Xorg is dead. Long live XLibre

[Brian MasinickModerator]

Thanks for providing these details; though this technology has promise and upside you are showing that it really is not ready for widespread deployment yet.

You do think that it’s promising; it’s been working well for me in my fairly simple every day environment.

@anticapitalista is absolutely correct to proceed gradually and with appropriate care and testing. We need much more testing and evaluating. I do hope that we continue building and testing it in the meantime.

--
Brian Masinick
Alternate "Search B":
This search page

[Xunzi_23Member]

Larrys4227 wrote
Duck-Duck had some really bad suggestions for me to try. LOL

Been there. below added to LibreWolf and/or UngoogledChromium helps.helps.
https://noai.duckduckgo.com/search?q=%s

Some other AI assisted search engines now also offer unlimited trash results.

Regarding not much feedback on the reported memory usage.

What is the GPU and for completeness CPU temperature when 100% memory usage
is reported.

Is the cooling fan screaming away at max while venting hot air, when working on
freshly setup laptop and desktop devices that is one clear indication something is
wrong.

Either will reliably tell you if the device really is working at maximum GPU load.

If available A wattmeter and comparison of power draw would also give reliable information
on the real state of affairs.

If GPU temp stays in a non critical range the software tools are doing as Wallon
described, reporting reserved memory and XLibre is not causing a problem.

I gave no feedback because my Thinkpad does not have an unusually high GPU temp
while running antiX 26 with the XLibre Packages provided by anticapitalista and
ZRAM setup.

[olsztynMember]

anti-apXos wrote:

Maybe others are already familiar with the benchmark test you used, but I’m not, so I wonder what the usual range of variation from one test to another is when you keep the same configuration. Without that info, it’s impossible to understand whether any difference in the numbers is actually significant when you change the configuration.

Just as I mentioned before, this benchmarking, using Hardinfo2 one of options of Hardinfo2 tool, is very rudimentary, so it is not full blown benchmarking. This tool is the hardware specs reporting tool, so benchmarking is just one of many functions and possibly does not measure up to the specialized benchmarking out there.
The version I used was the one you get in antiX with the standard set of repositories. I did not use the later version, which is available in MX.

So the comment of @anti-apXos is quite valid one, namely how much can you depend on this tool. Not much indeed, this is just a ballpark measurement, having run a few times on each piece of hardware and taking some middle readings, just to get some sense of numbers, which may or may not support a subjective sense of faster or slower experience.

My experiment and all testing of XLibre on multiple hardware is mostly to determine whether it seems a viable option for the things to come, whether there is any competition to Wayland that can be suitable for antiX and other distros or are we doomed to adopt Wayland no matter what and face the end of the road for the old WMs.
Although I am not depending on the old WMs and have Waylaad as one of selections on my login manager, I think Wayland is not ready yet for the majority of of Linux users and very slow cooking, so it may take still a very long time before it is well done.
XLibre may prove it is the real competition but it needs testing on real hardware, not in Virtual Box, to have any meaning to end users. If it does not prove such, then we are facing the end of the road with old WMs and better start testing options with Wayland…

• This reply was modified 4 months, 4 weeks ago by olsztyn.
• This reply was modified 4 months, 4 weeks ago by olsztyn.

Live antiX Boot Options (Previously posted by Xecure):
http://antixlinuxfan.miraheze.org/wiki/Table_of_antiX_Boot_Parameters

[olsztynMember]

Brian Masinick wrote:

Thanks for providing these details; though this technology has promise and upside you are showing that it really is not ready for widespread deployment yet.

U do think that it’s promising; it’s been working well for me in my fairly simple every day environment.

Yes, indeed. It is not for widespread use yet, as it needs more testing on real hardware. The arguments that it has issues with defaults on virtual Box or QEMU do not convince me though that there is a problem implementing this on real hardware, which real hardware counts in the end user experience, not Virtual Box, when it comes to real users. Negative experience with using such critical piece of infrastructure in Virtual Box means little to me. Just real hardware counts.
I have been testing on many laptops, not all yet of those I have for testing and if I find one where it does not work, this is what counts for me, not whether it works well in Virtual Box…

Live antiX Boot Options (Previously posted by Xecure):
http://antixlinuxfan.miraheze.org/wiki/Table_of_antiX_Boot_Parameters

[WallonMember]

Bonjour,

I noticed that the « TearFree » option (anti-tearing) is already active on my antiX 26 system without any manual configuration file. Here is how to check:

sudo grep -i tearfree /var/log/Xorg.0.log

Result:

(II) modeset(0): TearFree: enabled

• (II) = information (not an error)
• modeset(0) = graphics driver in use
• enabled = anti-tearing synchronisation is active

There is no « .conf » file in « /etc/X11/xorg.conf.d/ ». The default configuration of XLibre (the X server used by antiX) therefore enables « TearFree » for the « modesetting » driver.

