Nvidia GPU Randomly stops working

I recently installed Nvidia drivers using the official HowTo. I have secure boot enabled to I did these steps first. These steps worked beautifully and upon rebooting I was able to run the nvidia-smi results with everything looking good.

ruben@p1g7:~$ nvidia-smi
Sun Jul 19 13:13:50 2026       
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 610.43.03              KMD Version: 610.43.03     CUDA UMD Version: 13.3     |
+-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce RTX 4070 ...    Off |   00000000:01:00.0 Off |                  N/A |
| N/A   38C    P3             14W /   65W |      40MiB /   8188MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|    0   N/A  N/A            2574    C+G   /usr/bin/kwin_wayland                     7MiB |
+-----------------------------------------------------------------------------------------+

I was even able to install CUDA and run some matrix calculations on Julia, play some games and see useage stats on system monitor.

However, randomly through the user session, the GPU stops working. When gaming, the GPU is no longer used, the CPU is only used. When trying to do a GPU calculation, I get a CUDA error: unknown error (code 999, ERROR_UNKNOWN). If I run nvidia-smi I get an error:

nvidia-smi
Unable to determine the device handle for GPU0: 0000:01:00.0: Unknown Error
No devices were found

I’m not sure what’s happening here. If I reboot, everything works again until it randomly stops working. I’m running:

  • Fedora 44 KDE Plasma Edition
  • KDE Plasma Version: 6.7.3
  • KDE Frameworks Version: 6.28.0
  • Qt Version: 6.11.1
  • Kernel Version: 7.1.3-201.fc44.x86_64 (64-bit)
  • Graphics Platform: Wayland
  • Processors: 22 × Intel® Core™ Ultra 7 165H
  • Memory: 32 GiB of RAM (30.8 GiB usable)
  • Graphics Processor 1: Mesa Intel® Arc
  • (before failure) Graphics Processor 2: NVIDIA GeForce RTX 4070 Laptop GPU/PCIe/SSE2
  • (after failure) Graphics Processor 2: Mesa Intel® Arc
  • Manufacturer: LENOVO
  • System Version: ThinkPad P1 Gen 7

Even stranger, I can often close my laptop lid, put the system to sleep, wake up the system and the GPU is still working. However, later on it will stop working with little to no warning, sometimes even between two interactive CUDA calculations in the span of 5 minutes.

Fire up journalctl and have a look at the timeframe where the GPU fails and see why it’s dropping off the bus.

Something like journalctl -b 0 -g nvidia should throw up everything that the GPU reported - pluck out the pertinent error lines and post 'em.

I trimmed down the last message by getting it to fail again. The failure was noticed around 14:22. Apparently the GPU has fallen off the bus

