DELL G3/NVIDIA GTX 1050 LINUX PROBLEMS RESOLVED
Overview
The Dell G3 gaming laptop came with an NVIDIA GTX 1050 video card and it would not leave the D3COLD state for D0.
This guide addresses solving that.
Apparently there are 2 levels of power management and Linux doesn't manage them unless you make a simple change to GRUB parameters.
Indicators
- Cannot transition from D3cold to D0
- Cannot Access Any NVIDIA Device
- NVIDIA Drvier dropped off the bus
Solution
RESOLUTION: NVIDIA GTX 1050 "D3cold / fallen off the bus" on Dell G3 Linux
SYSTEM
------
Dell G3 laptop
Intel UHD Graphics 630 + NVIDIA GeForce GTX 1050 3GB Max-Q
Ubuntu/Mint-based Linux
Kernel: 6.8.0-138-generic
NVIDIA driver: 580.173.02
PRIME mode: NVIDIA Performance
PROBLEM
-------
After a fresh Linux installation, the NVIDIA GTX 1050 would not initialize.
The machine had previously worked correctly under both Windows 10 and
Windows 11 immediately before Linux was installed, so a hardware failure
was considered unlikely.
Linux detected the PCI device:
01:00.0 NVIDIA Corporation GP107M [GeForce GTX 1050 3 GB Max-Q]
PCI ID: 10de:1c91
However, the NVIDIA driver could not bind to it.
Typical kernel errors were:
nvidia 0000:01:00.0:
Unable to change power state from D3cold to D0,
device inaccessible
NVRM:
The NVIDIA GPU 0000:01:00.0 has fallen off the bus
and is not responding to commands.
NVRM:
None of the NVIDIA devices were initialized.
The PCI device remained in:
power_state = D3cold
and:
nvidia-smi
failed because it could not communicate with the NVIDIA driver.
An additional symptom was extremely slow booting. gpu-manager.service
could take approximately 80 seconds while repeatedly attempting to
initialize the inaccessible NVIDIA GPU.
INVESTIGATION
-------------
The NVIDIA 580.173.02 DKMS driver was correctly installed for the
running kernel.
Nouveau was blacklisted.
Changing PRIME between on-demand and NVIDIA Performance did not solve
the problem.
Attempts were also made to prevent NVIDIA runtime power management,
including:
options nvidia NVreg_DynamicPowerManagement=0x00
and a udev rule forcing:
/sys/bus/pci/devices/0000:01:00.0/power/control = on
These did not solve the underlying problem. Even with power/control set
to "on", the GPU remained physically inaccessible in D3cold.
gpu-manager was also found to explicitly set the NVIDIA PCI device's
power/control to "auto". It was masked during troubleshooting:
systemctl mask gpu-manager.service
ACPI tables were dumped and examined because the behavior strongly
suggested a platform/PCIe power-management problem.
Testing:
acpi_osi=Linux-Dell-Video
did NOT solve the problem. The GPU still entered D3cold and NVIDIA
continued reporting "fallen off the bus."
SOLUTION
--------
The successful workaround was to disable PCIe Active State Power
Management (ASPM) at the Linux kernel level.
Set in /etc/default/grub:
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash pcie_aspm=off"
Then:
sudo update-grub
sudo poweroff
The laptop was then cold-started.
RESULT
------
After booting with pcie_aspm=off:
power/control = on
power/runtime_status = active
power_state = D0
The NVIDIA PCI device successfully bound to the proprietary driver:
Kernel driver in use: nvidia
All NVIDIA modules loaded successfully:
nvidia
nvidia_modeset
nvidia_drm
nvidia_uvm
nvidia-smi now works correctly and reports:
NVIDIA-SMI 580.173.02
Driver Version: 580.173.02
CUDA Version: 13.0
NVIDIA GeForce GTX 1050
3072 MiB VRAM
Xorg, Cinnamon, and Firefox were confirmed using NVIDIA GPU memory.
The kernel log now shows normal NVIDIA initialization:
nvidia 0000:01:00.0: enabling device
NVRM: loading NVIDIA UNIX x86_64 Kernel Module 580.173.02
nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver
[drm] Initialized nvidia-drm
There are no longer D3cold -> D0 failures or "GPU fallen off the bus"
errors.
IMPORTANT DIAGNOSTIC CLUE
-------------------------
With pcie_aspm=off the kernel reports:
PCIe ASPM is disabled
ACPI FADT declares the system doesn't support PCIe ASPM,
so disable it
FADT indicates ASPM is unsupported, using BIOS configuration
This strongly indicates the underlying problem is interaction between
Linux PCIe power management and the Dell G3 firmware/platform power
management for the discrete NVIDIA GPU.