Is this option also enabled by default on your computer?

Best,
Wallon

• This reply was modified 4 months, 4 weeks ago by Wallon.

[olsztynMember]

Wallon wrote:

I noticed that the « TearFree » option (anti-tearing) is already active on my antiX 26 system without any manual configuration file. Here is how to check:

sudo grep -i tearfree /var/log/Xorg.0.log

I am not getting any output running this…

Live antiX Boot Options (Previously posted by Xecure):
http://antixlinuxfan.miraheze.org/wiki/Table_of_antiX_Boot_Parameters

[WallonMember]

olsztyn wrote:

So far I have been testing XLibre on my various laptops, to make sure all works well. So far it appears to have better support for old laptops, where, as reported before, it is a significantly better performer on my Thinkpad X61 Core2Duo from 2007-2008.
I just have run some benchmark on my midrange laptop – Apple Macbook Pro 15 from 2013, comparing antiX 26 instance running Xorg against XLibre, otherwise identical images.
For benmarking of graphics performance I was using antiX – native Hardinfo2, running GPU Drawing on both Xorg and XLibre antiX.
The results exceeded my expectations:

Xorg – 1125
XLibre -1313

This is a significant difference in favor of XLibre…

On my oldest laptop though Thinkpad X61 Core2Duo from 2007-2008 the benchmark result was not impressive though:
Xorg – 465
XLibre – 489

This small increase in performance is surprising, as running Google Earth the difference in performance was significant, where XLibre made it actually usable.

Thank you @olsztyn for sharing your test results. You must have spent a lot of time and carried out several installations to obtain the figures for different configurations.
This is really very encouraging for XLibre.

[WallonMember]

Bonjour,

According to my readings, XLibre should enable this option by default, which is why I ask to check in the Xorg log if it is indeed the case. If it is not enabled, you can perhaps add and test the following files according to your Intel GPU or AMD APU.

If the « TearFree » option is not enabled by default on your system, or if you experience screen tearing or flickering, you can create a custom configuration file to force it manually.

Here are the most common configurations:

1. For MOST modern Intel GPUs (using the « modesetting » driver)

This is the recommended driver for all recent Intel hardware. It is the default on antiX 26. Even though the legacy Intel driver (« xserver-xorg-video-intel ») is available in the repositories, it is no longer recommended for recent GPUs (including my i5-8350U) because « modesetting » is more stable and performs better.

Create the file:

sudo mkdir -p /etc/X11/xorg.conf.d/

sudo nano /etc/X11/xorg.conf.d/20-modesetting.conf

Add this content:

Section "Device"
       Identifier "Intel Graphics"
       Driver "modesetting"
       Option "TearFree" "true"
EndSection

2. For older Intel GPUs (using the legacy « intel » driver)

If you have very old Intel hardware (pre-2012), you may try the dedicated driver. Install it first if needed:

sudo apt install xserver-xorg-video-intel

Then create the file:

sudo nano /etc/X11/xorg.conf.d/20-intel.conf

Add this content:

Section "Device"
       Identifier "Intel Graphics"
       Driver "intel"
       Option "TearFree" "true"
EndSection

3. For AMD GPUs (using the « amdgpu » driver)

The required package « xserver-xorg-video-amdgpu » is available and often preinstalled on antiX 26.

Create the file:

sudo nano /etc/X11/xorg.conf.d/20-amdgpu.conf

Add this content:

Section "Device"
       Identifier "AMD Graphics"
       Driver "amdgpu"
       Option "TearFree" "true"
EndSection

After creating the file, save your changes and reboot (or restart Xorg) for the setting to take effect.