journalctl -b 0 -g nvidia
Jul 19 14:19:46 p1g7 kernel: nvidia: loading out-of-tree module taints kernel.
Jul 19 14:19:46 p1g7 kernel: nvidia-nvlink: Nvlink Core is being initialized, major device number 509
Jul 19 14:19:46 p1g7 kernel: nvidia 0000:01:00.0: enabling device (0000 -> 0003)
Jul 19 14:19:46 p1g7 kernel: NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64  610.43.03  Release Build  (dvs-builder@U22-I3-AD05-33-5)  Tue Jun 30>
Jul 19 14:19:47 p1g7 kernel: nvidia-modeset: Loading NVIDIA UNIX Open Kernel Mode Setting Driver for x86_64  610.43.03  Release Build  (dvs-builder@U22-I3>
Jul 19 14:19:47 p1g7 kernel: [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
Jul 19 14:19:47 p1g7 systemd[1]: Started nvidia-powerd.service - nvidia-powerd service.
Jul 19 14:19:47 p1g7 audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=nvidia-powerd comm="sys>
Jul 19 14:19:47 p1g7 nvidia-powerd[1358]: nvidia-powerd version:2.0 (build 1)
Jul 19 14:19:49 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 14:19:49 p1g7 kernel: [drm] Initialized nvidia-drm 0.0.0 for 0000:01:00.0 on minor 0
Jul 19 14:19:49 p1g7 kernel: nvidia 0000:01:00.0: [drm] No compatible format found
Jul 19 14:19:49 p1g7 kernel: nvidia 0000:01:00.0: [drm] Cannot find any crtc or sizes
Jul 19 14:19:49 p1g7 systemd[1]: nvidia-fallback.service - Fallback to nouveau as nvidia did not load skipped, unmet condition check ConditionPathExists=!>
Jul 19 14:19:51 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 14:19:57 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 14:19:59 p1g7 systemd[2284]: Starting app-nvidia\x2dsettings\x2duser@autostart.service - nvidia-settings...
Jul 19 14:19:59 p1g7 systemd[2284]: Started app-nvidia\x2dsettings\x2duser@autostart.service - nvidia-settings.
Jul 19 14:20:01 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 14:22:45 p1g7 kernel: NVRM: Xid (PCI:0000:01:00): 79, pid=4473, name=nvidia-smi, GPU has fallen off the bus.
Jul 19 14:22:45 p1g7 kernel: CPU: 3 UID: 1000 PID: 4478 Comm: nvidia-smi Tainted: G           O        7.1.3-201.fc44.x86_64 #1 PREEMPT(lazy) 
Jul 19 14:22:45 p1g7 kernel:  _kgspRpcRecvPoll+0x716/0x13e0 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  ? os_get_system_time+0x3b/0x90 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  ? _kgspRpcSendMessage+0xd3/0x190 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  _issueRpcAndWait+0xdd/0x920 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  ? FinnRmApiGetSerializedSize+0x3c/0xb0 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  ? osGetCurrentThread+0x26/0x60 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  rpcRmApiControl_GSP+0x76f/0x940 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  getGpuInfos+0x828/0x960 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  resControl_IMPL+0x1b1/0x1c0 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  gpuresControl_IMPL+0x4c/0xa0 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  serverControl+0x47e/0x590 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  _rmapiRmControl+0x4f2/0x820 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  rmapiControlWithSecInfo+0x79/0x140 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  rmapiControlWithSecInfoTls+0x8f/0xf0 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  _nv04ControlWithSecInfo+0x8d/0xa0 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  RmIoctl+0x934/0xda0 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  ? os_acquire_spinlock+0x12/0x30 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  ? portSyncSpinlockAcquire+0x18/0x30 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  ? rm_ioctl+0x52/0x4e0 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  rm_ioctl+0x66/0x4e0 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  nvidia_ioctl.isra.0+0x797/0xa70 [nvidia]
Jul 19 14:22:45 p1g7 kernel:  nvidia_unlocked_ioctl+0x1d/0x30 [nvidia]

This is where the card fails:

Jul 19 13:41:28 p1g7 kernel: NVRM: Xid (PCI:0000:01:00): 79, pid=4566, name=nvidia-smi, GPU has fallen off the bus.

Remove the card, clean the contacts with some isoprop alcohol and reseat it, making sure it’s fully slotted home and secure.

Also post the handful of lines prior to this message and about 20 or so following it.

For some reason I wasn’t able to post my other lines (eventually it might show up in this discussion). I don’t think this is a hardware issue though. This laptop GPU has been running very nicely in Windows. Moreover, even on Linux, I’ve been able to play for 40 minutes straight with no interruptions. It’s only when I stop doing things that require a GPU when this starts to happen. If the other posts don’t show up, I’ll try to send it again.

At least one of your posts appeared, as I quote it, and now it’s apparently vanished.

I mention cleaning and reseating the card to remove the possibility of connectivity errors - your card dropped off the bus, which means it effectively vanished according to the kernel… and as you post in your initial post, it’s not physically detected after the issue:

I await the entries from journalctl - if they appear we can look into the issue further but without any evidence, it’s hard to build a case.

