what about implementing in this package this commit:
that is for sleep to work well with systemd 256, see this:
Git Clone URL: | https://aur.archlinux.org/nvidia-470xx-utils.git (read-only, click to copy) |
---|---|
Package Base: | nvidia-470xx-utils |
Description: | OpenCL implemention for NVIDIA |
Upstream URL: | http://www.nvidia.com/ |
Keywords: | driver nvidia video |
Licenses: | custom |
Conflicts: | opencl-nvidia |
Provides: | opencl-driver, opencl-nvidia |
Submitter: | jonathon |
Maintainer: | Sinyria (cysp74, SoftExpert) |
Last Packager: | SoftExpert |
Votes: | 103 |
Popularity: | 1.94 |
First Submitted: | 2021-10-31 00:50 (UTC) |
Last Updated: | 2025-02-03 19:42 (UTC) |
« First ‹ Previous 1 .. 4 5 6 7 8 9 10 11 12 13 14 .. 44 Next › Last »
what about implementing in this package this commit:
that is for sleep to work well with systemd 256, see this:
@marco.righi: The possibility of fixing something in the driver is almost zero.
Newer libraries will start misbehaving with old drivers at some point in time; there could be / are bugs introduced by the latest libraries that might trigger previously unknown / unseen bugs in the driver, or it's just that some implementations of the libraries expect the driver to behave in a way which corresponds to a newer version which will never be available with the 470xx series.
Currently we are lucky to still get some updates - which are only meant to keep the driver compatible with the latest kernel changes. Though, I'm kind of worried, since the Nvidia forum page that summarizes the latest versions has not been updated for our current version.
As previously recommended, try to change a single parameter at a time and test the effect it has. It's long, but it will provide the best result you could get with your own configuration - and you will learn new things from it.
The upgrade does not resolve the problem @marco.righi/2024-05-28 13:58
@SinClaus, @a821 : sorry about that, I forgot to include the patches as well. I pushed a new release.
CRC error on install, installation filed!
@gbr: thanks for the heads-up; the Current graphics driver releases page is not yet updated. I'm currently building the new version to see if the patches still apply ...
edit:
This new version still requires the previous patches. It seems that it comes with a single fix: Fixed a bug that could cause the X server to crash when graphics applications requested single-buffered drawables while certain features (such as Vulkan sharpening) are enabled.
@marco.righi I'm not 100% sure, but I feel that your configuration mixes non-nvidia settings with nvidia-specific settings and, as a result, you might get interference between two different kinds of modules, fighting over the same ressources.
Clearly, the best answer you would get comes from meticulous experimentation (changing a single option at a time and testing the result); one of the reasosns is that the hardware configuration you have is specific to your system and what is good for one set of hardware components might not be optimal for another.
Nvidia drivers are kind of picky - it matters if you mix Mesa modules with them (the consensus would be don't
).
Maybe someone has an identical harware configuration and could provide a more precise answer.
P.S. My config:
Graphics:
Device-1: NVIDIA GK104GLM [Quadro K3000M] driver: nvidia v: 470.239.06
Device-2: Sunplus Innovation HP HD Webcam [Fixed] driver: uvcvideo
type: USB
Display: server: X.Org v: 21.1.13 with: Xwayland v: 24.1.0 driver: X:
loaded: N/A failed: nvidia gpu: nvidia,nvidia-nvswitch resolution:
1: 2560x1440 2: 1920x1080~60Hz
API: EGL v: 1.5 drivers: nvidia platforms: x11
API: OpenGL v: 4.6.0 vendor: nvidia v: 470.239.06
renderer: Quadro K3000M/PCIe/SSE2
API: Vulkan v: 1.3.279 drivers: nvidia surfaces: xcb,xlib
And this is the relevant part of my xorg.conf
:
Section "Device"
Identifier "Device0"
Driver "nvidia"
VendorName "NVIDIA Corporation"
BoardName "Quadro K3000M"
Option "TripleBuffer" "On"
Option "AllowIndirectGLXProtocol" "off"
Option "ConnectToAcpid" "Off"
Option "DPI" "108x108"
Option "RegistryDwords" "EnableBrightnessControl=1"
Option "Backlight" "nvidia_backlight"
EndSection
Section "Screen"
Identifier "Screen0"
Device "Device0"
Monitor "Monitor0"
# 24 for 8-bit or 30 for 10-bit
DefaultDepth 24
Option "Stereo" "0"
Option "nvidiaXineramaInfoOrder" "DFP-6, DFP-0"
Option "metamodes" "LVDS-0: nvidia-auto-select +2560+0 {ForceCompositionPipeline=On}, DP-5: 2560x1440_75 +0+0 {ForceCompositionPipeline=On}"
Option "SLI" "Off"
Option "MultiGPU" "Off"
Option "BaseMosaic" "off"
Option "TripleBuffer" "On"
Option "AllowIndirectGLXProtocol" "off"
Option "ForceCompositionPipeline" "on"
Option "Coolbits" "8"
Option "ColorSpace" "RGB"
Option "ColorRange" "Full"
SubSection "Display"
Depth 24
EndSubSection
EndSection
Section "Extensions"
Option "COMPOSITE" "Enable"
EndSection
@Maniaxx sorry for the following question.
Which is the solution that guarantee the best performance? My solution or GSK_RENDERER=gl
referred by your link?
The referred solution GSK_RENDERER=gl
seems not working in every case...
Thanks in advance
Pinned Comments
SoftExpert commented on 2025-02-09 10:20 (UTC)
For the beginners in need of assistance, first take a look at this recommended guide https://github.com/korvahannu/arch-nvidia-drivers-installation-guide - maybe it helps to solve the issues your are having with this version of the driver.