GPU Unused: Nvidia (Help, desperate)

helplps · November 5, 2024, 6:51am

I’m at the end of my rope. I’ve tried reinstalling, uninstalling, open and proprietary Nvidia drivers. Now I realize even Nouveau doesn’t even work anymore, even if I modprobe, even if I un-blacklist. I have already posted a few posts on here about my problem (this, and this2), and all that’s made me realize is I’ve been trying to fix this problem for exactly 2 weeks from now. I also have tried to Snapper rollback, except even in the read-only backups, I could not get any driver to work for my GPU. Furthermore, I even tried to reinstall OpenSUSE from a USB Flash, but I couldn’t bring myself to wipe my hard drive. Although, in the Live Bootable USB of OpenSUSE, the Nouveau driver DOES work and uses the GPU, which fairly confuses me since I can’t even use Nouveau on my main system. And thing is, I used to be able to use my GPU on this system, until suddenly I just can’t after an update.

I’m here to ask what the best possible action to take here. I could reinstall OpenSUSE by wiping the hard drive, but I’ll need to know how to FULLY back everything up to another storage system, and for that, I’ll need help to know what exactly I should back up. I could also move over to Fedora, since I have another system that uses it, and so far I’ve been getting a good experience, and it’s been getting better. I’m open to more options to take, if anyone can provide one.

What’s the best possible action to take?

mrmazda · November 5, 2024, 8:33am

[tongue-in-cheek]
1.Never install NVidia proprietary drivers in the first place.
2.Don’t buy NVidia hardware in the first place.
[/tongue-in-cheek]
Note that I’ve never installed NVidia’s proprietary drivers on anything I own. I did once install for a customer. My own simply don’t need them. I’ve never used Optimus or a related AMD/NVidia GPU pair either.

The problem with NVidia’s drivers traditionally has been it’s not just a matter of simply installing an rpm, and if it doesn’t produce satisfaction, removing that rpm. They’re reputed to provide removal instructions along with their installation instructions, because under the covers, it’s not one simple process to install, thus typically is not one simple process to reverse. Installation affects initrds, so removal requires undoing that. Typically, installation installs otherwise unnecessary files in /etc/X11/xorg.con* that are inconsistent with FOSS driver usage, so on removal they need purging too. Typically, installation blocks use of the FOSS drivers, typically by blacklisting, sometimes via cmdlne options, so those blockages, if present, must be undone. All this, and possibly more, especially when multiple GPUs are present, should be in the removal instructions, if they can be located. Not infrequently it turns out reinstalling NVidia proprietary drivers, or installing a different version of them, will resolve the issue. In the end, it often turns out to be a fresh installation of the whole operating system is easier than a purge.

I used the word “traditionally” because for the newest NVidia GPUs, NVidia itself is now a participant in FOSS driver development for its newest products, so that the proprietary drivers can be unnecessary to satisfy more users than the older reverse-engineered FOSS drivers could. If yours is new enough, nvidia-open-driver-G06* should provide an alternative to proprietary.

When an NVidia GPU becomes old enough, proprietary drivers cease to be available. All the NVidia GPUs I own, approximately 10 fit for use with current software, fit the latter category. With one recently arising exception ( GK107 10de:0fc1 Kepler; Zotac GeForce GT 640), all work as expected in TW using only FOSS.

Understanding where I’m coming from, I suggest best next step may be finding those installation instructions’ removal instructions and trying them, especially if yours qualifies to use nvidia-open-driver-G06*.

malcolmlewis · November 5, 2024, 12:40pm

@helplps Hi and welcome to the Forum
Can you please post the output from inxi -GSaz I’m assuming you installed the rpms?

helplps · November 5, 2024, 1:04pm

Thank you for this bit of wisdom. As much as I want to stop using Nvidia, it’s just the best in terms of performance when it comes to video editing or 3D rendering. I knew I should have gone for the open drivers, but when I installed it and started using OpenSUSE for the very first time, the proprietary driver instructions from the official OpenSUSE site were the first to show up.

