeless though, so time to bow out for real.