Why the windows graphical interface still slightly faster than all linux DE?

I play chess bullet 1 min x 1 min per player, I feel a drop in graphic performance when drawing the chess pieces when they are dragged too quickly, in windows that doesn't happen. Tested in all web browser and DE.

Thks

Is it? Evidence?

2 members found this post helpful.

Indeed, what evidence do you have to support the claim that their interface is "slightly" faster than all Linux DE's?

What hardware? What test conditions? What Linux DE's have you compared against?

cpu i7 3770k 4.1GHZ
Nvidia GTX 680
8GB Ram Gskill 2133mhz
HDD 7200RPM Black Edition 64mb cache

Fedora 31 + Plasmas (or mate, xfce, gnome)
Nvidia driver 440.31
Display 120hz

Browsers : Firefox 70, Opera, Chrome and Vivaldi

All system updated.

And you're somehow running the same exact program on all of those DEs and browsers, including on Windows? I think you need to provide more information about what exactly you're running to obtain these results.

Have you tried a different distro other than Fedora? Have you considered you might have a graphics problem on Fedora if not?

1 members found this post helpful.

yes exact hardware and softwares versions.
An advanced user in the linux telegram group said X.Org has a rather convoluted rendering pipeline.

Your title is incorrect. Please continue to specify your particular installations and their settings. Or, just don't bother and ask a real question.

Windows has some performance boost from a lot of non-essential items being done in the kernel; font-rendering jumps to mind as well as lots of image rendering.
MS made the choice long ago to sacrifice security for speed.

With that said, you have to provide some benchmark to show this lag between things, but not sure how to do that in a browser. Is your Linux system using X.org or Wayland, that is also a big thing.

This seems to me a deliberate effort to provoke. The thread is probably best left alone.

1 members found this post helpful.

Screen recordings, maybe?

Impossible to record in video.

I'm back to Gnome and I can see now a very good improvement without this problems but the problem still in Plasma Desktop.

dconf write /org/gnome/mutter/experimental-features ['rt-scheduler']

I have certainly never seen what you are describing, but there's something you should know about: CPU frequency scaling.

Most distros ship with their CPU frequency scalings set to use the Ondemand governor, which causes the CPU to be downclocked to half-speed pretty much all the time. Also available, typically, are the Performance governor (which I always switch to immediately), and the Powersave governor (which Linux is usually set up to switch to if your laptop is running off batteries). Windows, of course, would have differently tuned CPU frequency scaling settings by default.

You can test by yourself, ..
1) just install windows in some machine (no virtual), install firefox or another web navigator;
2) install debian, fedora or ubuntu and firefox or another web navigator;
( all systems updated and with the last release of softwares and drivers and tearing and pipeline tricks applied )
3) place the two PC side-by-side;
4) Start the test: open some windows (in the two OS) like file manager, text editor, ... drag the window around the desktop;
5) open www.lichess.com or chess.com create a new chess board and drag the pieces around the board;

Doing the steps 4 and 5 you will feel the graphics performance in windows is smoother and slightly faster.

Obs: the last release of gnome in fedora 31 with the experimental-feature did a very good improvement but the problem still in the Plasma Desktop, Mate and Xfce.
dconf write /org/gnome/mutter/experimental-features ['rt-scheduler']

1. I've already covered this. I told you that I have never perceived the difference that you claim to. So don't tell me to "try it myself" because I already told you I have.

2. I can't help you if I make a technical point and you completely ignore it, dude.

3. Your proposed test has Windows and Linux on two different PCs?!!

4. If the actual data here is "I feel it" and the difference is "slight" and too small to show up on video, then no we cannot try it ourselves. Actually reproducing your results would require starting with your expectations. They obviously have a bigger weight than anything actually technical.

5. The first URL should be org, not com.

1 members found this post helpful.

that may also depend on the [properly installed] video driver, the other [background] processes and a lot of other things [like RAM and swap].
So your question is meaningless as long as you won't describe the exact hardware, the installed software, the actual configuration and the steps to reproduce/measure the phenomenon.