It was caught by the span filter for some reason. I managed to simulate it again. The GPU seems to fall off the bus when I’m not using it. Here’s the logs from the beginning of the session.

journalctl -b 0 -g nvidia
Jul 19 16:19:42 p1g7 kernel: nvidia: loading out-of-tree module taints kernel.
Jul 19 16:19:42 p1g7 kernel: nvidia-nvlink: Nvlink Core is being initialized, major device number 509
Jul 19 16:19:42 p1g7 kernel: nvidia 0000:01:00.0: enabling device (0000 -> 0003)
Jul 19 16:19:42 p1g7 kernel: NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64  610.43.03  Release Build  (dvs-builder@U22-I3-AD05-33-5)  Tue Jun 30 23:47:20 UTC 2026
Jul 19 16:19:42 p1g7 kernel: nvidia-modeset: Loading NVIDIA UNIX Open Kernel Mode Setting Driver for x86_64  610.43.03  Release Build  (dvs-builder@U22-I3-AD05-33-5)  Tue Jun 30 23:32:22 UTC 2026
Jul 19 16:19:42 p1g7 kernel: [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
Jul 19 16:19:42 p1g7 systemd[1]: Started nvidia-powerd.service - nvidia-powerd service.
Jul 19 16:19:42 p1g7 audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=nvidia-powerd comm="systemd" exe="/usr/lib/systemd/systemd" hos>
Jul 19 16:19:42 p1g7 nvidia-powerd[1432]: nvidia-powerd version:2.0 (build 1)
Jul 19 16:19:44 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 16:19:44 p1g7 kernel: [drm] Initialized nvidia-drm 0.0.0 for 0000:01:00.0 on minor 0
Jul 19 16:19:44 p1g7 kernel: nvidia 0000:01:00.0: [drm] No compatible format found
Jul 19 16:19:44 p1g7 kernel: nvidia 0000:01:00.0: [drm] Cannot find any crtc or sizes
Jul 19 16:19:44 p1g7 systemd[1]: nvidia-fallback.service - Fallback to nouveau as nvidia did not load skipped, unmet condition check ConditionPathExists=!/sys/module/nvidia
Jul 19 16:19:46 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 16:19:53 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 16:19:55 p1g7 systemd[2361]: Starting app-nvidia\x2dsettings\x2duser@autostart.service - nvidia-settings...
Jul 19 16:19:55 p1g7 systemd[2361]: Started app-nvidia\x2dsettings\x2duser@autostart.service - nvidia-settings.
Jul 19 16:19:56 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 16:23:44 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 16:24:04 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 16:25:14 p1g7 kernel: NVRM: Xid (PCI:0000:01:00): 79, pid=4503, name=nvidia-smi, GPU has fallen off the bus.
Jul 19 16:25:14 p1g7 kernel: CPU: 4 UID: 1000 PID: 4508 Comm: nvidia-smi Tainted: G           O        7.1.3-201.fc44.x86_64 #1 PREEMPT(lazy) 
Jul 19 16:25:14 p1g7 kernel:  _kgspRpcRecvPoll+0x716/0x13e0 [nvidia]
Jul 19 16:25:14 p1g7 kernel:  ? os_get_system_time+0x3b/0x90 [nvidia]
Jul 19 16:25:14 p1g7 kernel:  ? _kgspRpcSendMessage+0xd3/0x190 [nvidia]
Jul 19 16:25:14 p1g7 kernel:  _issueRpcAndWait+0xdd/0x920 [nvidia]
Jul 19 16:25:14 p1g7 kernel:  ? FinnRmApiGetSerializedSize+0x3c/0xb0 [nvidia]
Jul 19 16:25:14 p1g7 kernel:  ? osGetCurrentThread+0x26/0x60 [nvidia]
Jul 19 16:25:14 p1g7 kernel:  rpcRmApiControl_GSP+0x76f/0x940 [nvidia]
Jul 19 16:25:14 p1g7 kernel:  getGpuInfos+0x828/0x960 [nvidia]
Jul 19 16:25:14 p1g7 kernel:  resControl_IMPL+0x1b1/0x1c0 [nvidia]