As for finding the uninstallation instructions, I’m still currently in the phase of doing so, however it is looking rather bleak after a few searches…

helplps · November 5, 2024, 1:05pm

Hello, and thank you for replying. This is the output, as asked:

System:
  Kernel: 6.11.5-2-default arch: x86_64 bits: 64 compiler: gcc v: 14.2.1
    clocksource: hpet avail: acpi_pm
    parameters: BOOT_IMAGE=/boot/vmlinuz-6.11.5-2-default
    root=/dev/mapper/lvmlmde-root splash
    resume=/dev/disk/by-uuid/44375b7f-170a-41a0-ac74-6b7ff8ef1b79 psi=1
    mitigations=auto security=apparmor amd_iommu=on kvm.ignore_msrs=1
  Desktop: GNOME v: 47.1 tk: GTK v: 3.24.43 wm: gnome-shell
    tools: gsd-screensaver-proxy avail: xscreensaver dm: GDM v: 47.0
    Distro: openSUSE Tumbleweed 20241103
Graphics:
  Device-1: NVIDIA GA104M [GeForce RTX 3070 Mobile / Max-Q] vendor: Lenovo
    driver: N/A alternate: nouveau, nvidia_drm, nvidia non-free: 550.xx+
    status: current (as of 2024-09; EOL~2026-12-xx) arch: Ampere code: GAxxx
    process: TSMC n7 (7nm) built: 2020-2023 pcie: gen: 3 speed: 8 GT/s
    lanes: 8 link-max: gen: 4 speed: 16 GT/s lanes: 16 bus-ID: 01:00.0
    chip-ID: 10de:24dd class-ID: 0300
  Device-2: Advanced Micro Devices [AMD/ATI] Cezanne [Radeon Vega Series /
    Radeon Mobile Series] vendor: Lenovo driver: amdgpu v: kernel arch: GCN-5
    code: Vega process: GF 14nm built: 2017-20 pcie: gen: 3 speed: 8 GT/s
    lanes: 16 link-max: gen: 4 speed: 16 GT/s ports: active: eDP-1 empty: none
    bus-ID: 05:00.0 chip-ID: 1002:1638 class-ID: 0300 temp: 47.0 C
  Display: wayland server: X.org v: 1.21.1.12 with: Xwayland v: 24.1.3
    compositor: gnome-shell driver: X: loaded: modesetting failed: nvidia
    gpu: amdgpu display-ID: 0
  Monitor-1: eDP-1 model-id: CSO 0x1609 built: 2021 res: 2560x1600 dpi: 188
    gamma: 1.2 size: 345x215mm (13.58x8.46") diag: 407mm (16") ratio: 16:10
    modes: max: 2560x1600 min: 640x480
  API: OpenGL v: 4.6 vendor: amd mesa v: 24.1.7 glx-v: 1.4 es-v: 3.2
    direct-render: yes renderer: AMD Radeon Graphics (radeonsi renoir LLVM
    18.1.8 DRM 3.59 6.11.5-2-default) device-ID: 1002:1638 memory: 3.91 GiB
    unified: no display-ID: :0.0
  API: EGL Message: EGL data requires eglinfo. Check --recommends.

helplps · November 5, 2024, 1:08pm

Hello, and thank you for replying. This is the output, as asked (Oops, I replied to myself…):

System:
  Kernel: 6.11.5-2-default arch: x86_64 bits: 64 compiler: gcc v: 14.2.1
    clocksource: hpet avail: acpi_pm
    parameters: BOOT_IMAGE=/boot/vmlinuz-6.11.5-2-default
    root=/dev/mapper/lvmlmde-root splash
    resume=/dev/disk/by-uuid/44375b7f-170a-41a0-ac74-6b7ff8ef1b79 psi=1
    mitigations=auto security=apparmor amd_iommu=on kvm.ignore_msrs=1
  Desktop: GNOME v: 47.1 tk: GTK v: 3.24.43 wm: gnome-shell
    tools: gsd-screensaver-proxy avail: xscreensaver dm: GDM v: 47.0
    Distro: openSUSE Tumbleweed 20241103
Graphics:
  Device-1: NVIDIA GA104M [GeForce RTX 3070 Mobile / Max-Q] vendor: Lenovo
    driver: N/A alternate: nouveau, nvidia_drm, nvidia non-free: 550.xx+
    status: current (as of 2024-09; EOL~2026-12-xx) arch: Ampere code: GAxxx
    process: TSMC n7 (7nm) built: 2020-2023 pcie: gen: 3 speed: 8 GT/s
    lanes: 8 link-max: gen: 4 speed: 16 GT/s lanes: 16 bus-ID: 01:00.0
    chip-ID: 10de:24dd class-ID: 0300
  Device-2: Advanced Micro Devices [AMD/ATI] Cezanne [Radeon Vega Series /
    Radeon Mobile Series] vendor: Lenovo driver: amdgpu v: kernel arch: GCN-5
    code: Vega process: GF 14nm built: 2017-20 pcie: gen: 3 speed: 8 GT/s
    lanes: 16 link-max: gen: 4 speed: 16 GT/s ports: active: eDP-1 empty: none
    bus-ID: 05:00.0 chip-ID: 1002:1638 class-ID: 0300 temp: 47.0 C
  Display: wayland server: X.org v: 1.21.1.12 with: Xwayland v: 24.1.3
    compositor: gnome-shell driver: X: loaded: modesetting failed: nvidia
    gpu: amdgpu display-ID: 0
  Monitor-1: eDP-1 model-id: CSO 0x1609 built: 2021 res: 2560x1600 dpi: 188
    gamma: 1.2 size: 345x215mm (13.58x8.46") diag: 407mm (16") ratio: 16:10
    modes: max: 2560x1600 min: 640x480
  API: OpenGL v: 4.6 vendor: amd mesa v: 24.1.7 glx-v: 1.4 es-v: 3.2
    direct-render: yes renderer: AMD Radeon Graphics (radeonsi renoir LLVM
    18.1.8 DRM 3.59 6.11.5-2-default) device-ID: 1002:1638 memory: 3.91 GiB
    unified: no display-ID: :0.0
  API: EGL Message: EGL data requires eglinfo. Check --recommends.

