Mehrere Bildschirme in KDE Debian 12 mit NVIDIA einrichten?


28.06.2024, 00:08

Genutzt wird KDE X11

1 Antwort

Nun werden aber in den Einstellungen der NVIDIA Software keine weiteren Monitore erkannt. Die zwei Bildschirme zeigen den Bootloader und alles was danach Terminaltechnisch passiert. Da ist der LG schwarz. Erst wenn ich den PC wieder aus mache sehe ich alles auf den zwei Bildschirmen stoppen und herunterfahren.

Benutzt du den Open Source Treiber von Linux selbst oder nutzt du den von Hersteller Nvidia angebotenen Teriber auf dessen Webseite ?

Bild zum Beitrag

 - (Grafikkarte, Nvidia, Treiber)

Ahegao785 
Beitragsersteller
 28.06.2024, 00:23

Den offiziellen von Nvidia.

Flips100  28.06.2024, 00:28
@Ahegao785

Diesen sollte man besser nicht Nutzen. da dieser unsauber Funktioniert. Viele Nutzer unter Linux machen den Fehler und wundern sich warum der Herstellereigene Treiber so schlecht ist. Vor einiger Zeit hat Nvidia mal einen Linux Kernel so Modifiziert und an den Urvater von Linux also Linus Torwalds geschickt. Dieser war so schockiert ,das er diesen Kernel glatt Verboten hat. Durch die Unkooperative Haltung von Nvidia zu Linux kam es mehr oder weniger zum Streit.

Endergebnis ist die Haltung von Linux Torwalds betreffend Nvidia.

https://www.youtube.com/watch?v=xPh-5P4XH6o

Deinstalliere den Treiber vom Hersteller und Nutze den Open Source der von Linux angeboten wird. Er Funktioniert schlichtweg besser als der vom Hersteller Nvidia.

Ahegao785 
Beitragsersteller
 28.06.2024, 00:37
@Flips100

Ich hatte das auch schon einige Male gelesen, dachte aber nie das es so einen großen Unterschied machen würde.Danke für die Info. Ich werd mal die Treiber wechseln und melde mich dann wieder.

Ahegao785 
Beitragsersteller
 28.06.2024, 01:56
@Flips100

Also, Treiber plus xorg conf entfernt und xorg conf neu konfiguriert. Alles funktioniert, aber ich nutze nicht die Graka. Ich installiere die Open-Source Treiber und sofort fangen wieder die Probleme an. Der LG in der Mitte funktioniert nicht mehr und nvidia-smi funktioniert auch nicht. Auch nvidia-settings spuckt nichts mehr aus.

Flips100  28.06.2024, 08:31
@Ahegao785

Lass die Konfig von xorg.conf in ruhe ,nicht dran rumfummeln oder was verändern. Mitunter sind die Werte dort auch Maßgebend das dein LG da nicht mitmacht.

cp xorg.conf xorg.conf.bak

rm -rf xorg.conf

lass sie neu sich erstellen keine Manuelle Konfigs machen.

Hilft das nicht Daten sichern und Linux neu aufsetzen . die Datei im detail Punkt für Punkt durcgehen dauert Stunden. Da geht eine Neuinstallation schneller .

Ahegao785 
Beitragsersteller
 28.06.2024, 15:13
@Flips100

Ich hatte die Config gelöscht, vorher natürlich ein Backup erstellt, habe dann das ganze komplett neuinstalliert und automatisch konfigurieren lassen. Nach einem Neustart bringt das auch wieder alles zum Leben. Auf dem Stand bin ich gerade wieder. Ich hatte davor die Open-Source Variante von Nvidia mit Cuda installiert, ich probier mal ohne. Ich hab genau dieses Problem übrigens bisher jedes mal gehabt auf mehreren Rechnern, eine neuinstallation hilft da also meiner Erfahrung nach leider nicht viel.

Ahegao785 
Beitragsersteller
 28.06.2024, 15:24
@Flips100

Immernoch genau dasselbe Problem, funktioniert mit Wayland vllt. besser?
Irgendwie scheint er ja die Nvidia Treiber nicht richtig zu laden, da ja smi und die Settings beide nicht funktionieren.

Ahegao785 
Beitragsersteller
 28.06.2024, 15:30
@Flips100

Die einzige xorg.conf Datei die ich außer meinem Backup finde ist xorg.conf.nvidia-xconfig-original und diese ist leer.

Flips100  28.06.2024, 20:00
@Ahegao785
Also, Treiber plus xorg conf entfernt und xorg conf neu konfiguriert. Alles funktioniert, aber ich nutze nicht die Graka. Ich installiere die Open-Source Treiber und sofort fangen wieder die Probleme an.

Linux erkennt die Hardware automatisch man muss nichts dazu Installieren.Die Treiber kommen vom Kernel nicht von einer Externen Quelle. Wenn man was Verpfuscht hat dazu , wäre es besser von Grund auf neu zu Installieren.