Here’s all the journalctl entries from a time range up to falling off the bus (not just nvidia)

journalctl --since "Jul 19 16:24:04" --until "Jul 19 16:25:15"
Jul 19 16:24:04 p1g7 kernel: nvidia-modeset: WARNING: GPU:0: Correcting number of heads for current head configuration (0x00)
Jul 19 16:24:04 p1g7 kinfocenter[4881]: Action "org.kde.kinfocenter.dmidecode.systeminformation" has no parentWindow, assuming QWidgetWindow(0x5580572e5ee0, name="MainWindow#1Window")
Jul 19 16:24:04 p1g7 systemd[1]: Started dbus-:1.3-org.kde.kinfocenter.dmidecode@1.service.
Jul 19 16:24:04 p1g7 audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=dbus-:1.3-org.kde.kinfocenter.dmidecode@1 comm="systemd" exe="/>
Jul 19 16:24:07 p1g7 systemd[2361]: app-org.kde.kinfocenter@93507bbb11964f0ebd852bbc0a906533.service: Consumed 1.693s CPU time over 23.258s wall clock time, 245.4M memory peak.
Jul 19 16:24:14 p1g7 systemd[1]: dbus-:1.3-org.kde.kinfocenter.dmidecode@1.service: Deactivated successfully.
Jul 19 16:24:14 p1g7 audit[1]: SERVICE_STOP pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=dbus-:1.3-org.kde.kinfocenter.dmidecode@1 comm="systemd" exe="/u>
Jul 19 16:24:52 p1g7 systemd[2361]: Starting systemd-tmpfiles-clean.service - Cleanup of User's Temporary Files and Directories...
Jul 19 16:24:52 p1g7 systemd[2361]: Finished systemd-tmpfiles-clean.service - Cleanup of User's Temporary Files and Directories.
Jul 19 16:24:56 p1g7 PackageKit[3470]: uid 1000 is trying to obtain org.freedesktop.packagekit.system-sources-refresh auth (only_trusted:0)
Jul 19 16:24:56 p1g7 PackageKit[3470]: uid 1000 obtained auth for org.freedesktop.packagekit.system-sources-refresh
Jul 19 16:24:57 p1g7 PackageKit[3470]: refresh-cache transaction /1533_acbceabb from uid 1000 finished with success after 419ms
Jul 19 16:24:58 p1g7 PackageKit[3470]: get-updates transaction /1534_bbadeadc from uid 1000 finished with success after 226ms
Jul 19 16:25:14 p1g7 kernel: NVRM: GPU at PCI:0000:01:00: GPU-e7cd9cc9-dabf-7469-55d0-0a51a40e040b
Jul 19 16:25:14 p1g7 kernel: NVRM: Xid (PCI:0000:01:00): 79, pid=4503, name=nvidia-smi, GPU has fallen off the bus.

Is this a relatively new issue? i.e. this card was working without problems in this machine and it appeared when you updated to the 610 drivers? Or, have you always had this issue?

If the card is not physically disconnecting, and you say it’s stable in windows, then one would have to assume it’s an issue with the driver/kernel interface; ergo, reverting to an older driver which did not exhibit this issue would be prudent until a fix is shipped. Card dropping off the bus due to Nvidia software is not going to be tolerated for long, if it is a software issue.

I installed Fedora 44 about a month ago to replace Kubuntu (which I used previously, mostly for building Docker images without running out of Ram) but I only installed the Nvidia drivers yesterday. I’m doing some heavy CUDA computations next week so I figured I’d prep my machine for it; previously I’ve been able to use CUDA on both Windows and Kubuntu 24.04.

