NVIDIA GPU Fails to power off (PRIME) Razer Blade 14 2022

I’ve installed Fedora 37 WorkStation KDE on my Razer Blade 14 2022 and have followed the documentation to enable RPM fusion and install the nvidia driver, and everything is up to date and appears installed properly.

The NVIDIA GPU is able to power off into D3cold status according to cat /sys/class/drm/card*/device/power_state when on AC, but on battery it switches to D0. I would like it to power off and run in D3cold all the time unless prompted to turn on by an application in order to save battery life. I have verified that D3cold works on battery in ubuntu, so it seems to be a matter of proper software configuration.

Please help me set it up so the NVIDIA GPU is powered off on battery! Thanks :)

1 Like

There’s not really anything to configure, rather looks like a bug, possibly with Wayland and/or KDE.
Please run nvidia-bug-report.sh as root and attach the resulting nvidia-bug-report.log.gz file to your post.

Thanks for your response!
Here it is.
nvidia-bug-report-wayland.log.gz (333.5 KB)

nvidia-bug-report-xorg.log.gz (401.3 KB)
I also have one done in xorg. For the record, the issue is the same in both.

It’s a bit odd, the gpu shows some activity and the kernel reports pcie errors. Since it worked with Ubuntu, I suspect problems with the latest 6.2 kernel Fedora uses.
Did you already check for a bios update?

I’ve checked for a BIOS update and am on the latest one. Something odd I noticed is that when I tested opensuse tumbleweed it worked on there with the same kernels as fedora, but importantly they don’t have the nvidia 530 driver yet, they had 525 (and so did ubuntu). This issue did occur in arch with 530 driver as well, so it may be a regression in the newer driver version. I’m not sure though.

I’ve discovered something quite interesting. I ended up installing arch since I am much more familiar with the package manager on there and have discovered the source of the problem. I tried downgrading the nvidia packages to a few prior versions and managed to get it to work on version 525.85.05, the same one that worked on mint and opensuse. Every version after that is broken with regards to this issue. Please let me know what I should do next to hopefully get this fixed.

That’s at least a hint, though for me 525.105.17 still works with a 3050ti so doesn’t seem to be a general bug but something machine specific.
You might want to send your bug-report.log to linux-bugs[at]nvidia.com though chances are low to get some more attention.

I have filed a bug 4084236 internally for tracking purpose.

I tried multiple different notebooks like ThinkPad P1 Gen 6; HP ZBook Fury 16 G9 Mobile Workstation PC and Razer Notebook 16 and installed Fedora 37 which has kernel version 6.0 but could not repro issue.

Later I installed Arch Linux which has kernel version 6.2 on Razer Notebook 16 but did not observe any issue, GPU can enter into D3cold in both AC/DC as well.

I will spend few more cycle on it and update and if required more information/logs from your end, I will let you know.

Thanks in advance.

Thank you for doing that. A user with the same laptop model as me has been able to reproduce this issue on the fedora forums. Perhaps this can help narrow down the issue?

is there a way for me to check if there’s any progress on this issue?

Installed Fedora KDE 37 with kernel 6.4.10-100.fc37.x86_64 on Razer Blade 15 + RTX3070 Ti system but not able to repro issue with driver 535.98, GPU can enter into D3cold in both AC/DC as well.
Could you please check once with driver 535.98 and update.

Hello, I’m the user from the Fedora forums with the same laptop model / specs. Sadly, the issue seems to still be here. I’m running opensuse running nvidia 535.104.05-11.1 kernel 6.4.12-1-default.
nvidia-bug-report.log.gz (838.7 KB)

Hi there,

I’ve been facing the same problem on my Razer Blade 14 2022 (AMD 5900HX - Nvidia RTX 3070) for several months, where the GPU seemingly cycles between the active power state into the suspended state for about a second, and then back into the active state again for about 10+ seconds. The only thing that I’ve found that has worked on Arch Linux for me is downgrading back to the 525xx drivers on the AUR. I believe this issue started when I upgrade past the official 525 drivers onto 530, and I have not been able to get it to work with any newer version since.

Kernel version:

❯ uname -r

Current Nvidia packages:

❯ sudo pacman -Q | grep nvidia
lib32-nvidia-utils 535.104.05-1
lib32-opencl-nvidia 535.104.05-1
nvidia 535.104.05-7
nvidia-prime 1.0-4
nvidia-settings 535.104.05-1
nvidia-utils 535.104.05-1
opencl-nvidia 535.104.05-1

Udev rules as per Arch wiki:

❯ cat /etc/udev/rules.d/80-nvidia-pm.rules
# Enable runtime PM for NVIDIA VGA/3D controller devices on driver bind
ACTION=="bind", SUBSYSTEM=="pci", ATTR{vendor}=="0x10de", ATTR{class}=="0x030000", TEST=="power/control", ATTR{power/control}="auto"
ACTION=="bind", SUBSYSTEM=="pci", ATTR{vendor}=="0x10de", ATTR{class}=="0x030200", TEST=="power/control", ATTR{power/control}="auto"

# Disable runtime PM for NVIDIA VGA/3D controller devices on driver unbind
ACTION=="unbind", SUBSYSTEM=="pci", ATTR{vendor}=="0x10de", ATTR{class}=="0x030000", TEST=="power/control", ATTR{power/control}="on"
ACTION=="unbind", SUBSYSTEM=="pci", ATTR{vendor}=="0x10de", ATTR{class}=="0x030200", TEST=="power/control", ATTR{power/control}="on"

Nvidia module settings:

❯ cat /etc/modprobe.d/nvidia-pm.conf
options nvidia "NVreg_DynamicPowerManagement=0x02"

Note that this issue still occurs even without a desktop environment launched such as Gnome in my case. The below bug report was generated through the system console with all desktop environments disabled, and the issue is still present.

nvidia-bug-report.log.gz (323.0 KB)

Confirming this Bug on a DELL Precision 5750 (aka XPS 17), with a Quadro RTX 3000 Max-Q

Can you please share nvidia bug report.

Can you please share nvidia bug report.

I am not sure if I am in the same boat to be honest. With more recent driver versions (I cannot tell when it changed), the card seems to stay active most of the times. However, sometimes (No idea how to reproduce), it turns off and behaves as I expect.

Happy to share bug report, but don’t know how to do so.

Please run nvidia-bug-report.sh as root and attach the resulting nvidia-bug-report.log.gz file to your post.

nvidia-bug-report.log.gz (1.9 MB)