ASUS TUF A15 (FA506NCR) + RTX 3050 (dGPU only) - Suspend worked on Fedora 44 before, now black screen after fresh install

Hi everyone,

I’m trying to figure out whether this is a kernel/NVIDIA regression or if I’m missing something.

Laptop

  • ASUS TUF A15 FA506NCR
  • AMD Ryzen 7 7435HS (no iGPU)
  • NVIDIA RTX 3050 Laptop GPU (dGPU only)
  • Fedora 44 Workstation
  • Kernel: 7.0.12-201.fc44.x86_64
  • NVIDIA Driver: 595.80 (RPM Fusion)

Timeline

  • Fedora 43 → Suspend never worked.
  • Upgraded to Fedora 44 when it was released → Suspend suddenly started working perfectly.
  • Used it for about 2–3 months without any suspend/resume issues.
  • Recently wiped the entire SSD and performed a clean Fedora 44 installation from the official USB.
  • Installed the NVIDIA driver again from RPM Fusion.
  • Suspend is broken again.

So this exact laptop has worked before with Fedora 44, but I cannot reproduce that behavior after a clean installation.

Current behavior

When I suspend:

  • Laptop appears to enter suspend.
  • Pressing the power button wakes it.
  • Caps Lock responds after wake.
  • The internal display stays completely black.
  • Keyboard backlight never comes back.
  • I have to force power off.

It looks like the system resumes but the display/GPU never recovers.

What I’ve already checked

  • NVIDIA installed from RPM Fusion.
  • nvidia-smi works.
  • nvidia-drm.modeset=1 is enabled.
  • NVreg_PreserveVideoMemoryAllocations=1
  • NVIDIA suspend/resume services are enabled.
  • Tried disabling GSP firmware (NVreg_EnableGpuFirmware=0).
  • Rebuilt initramfs with dracut -f.
  • BIOS is updated to version 304 (01/08/2025).

Current info

Fedora 44
Kernel: 7.0.12-201.fc44.x86_64
NVIDIA: 595.80

cat /sys/power/state
freeze mem disk

cat /sys/power/mem_sleep
[s2idle]

cat /proc/driver/nvidia/gpus/0000:01:00.0/power

Runtime D3 status: Enabled (fine-grained)
S0ix Platform Support: Supported
S0ix Status: Disabled

My questions

  1. Has anyone with an ASUS TUF (especially FA506NCR) or another dGPU-only NVIDIA laptop experienced this recently?
  2. Is this a known regression in Linux 7.0.x or NVIDIA 595.80?
  3. Is there any kernel parameter, ACPI workaround, or NVIDIA setting that I’m missing?
  4. Has anyone successfully fixed suspend/resume on this exact hardware?

I’d appreciate any ideas. At this point I’m trying to determine whether this is a configuration issue or simply a kernel/NVIDIA regression.

Thanks!

I would search for reports of suspend/resume issues with your model across other linux distros as your issue is probably not specific to Fedora. Assuming you preserved or restored your home directory, you could try creating a “test” login to see if the issue is linked to your home directory.

Thanks. It was a completely fresh installation after wiping the SSD, so this isn’t the same home directory from my previous Fedora installation. The issue occurs with the fresh user created during installation as well. That’s why I’m leaning toward a kernel/NVIDIA/firmware issue rather than user-specific configuration.

Update: I finally fixed the suspend/resume issue on my ASUS TUF A15 (FA506NCR) with RTX 3050 (dGPU only)

After several hours of debugging, I finally found a solution that consistently fixed suspend/resume on my laptop.

Hardware

  • ASUS TUF A15 FA506NCR
  • AMD Ryzen 7 7435HS (no iGPU)
  • NVIDIA RTX 3050 Laptop GPU (dGPU only)
  • Fedora 44 Workstation
  • Kernel 7.0.12
  • NVIDIA 595.80 (RPM Fusion)

Symptoms

  • Suspend appeared to work.
  • Pressing the power button woke the laptop.
  • Caps Lock responded.
  • Internal display remained black.
  • Keyboard backlight never came back.
  • Only a forced shutdown recovered the system.

What did NOT fix it

  • NVreg_PreserveVideoMemoryAllocations=1
  • NVreg_TemporaryFilePath=/var/tmp
  • nvidia-suspend.service
  • nvidia-resume.service
  • nvidia-drm.modeset=1
  • Disabling GSP firmware (NVreg_EnableGpuFirmware=0)

None of these solved the problem by themselves.

What finally fixed it

I edited:

/usr/lib/modprobe.d/nvidia-power.conf

and added:

options nvidia NVreg_PreserveVideoMemoryAllocations=1
options nvidia NVreg_TemporaryFilePath=/var/tmp
options nvidia NVreg_EnableS0ixPowerManagement=1
options nvidia NVreg_DynamicPowerManagement=0x00

Then rebuilt the initramfs:

sudo dracut --force

Rebooted the machine.

Suspend/resume now works correctly again.

Important note

Editing files under /usr/lib/modprobe.d/ is not recommended because package updates may overwrite them.

The proper long-term solution is to create your own override:

sudo nano /etc/modprobe.d/99-nvidia-power.conf

with the same contents:

options nvidia NVreg_PreserveVideoMemoryAllocations=1
options nvidia NVreg_TemporaryFilePath=/var/tmp
options nvidia NVreg_EnableS0ixPowerManagement=1
options nvidia NVreg_DynamicPowerManagement=0x00

Then rebuild the initramfs again:

sudo dracut --force

Why this might help

My laptop only supports s2idle (S0ix) and has no integrated GPU. The internal display is driven entirely by the NVIDIA GPU. Enabling NVIDIA’s S0ix power management and disabling dynamic runtime power management appears to allow the GPU to resume correctly after suspend.

Hopefully this helps anyone with a similar ASUS TUF laptop or another NVIDIA dGPU-only system experiencing black-screen resume issues.

please read the Nvidia documentation explaining Power Management and try to undestand what those lines do.

'your_preferred_browser'  /usr/share/doc/xorg-x11-drv-nvidia/html/dynamicpowermanagement.html 

Most of those lines make no sense and were probably presented to you by some ‘stupid’ LLM, because those exact four lines were previously posted here by an user on this forum.

The issue is a known regression in the 595.80 nvidia drivers and was fixed in 595.84. For some reason the rpmfusion team has not packaged this driver yet. Maybe open a ticket at rpmfusion and request politely an driver update.

All you need is one line changing DynamicPowerManagent from 3 (default) to 2, 0 disables dynamic powermanagement.

options nvidia NVreg_DynamicPowerManagement=0x02