I suggest filing a report on the NVIDIA linux forum: Linux - NVIDIA Developer Forums

But first, make sure all of your laptop’s BIOS/firmware is up to date… just to eliminate that as a potential cause.

Then, I would work to reproduce this with a clean install of Fedora Workstation + drivers from the official NVIDIA repo for Fedora 44: Fedora — NVIDIA Driver Installation Guide

This will get you 1) clean repro logs, and 2) you’d have followed NVIDIA’s official installation procedure, which makes the setup more actionable for them.
Finally, from the repro state, grab a bug report file: »»»»»»»»»» If you have a problem, PLEASE read this first «««««««««« - Linux - NVIDIA Developer Forums

No guarantee that an NVIDIA employee will action your report, but it does happen.

@rgonthenoble I’m sorry, but your messages ended up in the spam filter. I freed the first, its visible now: Nvidia GPU Randomly stops working - #3 by rgonthenoble

Unfortunately, we cannot avoid that occasionally that happens. That is not super serious because everything that is considered spam by the system is reviewed by a moderator. So the issue is only that it can take some hours, at the max a day until your message is freed/published.

Sorry for the inconvenience.

Just to be sure: I freed your first message but not the latter messages → at first glance it looks to me they are subsequent attempts to post the equal content. If I missed something, let me know, I can still free another message(s).

does option "NVreg_DynamicPowerManagement=0x02 make any difference? This will keep the gpu powered on when not used.
Default is 0x03 see nvidia docs for more details

Or configure graphics to discrete only in UEFI settings.

This does make a difference. It doesn’t completely solve the problem, but I was able to run some experiments. First, Fedora had some updates including the kernel so I updated it to
7.1.4-200.fc44.x86_64. Then ran the following experiments on some of the settings. It looks like the setting you recommended performed the best. The GPU fell off the bus much later than the other settings. Considering how this problem seems to manifest itself faster when the GPU is idle, I’m pretty convinced this is an issue with laptop power management software.

modprobe nvidia "NVreg_DynamicPowerManagement=0x00"
(Consumes 2.5W-3.0W when idle, Failed after 9 minutes)
journalctl -b 0 -g NVRM:

Jul 20 12:17:28 p1g7 kernel: NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64  610.43.03  Release Build  (dvs-builder@U22-I3-AD05-33-5)  Tue Jun 30 23:47>
Jul 20 12:17:31 p1g7 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: PRH failed to update thermal limit! @ platform_request_handler.c:855
Jul 20 12:26:28 p1g7 kernel: NVRM: GPU at PCI:0000:01:00: GPU...
Jul 20 12:26:28 p1g7 kernel: NVRM: Xid (PCI:0000:01:00): 79, GPU has fallen off the bus.
Jul 20 12:26:28 p1g7 kernel: NVRM: GPU 0000:01:00.0: GPU has fallen off the bus.
Jul 20 12:26:28 p1g7 kernel: NVRM: krcRcAndNotifyAllChannels_IMPL: RC all channels for critical error 79.


modprobe nvidia "NVreg_DynamicPowerManagement=0x01"
(Consumes 2.5W-3.0W when idle, Failed almost instantly)

journalctl -b 0 -g NVRM:
Jul 20 12:31:43 p1g7 kernel: NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64  610.43.03  Release Build  (dvs-builder@U22-I3-AD05-33-5)  Tue Jun 30 23:47>
Jul 20 12:31:46 p1g7 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: PRH failed to update thermal limit! @ platform_request_handler.c:855
Jul 20 12:32:04 p1g7 kernel: NVRM: GPU at PCI:0000:01:00: GPU...
Jul 20 12:32:04 p1g7 kernel: NVRM: Xid (PCI:0000:01:00): 79, GPU has fallen off the bus.
Jul 20 12:32:04 p1g7 kernel: NVRM: GPU 0000:01:00.0: GPU has fallen off the bus.
Jul 20 12:32:04 p1g7 kernel: NVRM: krcRcAndNotifyAllChannels_IMPL: RC all channels for critical error 79.