The NVIDIA 580 driver itself was not the fundamental problem.
BOOT PERFORMANCE
----------------
Before the fix:
Boot time: approximately 1 minute 50 seconds
gpu-manager.service: approximately 1 minute 20 seconds
After the fix:
Total boot: 31.350 seconds
graphical.target: 12.202 seconds
The enormous gpu-manager delay disappeared once the NVIDIA GPU became
accessible.
CURRENT WORKING CONFIGURATION
-----------------------------
GRUB:
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash pcie_aspm=off"
NVIDIA:
Driver 580.173.02
PRIME NVIDIA Performance mode
GTX 1050 successfully bound to nvidia driver
GPU remains in D0 and is operational
NOTE
----
Several troubleshooting changes remain installed:
/etc/modprobe.d/nvidia-power.conf
options nvidia NVreg_DynamicPowerManagement=0x00
/etc/udev/rules.d/99-dell-g3-nvidia-power.rules
forces NVIDIA PCI power/control="on"
gpu-manager.service
masked
These should be removed/tested ONE AT A TIME later to determine the
minimum configuration required.
The confirmed critical change that made the GPU operational was:
pcie_aspm=off
SUSPEND/RESUME TEST
root@MissionControl:/home/warlock# echo "=== BEFORE SUSPEND ==="
date
cat /sys/bus/pci/devices/0000:01:00.0/power_state
nvidia-smi
systemctl suspend
=== BEFORE SUSPEND ===
Thu Aug 20 03:00:49 PM CDT 2026
D0
Thu Aug 20 15:00:49 2026
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 580.173.02 Driver Version: 580.173.02 CUDA Version: 13.0 |
+-----------------------------------------+------------------------+----------------------+
| 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 GTX 1050 Off | 00000000:01:00.0 Off | N/A |
| N/A 54C P3 N/A / 5001W | 315MiB / 3072MiB | 2% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 1075 G /usr/lib/xorg/Xorg 110MiB |
| 0 N/A N/A 1781 G cinnamon 23MiB |
| 0 N/A N/A 2673 G /usr/lib/firefox/firefox 176MiB |
+-----------------------------------------------------------------------------------------+
root@MissionControl:/home/warlock#
root@MissionControl:/home/warlock# echo "=== AFTER RESUME ==="
date
echo
echo "=== NVIDIA PCI ==="
lspci -nnk -s 01:00.0
echo
echo "=== POWER ==="
cat /sys/bus/pci/devices/0000:01:00.0/power/control
cat /sys/bus/pci/devices/0000:01:00.0/power/runtime_status
cat /sys/bus/pci/devices/0000:01:00.0/power_state
echo
echo "=== NVIDIA-SMI ==="
nvidia-smi
echo
echo "=== SUSPEND/RESUME/NVIDIA LOG ==="
dmesg -T | grep -iE \
'nvidia|NVRM|D3cold|D3hot|D0|fallen off|suspend|resume|PM:' | tail -150
=== AFTER RESUME ===
Thu Aug 20 03:01:38 PM CDT 2026
=== NVIDIA PCI ===
01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GP107M [GeForce GTX 1050 3 GB Max-Q] [10de:1c91] (rev a1)
Subsystem: Dell GP107M [GeForce GTX 1050 3 GB Max-Q] [1028:0949]
Kernel driver in use: nvidia
Kernel modules: nvidiafb, nouveau, nvidia_drm, nvidia
=== POWER ===
on
active
D0
=== NVIDIA-SMI ===
Thu Aug 20 15:01:38 2026
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 580.173.02 Driver Version: 580.173.02 CUDA Version: 13.0 |
+-----------------------------------------+------------------------+----------------------+
| 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 GTX 1050 Off | 00000000:01:00.0 Off | N/A |
| N/A 55C P3 N/A / 5001W | 349MiB / 3072MiB | 4% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 1075 G /usr/lib/xorg/Xorg 131MiB |