malcolmlewis · November 5, 2024, 1:16pm

@helplps Is suse-prime installed? zypper se -i suse-prime It seems strange that Nvidia is showing as device 1. Is there a BIOS setting for GPU order?

helplps · November 5, 2024, 2:05pm

I apologize, suse-prime was not installed, but I have now installed it. This is the result after a reboot:

System:
  Kernel: 6.11.5-2-default arch: x86_64 bits: 64 compiler: gcc v: 14.2.1
    clocksource: hpet avail: acpi_pm
    parameters: BOOT_IMAGE=/boot/vmlinuz-6.11.5-2-default
    root=/dev/mapper/lvmlmde-root splash
    resume=/dev/disk/by-uuid/44375b7f-170a-41a0-ac74-6b7ff8ef1b79 psi=1
    mitigations=auto security=apparmor amd_iommu=on kvm.ignore_msrs=1
  Desktop: GNOME v: 47.1 tk: GTK v: 3.24.43 wm: gnome-shell
    tools: gsd-screensaver-proxy avail: xscreensaver dm: GDM v: 47.0
    Distro: openSUSE Tumbleweed 20241103
Graphics:
  Device-1: NVIDIA GA104M [GeForce RTX 3070 Mobile / Max-Q] vendor: Lenovo
    driver: N/A alternate: nouveau, nvidia_drm, nvidia non-free: 550.xx+
    status: current (as of 2024-09; EOL~2026-12-xx) arch: Ampere code: GAxxx
    process: TSMC n7 (7nm) built: 2020-2023 pcie: gen: 3 speed: 8 GT/s
    lanes: 8 link-max: gen: 4 speed: 16 GT/s lanes: 16 bus-ID: 01:00.0
    chip-ID: 10de:24dd class-ID: 0300
  Device-2: Advanced Micro Devices [AMD/ATI] Cezanne [Radeon Vega Series /
    Radeon Mobile Series] vendor: Lenovo driver: amdgpu v: kernel arch: GCN-5
    code: Vega process: GF 14nm built: 2017-20 pcie: gen: 3 speed: 8 GT/s
    lanes: 16 link-max: gen: 4 speed: 16 GT/s ports: active: eDP-1 empty: none
    bus-ID: 05:00.0 chip-ID: 1002:1638 class-ID: 0300 temp: 47.0 C
  Display: wayland server: X.org v: 1.21.1.12 with: Xwayland v: 24.1.3
    compositor: gnome-shell driver: X: loaded: modesetting failed: nvidia
    gpu: amdgpu display-ID: 0
  Monitor-1: eDP-1 model-id: CSO 0x1609 built: 2021 res: 2560x1600 dpi: 188
    gamma: 1.2 size: 345x215mm (13.58x8.46") diag: 407mm (16") ratio: 16:10
    modes: max: 2560x1600 min: 640x480
  API: OpenGL v: 4.6 vendor: amd mesa v: 24.1.7 glx-v: 1.4 es-v: 3.2
    direct-render: yes renderer: AMD Radeon Graphics (radeonsi renoir LLVM
    18.1.8 DRM 3.59 6.11.5-2-default) device-ID: 1002:1638 memory: 3.91 GiB
    unified: no display-ID: :0.0
  API: EGL Message: EGL data requires eglinfo. Check --recommends.

