Slow graphics issues after update 2022/06/12. Perhaps NVIDIA driver related

[Mirdarthos]

Then,

@Mirdarthos

run it with:

glxheads

It might look like nothing’s happening, but it’ll exit when it’s done.

@TWrecks

Just need to clarify what you mean to run it with glxheads?

In the terminal.

[TWrecks]

Right. I thought you might be wanting me to add that command to the previous one or some such.

It’s running now.

[Mirdarthos]

Nah. Just like that.

Edit:

Just saw this, after a few seconds, press the elusive ANY to exit it, go back to the terminal, and copy the output from there.

[TWrecks]

Oh sorry just to clarify. I ran it properly. It’s just doing it’s thing.

Was your last message talking about if I’d run it incorrectly? Or were you saying that if I ran it properly then I should press the any key?

[Mirdarthos]

It shouldn’t take long, so see my edit:

@Mirdarthos

Edit:

Just saw this, after a few seconds, press the elusive ANY to exit it, go back to the terminal, and copy the output from there.

[TWrecks]

I feel like a total idiot lol. I don’t see an any key. What are you talking about?

Oh you mean like this?

    ~  glxheads                                                                     ✔ 
glxheads: exercise multiple GLX connections (any key = exit)
Usage:
  glxheads xdisplayname ...
Example:
  glxheads :0 mars:0 venus:1
Name: :0
  Display:     0x55c3c18421b0
  Window:      0x400002
  Context:     0x55c3c1992738
  GL_VERSION:  4.6.0 NVIDIA 525.60.11
  GL_VENDOR:   NVIDIA Corporation
  GL_RENDERER: NVIDIA GeForce RTX 3080/PCIe/SSE2
 %                 

[Mirdarthos]

In.deed.

Everything looks good to go, so I have no more help/assistance for you. You could always check that integrated graphics are disabled in the BIOS/UEFI - just for and giggles. Also to make sure it’s not that.

Else, I have to say, short of an exorcist, I have no idea. And even that has no guarantee.

Edit:

Try the same with glxgears and provide the terminal output here:

$ glxgears                                                                                                                                                                                                                                           1 ↵
Running synchronized to the vertical refresh.  The framerate should be
approximately the same as the monitor refresh rate.
X connection to :0 broken (explicit kill or server shutdown).
$

[TWrecks]

Damnit. Is there a way I can nuke my installation and start over but keep my data?

[Mirdarthos]

@TWrecks

Is there a way I can nuke my installation and start over but keep my data?

If you haven’t changed anything globally, all settings and data should be in your $HOME directory. So in theoryit should be as simple as:

cp -r ~ <backupPath>

Where:

• <backupPath> is the path where to backup your home directory;
• ~ is a built-in alias/shortcut that points to your home directory.

Also, let’s find out if some reinforcements doesn’t perhaps have a better idea of what’s going on.

*Sending the Manjaro-signal to @bogdancovaciu*

[TWrecks]

Can confirm, haven’t changed anything yet. Also it was working fine before the latest update, so it’s probably only a small issue.

[Mirdarthos]

@TWrecks

so it’s probably only a small issue

…one I’ll kick myself over for not thinking of…

[TWrecks]

Probably xD

[bogdancovaciu]

Hello,

From what i read and see there should be no issue, unless:

• some custom xorg.conf is lurking in the /etc/X11/ when there should be only /etc/X11/mhwd.d/nvidia.conf
• was the /etc/X11/mhwd.d/nvidia.conf manually altered?
• your system might require early KMS to be enabled

From what i know the RTX 3080 will perform faster on PCIe v3 or 4 … Not sure, is it connected on the proper port internally ?

@TWrecks

/PCIe/SSE2

Mine is quite old PC and GPU and is normal to have it like that and be slow in comparison with newer models and motherboards

Graphics:
  Device-1: NVIDIA GM206 [GeForce GTX 960] vendor: eVga.com. driver: nvidia
    v: 525.60.11 alternate: nouveau,nvidia_drm non-free: 520.xx+
    status: current (as of 2022-10) arch: Maxwell code: GMxxx
    process: TSMC 28nm built: 2014-19 pcie: gen: 2 speed: 5 GT/s lanes: 16
    ports: active: none off: DVI-D-1,DVI-I-1 empty: DP-1,HDMI-A-1
    bus-ID: 08:00.0 chip-ID: 10de:1401 class-ID: 0300
  Display: x11 server: X.Org v: 21.1.4 with: Xwayland v: 22.1.5
    compositor: kwin_x11 driver: X: loaded: nvidia gpu: nvidia,nvidia-nvswitch
    display-ID: :0 screens: 1
  Screen-1: 0 s-res: 3840x1080 s-dpi: 101 s-size: 966x272mm (38.03x10.71")
    s-diag: 1004mm (39.51")
  Monitor-1: DVI-D-1 mapped: DVI-D-0 note: disabled pos: right
    model: Cintiq 22HD serial: <filter> built: 2014 res: 1920x1080 hz: 60
    dpi: 102 gamma: 1.2 size: 476x268mm (18.74x10.55") diag: 546mm (21.5")
    ratio: 16:9 modes: max: 1920x1080 min: 640x480
  Monitor-2: DVI-I-1 note: disabled pos: primary,left
    model: LG (GoldStar) W2253 serial: <filter> built: 2009 res: 1920x1080 hz: 60
    dpi: 102 gamma: 1.2 size: 477x268mm (18.78x10.55") diag: 547mm (21.5")
    ratio: 16:9 modes: max: 1920x1080 min: 640x480
  API: OpenGL v: 4.6.0 NVIDIA 525.60.11 renderer: NVIDIA GeForce GTX
    960/PCIe/SSE2 direct render: Yes