| 0 N/A N/A 1781 G cinnamon 23MiB |
| 0 N/A N/A 2673 G /usr/lib/firefox/firefox 189MiB |
+-----------------------------------------------------------------------------------------+
=== SUSPEND/RESUME/NVIDIA LOG ===
[Thu Aug 20 14:45:36 2026] e820: update [mem 0x63b5d018-0x63b7ac57] usable ==> usable
[Thu Aug 20 14:45:36 2026] e820: update [mem 0x63b5d018-0x63b7ac57] usable ==> usable
[Thu Aug 20 14:45:36 2026] reserve setup_data: [mem 0x0000000040400000-0x0000000063b5d017] usable
[Thu Aug 20 14:45:36 2026] reserve setup_data: [mem 0x0000000063b5d018-0x0000000063b7ac57] usable
[Thu Aug 20 14:45:36 2026] ACPI: XSDT 0x00000000777410D0 00011C (v01 DELL CBX3 01072009 AMI 00010013)
[Thu Aug 20 14:45:36 2026] ACPI: UEFI 0x0000000077793CD0 000042 (v01 DELL\x CBX3 00000002 01000013)
[Thu Aug 20 14:45:36 2026] ACPI: Reserving UEFI table memory at [mem 0x77793cd0-0x77793d11]
[Thu Aug 20 14:45:36 2026] ACPI: HPET id: 0x8086a201 base: 0xfed00000
[Thu Aug 20 14:45:36 2026] PM: hibernation: Registered nosave memory: [mem 0x00000000-0x00000fff]
[Thu Aug 20 14:45:36 2026] PM: hibernation: Registered nosave memory: [mem 0x0009e000-0x0009efff]
[Thu Aug 20 14:45:36 2026] PM: hibernation: Registered nosave memory: [mem 0x000a0000-0x000fffff]
[Thu Aug 20 14:45:36 2026] PM: hibernation: Registered nosave memory: [mem 0x40000000-0x403fffff]
[Thu Aug 20 14:45:36 2026] PM: hibernation: Registered nosave memory: [mem 0x6cff1000-0x6cff2fff]
[Thu Aug 20 14:45:36 2026] PM: hibernation: Registered nosave memory: [mem 0x72aa1000-0x72b29fff]
[Thu Aug 20 14:45:36 2026] PM: hibernation: Registered nosave memory: [mem 0x7711a000-0x791fdfff]
[Thu Aug 20 14:45:36 2026] PM: hibernation: Registered nosave memory: [mem 0x791ff000-0xffffffff]
[Thu Aug 20 14:45:36 2026] ACPI: PM: Registering ACPI NVS region [mem 0x6cff1000-0x6cff1fff] (4096 bytes)
[Thu Aug 20 14:45:36 2026] ACPI: PM: Registering ACPI NVS region [mem 0x777a5000-0x778abfff] (1077248 bytes)
[Thu Aug 20 14:45:36 2026] PM: RTC time: 14:45:45, date: 2026-08-20
[Thu Aug 20 14:45:36 2026] ACPI: PM: (supports S0 S4 S5)
[Thu Aug 20 14:45:36 2026] pci_bus 0000:00: root bus resource [io 0x0d00-0xffff window]
[Thu Aug 20 14:45:36 2026] pci 0000:00:01.0: PME# supported from D0 D3hot D3cold
[Thu Aug 20 14:45:36 2026] pci 0000:00:12.0: BAR 0 [mem 0xa54ad000-0xa54adfff 64bit]
[Thu Aug 20 14:45:36 2026] pci 0000:00:14.0: PME# supported from D3hot D3cold
[Thu Aug 20 14:45:36 2026] pci 0000:00:16.0: PME# supported from D3hot
[Thu Aug 20 14:45:36 2026] pci 0000:00:17.0: PME# supported from D3hot
[Thu Aug 20 14:45:36 2026] pci 0000:00:1d.0: PME# supported from D0 D3hot D3cold
[Thu Aug 20 14:45:36 2026] pci 0000:00:1d.5: PME# supported from D0 D3hot D3cold
[Thu Aug 20 14:45:36 2026] pci 0000:00:1d.6: PME# supported from D0 D3hot D3cold
[Thu Aug 20 14:45:36 2026] pci 0000:00:1f.3: PME# supported from D3hot D3cold
[Thu Aug 20 14:45:36 2026] pci 0000:03:00.0: PME# supported from D0 D1 D2 D3hot D3cold
[Thu Aug 20 14:45:36 2026] pci 0000:04:00.0: PME# supported from D0 D3hot D3cold
[Thu Aug 20 14:45:36 2026] e820: reserve RAM buffer [mem 0x63b5d018-0x63ffffff]
[Thu Aug 20 14:45:36 2026] system 00:05: [mem 0xfd000000-0xfd69ffff] has been reserved
[Thu Aug 20 14:45:36 2026] clocksource: acpi_pm: mask: 0xffffff max_cycles: 0xffffff, max_idle_ns: 2085701024 ns