If you’d like, I could also send the installation text when I installed suse-prime in a pastebin.

malcolmlewis · November 5, 2024, 2:14pm

@helplps No, we don’t want it installed So please uninstall…

What about BIOS settings for iGPU and dGPU?

helplps · November 5, 2024, 2:31pm

Ah, I see. I apologize. There are only 2 options for the GPU, “Dynamic” or “Discrete” mode, where Dynamic uses both iGPU and dGPU depending on the performance needed, whilst Discrete only uses the Nvidia GPU. I have to use Dynamic mode or else I literally would not be able to see anything. I have however tested that the system is fully operable, even when I can’t see what’s happening.

malcolmlewis · November 5, 2024, 2:36pm

@helplps OK, so what nvidia drivers are installed, zypper se -si nvidia

helplps · November 5, 2024, 3:23pm

S  | Name                               | Type    | Version              | Arch   | Repository
---+------------------------------------+---------+----------------------+--------+----------------------
i  | kernel-firmware-nvidia-gspx-G06    | package | 565.57.01-1          | x86_64 | cuda
i  | nvidia-compute-G06                 | package | 565.57.01-1          | x86_64 | cuda
i  | nvidia-compute-G06-32bit           | package | 565.57.01-1          | x86_64 | cuda
i  | nvidia-compute-utils-G06           | package | 565.57.01-1          | x86_64 | cuda
i  | nvidia-drivers-G06                 | package | 565.57.01-1          | x86_64 | cuda
i  | nvidia-gl-G06                      | package | 565.57.01-1          | x86_64 | cuda
i  | nvidia-gl-G06-32bit                | package | 565.57.01-1          | x86_64 | cuda
i+ | nvidia-open                        | package | 565.57.01-1          | x86_64 | cuda
i  | nvidia-open-driver-G06-kmp-default | package | 565.57.01-1          | x86_64 | cuda
i  | nvidia-utils-G06                   | package | 565.57.01-1          | x86_64 | cuda
i  | nvidia-video-G06                   | package | 565.57.01-1          | x86_64 | cuda
i  | nvidia-video-G06-32bit             | package | 565.57.01-1          | x86_64 | cuda
i+ | openSUSE-repos-Tumbleweed-NVIDIA   | package | 20240712.dd8c2eb-1.2 | x86_64 | Main Repository (OSS)

malcolmlewis · November 5, 2024, 3:57pm

@helplps So your running the open version, I do here but use the cuda run file and also driver updates via the run file.

Can you show the output from /sbin/lspci -nnk | grep -EA3 "VGA|Display|3D"

Your using psi=1 grub option, any particular reason?

helplps · November 5, 2024, 5:47pm

I apologize for the late reply. I also won’t be able to reply for the next 5-8 hours, so I hope you won’t mind

01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GA104M [GeForce RTX 3070 Mobile / Max-Q] [10de:24dd] (rev a1)
	Subsystem: Lenovo Device [17aa:3a54]
	Kernel modules: nouveau, nvidia_drm, nvidia
01:00.1 Audio device [0403]: NVIDIA Corporation GA104 High Definition Audio Controller [10de:228b] (rev a1)
--
05:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Cezanne [Radeon Vega Series / Radeon Vega Mobile Series] [1002:1638] (rev c5)
	Subsystem: Lenovo Device [17aa:3a54]
	Kernel driver in use: amdgpu
	Kernel modules: amdgpu

malcolmlewis · November 5, 2024, 7:07pm

@helplps No worries, what about the grub option?

helplps · November 6, 2024, 3:55am

I added it as instructed here by the Waydroid OpenSUSE installation instructions. I’m not quite sure what it does, though.

malcolmlewis · November 6, 2024, 12:40pm

@helplps So are you using this repo, what other non standard repos are present?