One thing i do in KDE Plasma is to add:
ForceCompositionPipeline=On
to Section "Screen" to my both monitors set as fixed layout, and

    Option         "TripleBuffer"  "On"
    Option  "ConnectToAcpid"    "Off"

in Section "Device" of my /etc/X11/mhwd.d/nvidia.conf and i use the /etc/profile.d/kwin.sh with:

export KWIN_TRIPLE_BUFFER=1

in it.

In System Settings i make sure the Compositor is set like this

@TWrecks - have you tried a newer kernel? See with the 6 one.

[TWrecks]

I’m sorry bogdancovaciu, I’m having a bit of trouble parsing what exactly you want me to do.

How do I answer you here? Would you like me to dive right into updating my kernel?

The graphics card was working fine just before the most recent patch. I don’t think there’s an issue with how it’s plugged in. I haven’t changed it since it was working correctly.

[bogdancovaciu]

You initially had the video-nvidia-470xx installed and working fine. With the update, came a newer version of the 470xx nvidia drivers

@philm

NVIDIA got updated to 525.60.11 , 470.161.03 (legacy driver for GKxxx “Kepler” GPUs) and

So, prior to 470.161.03 nvidia driver family, there was 470.141.03 version, and that was working fine for you. After update you noticed a decrees in performance. Then you switched to video-nvidia = that is the latest family with the version 525.60.11 that i know for sure has multiple regressions in performance, vaapi mentioned here Issue after upgrading to 525 series drivers · Issue #126 · elFarto/nvidia-vaapi-driver · GitHub and also discussed on cuEGLStreamProducerConnect returns error 801 on 525.53 driver - #7 by elFarto - Linux - NVIDIA Developer Forums
and i doubt somehow that for some GPU models this issue got fixed

@philm

Nvidia 3060RTX cards may have issues with the 515.76 driver

• FS#75995 : [nvidia] Black X11 Screen and partial lockup when upgraded to 515.76 and dual RTX3060
• https://forums.developer.nvidia.com/t/bug-report-black-x11-screen-and-partial-lockup-when-upgraded-to-515-76-and-dual-rtx3060/228912

---

Somehow i was actually hoping that by mentioning all the above you will disprove my initial concerns, and you actually did, because i didn’t much say what you can do, but what to look for, and now, as conclusion - it seems that also the latest legacy drivers introduced regressions and we should go back to 470.141.03 driver version from the 470xx family, to regain the performance we used to have.

[TWrecks]

That sounds sensible. So just NIVIDIA ■■■■■■■ with us all again? xD

I will test this by backdating my drivers even further and then get back to you perhaps.

Uuuh if you don’t mind, could you help me write the correct console commands? I don’t want to accidentally do even more damage lol.

[bogdancovaciu]

Did you use some custom entries in /etc/X11/mhwd.d/nvidia.conf ?

Mind you, but now i’m investigating how to do it properly for me too, because i can also screw things up.
The thing is, we can’t use downgrade because now we switched to the video-nvidia. We need to switch back to video-nvidia-470xx and then downgrade video-nvidia-470xx from 470.161.03 to 470.141.03 version … but first i have to see how are the modules for the kennel, so i might actually try this on a separate install, once i get some time for it, and write how to do it (if nobody else will).
@Mirdarthos - have you done such thing before?

[TWrecks]

Right understood.

I don’t know what you mean by custom entries in /etc/X11/mhwd.d/nvidia.conf. I’m happy to try! But I pretty much need some hand holding. I hope that’s ok.

[Mirdarthos]

@bogdancovaciu

@Mirdarthos - have you done such thing before?

Nope. Although my GPU is also older, a GeForce GTX 960, it works flawlessly. So I’ve never had to do anything of the sort.

[bogdancovaciu]

@Mirdarthos

GeForce GTX 960

we have sisters GPUs

@bogdancovaciu

GeForce GTX 960

or are they brothers ?

@TWrecks

I’m happy to try!

In theory, you should go with:
sudo mhwd -r pci video-nvidia
then
sudo mhwd -i pci video-nvidia-470xx
to get back to the legacy drivers and then
sudo downgrade nvidia-470xx-utils

and chose from the list the 470.141.03 that was working fine for you prior to the update.