I remind you of my first post in this discussion thread: on Xorg, these different configuration files did not work at all and made antiX 26 unusable on my Dell Latitude 5590 laptop with i5-8350U CPU. That is why I installed XLibre.

Best regards,
Wallon

• This reply was modified 4 months, 4 weeks ago by Wallon.

[olsztynMember]

anticapitalista wrote:

New XLibre debs to hit antiX repos soon

Thanks again…
I suppose these are updates from two days ago. All upgraded fine except xserver-xlibre-video-amdgpu. It appears that it missed upload. It does list in synaptic as upgrade but it appears to not have been uploaded.
The older version runs fine though…

Live antiX Boot Options (Previously posted by Xecure):
http://antixlinuxfan.miraheze.org/wiki/Table_of_antiX_Boot_Parameters

[anticapitalistaForum Admin]

olsztyn wrote:

anticapitalista wrote:

New XLibre debs to hit antiX repos soon

Thanks again…
I suppose these are updates from two days ago. All upgraded fine except xserver-xlibre-video-amdgpu. It appears that it missed upload. It does list in synaptic as upgrade but it appears to not have been uploaded.
The older version runs fine though…

There hasn’t been any upgrade to xserver-xlibre-video-amdgpu.
The latest we have is the version here

https://repo.antixlinux.com/trixie/pool/main/x/xserver-xlibre-video-amdgpu/

Philosophers have interpreted the world in many ways; the point is to change it.

antiX with runit - leaner and meaner.

[c0pAMember]

anticapitalista wrote:

In that case you could

sudo apt purge xserver-xlibre-video-amdgpu

Today I finally tried this. Here’s the output:

$ sudo apt purge xserver-xlibre-video-amdgpu
REMOVING:                       
  xserver-xlibre-video-all*  xserver-xlibre-video-amdgpu*

Summary:
  Upgrading: 0, Installing: 0, Removing: 2, Not Upgrading: 0
  Freed space: 271 kB

Continue? [Y/n]

Is this safe to do?!

[anticapitalistaForum Admin]

Should be. But, make sure you know how reinstall it if it causes issues (chroot or from a live iso)

Philosophers have interpreted the world in many ways; the point is to change it.

antiX with runit - leaner and meaner.

[c0pAMember]

I carried on with that instruction and all is well at present.

I appreciate your input. Cheers!

[wildstar84Member]

re “Is this (TearFree) option also enabled by default on your computer?

No, but I’m using xserver-xlibre-video-intel (older Ironlake system that it works better on). I’ve never had pbms. w/tearing, and I read somewhere that XLibre is much better at avoiding tearing, particularly on older systems that had trouble with it, so, w/XLibre, you may want to try turning it off for performance. Perhaps XLibre may be turning the default on or off based on what hardware/driver you’re using?

[WallonMember]

Summary of results: comparison between Xorg and XLibre on antiX 26 Full (SysVinit)

Tests performed:
– glxinfo (OpenGL)
– glmark2 (X11 backend)
– x11perf (Copy 500×500)
– application launch latency (urxvt, 10 launches)

Hardware: Intel UHD Graphics 620
System: antiX 26 Full SysVinit

I also share the script used for these tests.
To run the full benchmark, two packages must be installed:
– glmark2-x11
– time

1. OpenGL (glxinfo)
Both servers use exactly the same rendering pipeline:
– Mesa Intel UHD 620
– OpenGL 4.6
– Mesa 25.0.7
=> No difference in driver or OpenGL pipeline.

2. glmark2 (Test “build”, 10 seconds)
– Xorg : 3417
– XLibre : 3891
=> Higher score under XLibre.

3. x11perf (Copy 500×500)
– Xorg : ~6500 copies/sec
– XLibre : ~6800 to 7500 copies/sec
=> XLibre is faster for X11 Core operations.

4. Launch latency (urxvt, 10 launches)
– Xorg : 0.72 s
– XLibre : 0.69 s
=> XLibre is slightly faster when creating an X11 window.

Conclusion
On this hardware and with antiX 26 Full SysVinit:
– OpenGL: identical
– glmark2: higher score under XLibre
– x11perf: XLibre is faster
– latency: XLibre is slightly faster

XLibre shows better overall responsiveness in X11 operations and window creation, while keeping the same OpenGL rendering as Xorg.

Attachments:

1. bench_xorg.sh_.zip

[Author]

Posts