Can you show the output from zypper lr -dE

helplps · November 6, 2024, 3:21pm

#  | Alias                                | Name                                             | Enabled | GPG Check | Refresh | Priority | Type   | URI                                                                                        | Service
---+--------------------------------------+--------------------------------------------------+---------+-----------+---------+----------+--------+--------------------------------------------------------------------------------------------+--------
 1 | NVIDIA                               | NVIDIA                                           | Yes     | (r ) Yes  | Yes     |   99     | rpm-md | https://download.nvidia.com/opensuse/tumbleweed                                            | 
 2 | cuda                                 | cuda                                             | Yes     | (r ) Yes  | No      |  100     | rpm-md | https://developer.download.nvidia.com/compute/cuda/repos/opensuse15/x86_64/                | 
 3 | download.opensuse.org-non-oss        | Main Repository (NON-OSS)                        | Yes     | (r ) Yes  | Yes     |   99     | rpm-md | http://download.opensuse.org/tumbleweed/repo/non-oss/                                      | 
 4 | download.opensuse.org-oss            | Main Repository (OSS)                            | Yes     | (r ) Yes  | Yes     |   99     | rpm-md | http://download.opensuse.org/tumbleweed/repo/oss/                                          | 
 5 | download.opensuse.org-tumbleweed     | Main Update Repository                           | Yes     | (r ) Yes  | Yes     |   99     | rpm-md | http://download.opensuse.org/update/tumbleweed/                                            | 
 6 | gitlab.com_paulcarroty_vscodium_repo | gitlab.com_paulcarroty_vscodium_repo             | Yes     | (r ) Yes  | No      |   99     | rpm-md | https://download.vscodium.com/rpms/                                                        | 
 7 | home_oertel_cnijfilter-3.80          | home:oertel:cnijfilter-3.80 (openSUSE_Factory)   | Yes     | (r ) Yes  | No      |   99     | rpm-md | https://download.opensuse.org/repositories/home:/oertel:/cnijfilter-3.80/openSUSE_Factory/ | 
 8 | home_runa-chin                       | Runa Inoue SUSE Repository (openSUSE_Tumbleweed) | Yes     | (r ) Yes  | No      |   99     | rpm-md | https://download.opensuse.org/repositories/home:/runa-chin/openSUSE_Tumbleweed/            | 
 9 | home_runa-chin_Waydroid              | RISR WayDroid Repository (openSUSE_Tumbleweed)   | Yes     | (r ) Yes  | No      |   99     | rpm-md | https://download.opensuse.org/repositories/home:/runa-chin:/Waydroid/openSUSE_Tumbleweed/  | 
11 | packman                              | Packman                                          | Yes     | (r ) Yes  | Yes     |   90     | rpm-md | https://ftp.fau.de/packman//suse/openSUSE_Tumbleweed/                                      | 
13 | repo-openh264                        | Open H.264 Codec (openSUSE Tumbleweed)           | Yes     | (r ) Yes  | Yes     |   99     | rpm-md | http://codecs.opensuse.org/openh264/openSUSE_Tumbleweed                                    |

malcolmlewis · November 6, 2024, 3:28pm

@helplps I suspect using the Nvidia cuda repo doesn’t do any nvidia module tweaks that the openSUSE rpms do…

In saying that I have my own for my Quadro T400.

What does the output as root user from lsinitrd | grep -E "nvidia|nouveau" show

helplps · November 6, 2024, 3:43pm

I think we’re finally getting somewhere… all this time of trying to install Nvidia drivers, it never failed to show that I didn’t have nvidia.ko, and yet it’s here…

-rw-r--r--   1 root     root           15 Sep 13 21:27 etc/ld.so.conf.d/nvidia-lib32.conf
-rw-r--r--   1 root     root           17 Sep 13 21:27 etc/ld.so.conf.d/nvidia-lib64.conf
-rw-r--r--   1 root     root         1755 Sep 13 21:27 etc/modprobe.d/50-nvidia-default.conf
-rw-r--r--   1 root     root           18 Sep 13 21:27 etc/modprobe.d/nvidia.conf
-rw-r--r--   1 root     root          129 Jul 11 16:37 usr/lib/modprobe.d/09-nvidia-modprobe-bbswitch-G04.conf
-rw-r--r--   1 root     root           49 Jul 11 16:37 usr/lib/modprobe.d/09-nvidia-modprobe-pm-G05.conf
drwxr-xr-x   2 root     root            0 Sep 13 21:27 usr/lib/modules/6.11.5-2-default/kernel/drivers/gpu/drm/nouveau
-rw-r--r--   1 root     root      1472591 Sep 13 21:27 usr/lib/modules/6.11.5-2-default/kernel/drivers/gpu/drm/nouveau/nouveau.ko.zst
-rw-r--r--   1 root     root        14061 Sep 13 21:27 usr/lib/modules/6.11.5-2-default/kernel/drivers/hid/hid-nvidia-shield.ko.zst
-rw-r--r--   1 root     root         3022 Sep 13 21:27 usr/lib/modules/6.11.5-2-default/kernel/drivers/usb/typec/altmodes/typec_nvidia.ko.zst