Jeder Versuch sich irgendwelche Vorteile mit Tools über Nvidia zu verschaffen werden unter Linux Scheitern. Dafür kannst dich beim Hersteller Bedanken das er so Weise und Flexibel bzw. so Kooperativ mit Linux ist (Nvidia)

Sobald man irgendwas wegen Nvidia Installiert ,fangen die Probleme erst wieder an.

Ahegao785 
Beitragsersteller
 30.06.2024, 17:06
@Flips100

Also, ich hatte die Frage auch auf einem anderen Forum gestellt, das ganze ist jetzt geklärt. Trotzdem danke für die Hilfe. Für jeden der die Antwort benötigt:
In the comments, you said the 

lspci -d ::0300

 output was:

00:02.0 VGA compatible controller: Intel Corporation HD Graphics 630 (rev 04) 
01:00.0 VGA compatible controller: NVIDIA Corporation GP106M [GeForce GTX 1060 Mobile] (rev a1)

indicating that your system has two GPUs: the CPU-integrated Intel GPU (iGPU for short) and the NVIDIA discrete GPU (dGPU for short), and the 

xrandr --listproviders

 output was:

Providers: number : 1 
Provider 0: id: 0x1b7 cap: 0x1, Source Output crtcs: 4 outputs: 3 associated providers: 0 name:NVIDIA-0

indicating that only the NVIDIA GPU is currently being used.

The problem

You have a dual-GPU laptop: NVIDIA calls this Optimus Technology. Unfortunately, the most common implementation of this does not include a multiplexer switch that would allow either GPU a physical connection to each display output connector. This is known as "muxless Optimus". When using a combination of display connections that are not directly wired to the active GPU, the missing connections must be substituted by passing rendered output data from one GPU to the other, which can then output the data to the displays it's physically connected to.

In laptops, a common muxless Optimus solution is as follows:

  • the laptop's own display and VGA output is wired to the iGPU (allowing power-efficient operation when no external displays are available, or when using the VGA output for presentation on a video projector)
  • any digital display outputs such as HDMI and/or DisplayPort are usually wired to the dGPU, for maximum performance when connected to external desktop display (and presumably external power supply is readily available, so being economical with battery power is not an issue).

In your specific case, it seems that only the HDMI output of the NVIDIA GPU chip is actually wired to a physical output connector.

The solution

First, you should ensure that you have the 

xserver-xorg-video-intel

 package installed. It contains the X11 driver for your Intel iGPU, which is required even just to pass display data through it. That driver, in turn, requires that the kernel module 'i915' is loaded, and that any firmware files it requires are present, i.e. the 

firmware-misc-nonfree

 package is installed.

Then, you should make a backup of your current 

/etc/X11/xorg.conf

 file (if it exists), and then change it to contain the following (creating the file if it doesn't exist):

Section "ServerLayout"
    Identifier "layout"
    Screen 0 "nvidia"
    Inactive "intel"
EndSection

Section "Device"
    Identifier "nvidia"
    Driver "nvidia"
    BusID "PCI:1:0:0"
EndSection

Section "Screen"
    Identifier "nvidia"
    Device "nvidia"
    Option "AllowEmptyInitialConfiguration"
EndSection

Section "Device"
    Identifier "intel"
    Driver "modesetting"
EndSection

Section "Screen"
    Identifier "intel"
    Device "intel"
EndSection

(This configuration snippet is directly from the NVIDIA GPU driver README file, with the BusID for your hardware already written in.)

This will tell the X server to start both drivers (even if no displays seem to be connected), and tell it to use the NVIDIA GPU as the primary one.

With modern enough version of the Xorg X server, this alone might be enough to make your extra displays available for use.

But if not, you might need two commands to explicitly feed display data from NVIDIA GPU to the Intel one:

xrandr --setprovideroutputsource modesetting NVIDIA-0
xrandr --auto

The first command tells the Intel GPU base driver (

modesetting

) to act as extra outputs for display data whose source is the NVIDIA GPU.

The second command makes the X server activate any displays that might have become available as the result of the previous command.

If you find the two 

xrandr

commands are needed, you could automate the process by writing them into e.g. 

/etc/X11/Xsession.d/92custom_xrandr

.

Flips100  30.06.2024, 17:59
@Ahegao785

Das Bestätigt sogar meine Meinung.

Auch die von Linus Towalds Betreffend Nvidia, also dem Urvater von Linux selbst.

https://www.youtube.com/watch?v=xPh-5P4XH6o

Besser man nutzt eine AMD Grafikkarte ,damit ist man besser Bedient. Oder man Boykottiert Nvidia indem man keine GK von denen Verwendet bis Sie einknicken und gescheite Treiber zu Ihren Grafikkarten liefern.