modprobe nvidia "NVreg_DynamicPowerManagement=0x02"
(Consumes 2.5W-3.0W when idle, Failed after 35 minutes)

journalctl -b  -g NVRM:
Jul 20 11:15:15 p1g7 kernel: NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64  610.43.03  Release Build  (dvs-builder@U22-I3-AD05-33-5)  Tue Jun 30 23:47>
Jul 20 11:15:18 p1g7 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: PRH failed to update thermal limit! @ platform_request_handler.c:855
Jul 20 11:51:21 p1g7 kernel: NVRM: GPU at PCI:0000:01:00: GPU...
Jul 20 11:51:21 p1g7 kernel: NVRM: Xid (PCI:0000:01:00): 79, GPU has fallen off the bus.
Jul 20 11:51:21 p1g7 kernel: NVRM: GPU 0000:01:00.0: GPU has fallen off the bus.

My Dell Inspiron 16 Plus 7620 with RTX 3060 had the exact same symptoms on F44. With 7.1 kernel, since last week or so, it would keep consuming power

It would also report after some use:

nvidia-smi
Unable to determine the device handle for GPU0: 0000:01:00.0: Unknown Error No devices were found

I also tried:
option "NVreg_DynamicPowerManagement=0x02

4 days ago it died. Most likely a VRAM module developed a short circuit and fried the power stage. RIP.

I thought it was all caused by a hardware error, but now, reading your report, it seems like the software side of it was unrelated.

Things got worse today. I tried to uninstall the Nvidia drivers and revert back to the Nouveau drivers using the official documentation. This ended up giving me black screens when I rebooted. I tried using the rescue option, but I couldn’t even input any commands. I had to reinstall the entire distro.

Now, however, the Nouveau drivers don’t result in my GPU dropping off. If I try playing games, the GPU1 stats still vary, and the Info Center doesn’t show double “Mesa Intel Arc” entries for the graphics processors. In addition, given that this card worked flawlessly with Nvidia drivers in Windows and Kubuntu for two years (and still works great in Windows) I’m pretty sure this is a software issue. Not sure if this is a Fedora/RPM, new Nvidia Driver, or linux kernel issue. Maybe I should have went back to Kubuntu, but I want to give Fedora a try for at least a year. I’ve heard horror stories about Nvidia drivers on Linux, and I guess I’m living through one right now. I think I might have to quit messing with Nvidia drivers while I have a working distro.

You could always try the 580xx drivers from earlier this year - they are the last set of drivers which are compatible with older Pascal cards and are kept as a separate driver package so that anyone with a Pascal card can continue to use that card despite Nv no longer supporting them.

They are stable, they’ve been around for a while, they have a lot of users; they may be more stable for you. If they are not, you can be relatively confident that that issue is not the drivers or “everyone” would have been having these issues you’re seeing, for months.

You could also try hanging back on an older version of the kernel, pre 7-1 to see if that helps.

You could install another distro which offers KDE and seeing if that works any better for you - it may be a combination of compilation flags, your hardware config, your BIOS settings, your overclock or undervolt settings, your PCIe timings, you actual physical motherboard has a dry joint, etc.

I had a 2080 card in this machine for years, and it’s currently in my wife’s desktop, running F44 with whatever driver she currently has in there - utterly rock solid. I’ve done nothing at all to make it that way - it’s just flawless, so there’s definitely combinations which work better than others - maybe you’re just a little unlucky right now with this card, mobo and setting combinations.

As an aside, have you been into the BIOS at all to “load default/optimal settings” or tweak anything to make things a little more stable?

Please remind that there is only one correct procedure to report issue related to the NVIDIA driver. This is documented at https://rpmfusion.org/Howto/NVIDIA#Bug_Report

Please don’t try random users comment in this thread.

Are you informing me or the OP?