[Thu Aug 20 14:45:36 2026] pci_bus 0000:00: resource 5 [io 0x0d00-0xffff window]
[Thu Aug 20 14:45:36 2026] pci 0000:01:00.1: extending delay after power-on from D3hot to 20 msec
[Thu Aug 20 14:45:36 2026] pci 0000:01:00.1: D0 power state depends on 0000:01:00.0
[Thu Aug 20 14:45:36 2026] ACPI: button: Lid Switch [LID0]
[Thu Aug 20 14:45:36 2026] Loaded X.509 cert 'Canonical Ltd. Secure Boot Signing: 61482aa2830d0ab2ad5af10b7250da9033ddcef0'
[Thu Aug 20 14:45:36 2026] PM: Magic number: 6:237:789
[Thu Aug 20 14:45:41 2026] Loaded X.509 cert 'wens: 61c038651aabdcf94bd0ac7ff06c7248db18c600'
[Thu Aug 20 14:45:41 2026] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card0/input35
[Thu Aug 20 14:45:41 2026] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card0/input36
[Thu Aug 20 14:45:41 2026] input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card0/input37
[Thu Aug 20 14:45:41 2026] input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card0/input38
[Thu Aug 20 14:45:42 2026] nvidia: loading out-of-tree module taints kernel.
[Thu Aug 20 14:45:42 2026] nvidia: module license 'NVIDIA' taints kernel.
[Thu Aug 20 14:45:42 2026] nvidia: module verification failed: signature and/or required key missing - tainting kernel
[Thu Aug 20 14:45:42 2026] nvidia: module license taints kernel.
[Thu Aug 20 14:45:42 2026] snd_hda_codec_realtek ehdaudio0D0: autoconfig for ALC3254: line_outs=1 (0x14/0x0/0x0/0x0/0x0) type:speaker
[Thu Aug 20 14:45:42 2026] snd_hda_codec_realtek ehdaudio0D0: speaker_outs=0 (0x0/0x0/0x0/0x0/0x0)
[Thu Aug 20 14:45:42 2026] snd_hda_codec_realtek ehdaudio0D0: hp_outs=1 (0x21/0x0/0x0/0x0/0x0)
[Thu Aug 20 14:45:42 2026] snd_hda_codec_realtek ehdaudio0D0: mono: mono_out=0x0
[Thu Aug 20 14:45:42 2026] snd_hda_codec_realtek ehdaudio0D0: inputs:
[Thu Aug 20 14:45:42 2026] snd_hda_codec_realtek ehdaudio0D0: Headset Mic=0x19
[Thu Aug 20 14:45:42 2026] snd_hda_codec_realtek ehdaudio0D0: Headphone Mic=0x1b
[Thu Aug 20 14:45:42 2026] nvidia-nvlink: Nvlink Core is being initialized, major device number 511
[Thu Aug 20 14:45:42 2026] nvidia 0000:01:00.0: enabling device (0000 -> 0003)
[Thu Aug 20 14:45:42 2026] nvidia 0000:01:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=none
[Thu Aug 20 14:45:42 2026] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 580.173.02 Tue Jun 23 08:38:17 UTC 2026
[Thu Aug 20 14:45:42 2026] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 580.173.02 Tue Jun 23 08:00:06 UTC 2026
[Thu Aug 20 14:45:42 2026] [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
[Thu Aug 20 14:45:43 2026] NVRM: nv_acpi_evaluate_dsm_method: DSM data invalid!
[Thu Aug 20 14:45:43 2026] [drm] Initialized nvidia-drm 0.0.0 20160202 for 0000:01:00.0 on minor 0
[Thu Aug 20 14:45:43 2026] nvidia 0000:01:00.0: [drm] Cannot find any crtc or sizes
[Thu Aug 20 14:45:43 2026] nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint.
[Thu Aug 20 15:00:50 2026] PM: suspend entry (s2idle)
[Thu Aug 20 15:00:56 2026] printk: Suspending console(s) (use no_console_suspend to debug)
[Thu Aug 20 15:01:09 2026] PM: suspend exit
root@MissionControl:/home/warlock#
NVIDIA: FIXED
Cold boot: WORKING
NVIDIA Performance mode: WORKING
Suspend/resume: WORKING
Driver 580.173.02: WORKING
GTX 1050 remains D0/active: WORKING
---
David Beecher (http://dbeecher.tekops.com)