They have a habit of not listing a large number of changes or fixes so likely not and we will have to test.
Update on driver 595.71.05 to my previous bug report:
- The Witcher 3: Wild Hunt still crashes - the issue appears to be triggered by enable Ray Tracing since the 590.44.01 branch.
- Indiana Jones and the Great Circle - HDR mode is reported as ânot supported by hardwareâ since 595.45.04 BETA.
RTX 5080, Ryzen 9800X3D, Fedora 44, Proton Experimental.
Nothing changed.
Forever Winter still freezes when drones are shot when running the 595 driver. Looks like the XID109 errors are going to continue for some time yet on UE5 games.
List of issues that still affect the 595 driver:
- Random black screens when playing intensive games, sometimes resulting in a partial GPU driver crash. It is sometimes possible to switch to a TTY, but attempting to switch back to a graphical session will crash the driver for good.
- Desktop memory usage is excessively high(over 2GB sometimes).
- A large amount of GPU Memory is unaccounted for in
nvidia-smi/ NVML. - Applications crash when VRAM is full(maybe fixed?).
- Shared RAM under Linux seems broken. Itâs almost like there is an incredibly small amount(256MB) and once thatâs gone things start crashing.
- DLSS3 / DLSS4 / Smooth Motion is broken. It causes massive input lag and micro stuttering.
- Gnome-shell / GDM can freeze on login from a fresh boot. Switching to a TTY and relogging in works around this.
- Ray Tracing is broken on a lot of games, resulting in an XID 109 or some other error.
- Some games show DLSS upscaling options but no frame generation(Portal 2: RTX). Maybe a Proton issue?
- Dragging the HDR slider under Gnome can cause a driver crash(maybe fixed?).
- DX12 games have a massive performance penalty.
- VRR causes backlight strobing/flickering.
- Shader cache doesnât save properly, forcing it to be computing again.
- Video decoding from a GTX 960 using Sunshine/Moonlight causes blocky artifacts.
- HDR is broken.
- Performance limiters are broken on Blackwell.
- PRAGMATA and other games have graphical glitches(maybe fixed?).
- Using a secondary dedicated GPU often fails with
[drm:__nv_drm_gem_nvkms_prime_dup [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000c00] Failed to import NVKMS memory to GEM object. This error can happen regardless of whether you usedxvk.deviceFilteror in-game display adapter settings. Other times games start but run at 5 FPS. - Display sometimes fails to wakeup from sleep unless switched to TTY.
Edit: flagged by trolls.
I had a run on the new driver coming from 590.48.1 to latest (595.58.03) exclusively for Death Stranding 2 as the new driver fixed a couple bugs for that game.
On the other hand, I started getting lower performances, which I didnât come to report though, but mainly this one when I was playing Sonic Frontiers (like the LODs were flipped):
A little update on the sharpness setting - I have now tried multiple live usb(endeavouros/cachyos/zorin) and in all it doesnât work.
Nvidia driver on endeavour/cachy was 590 - it takes the value doesnât prompt error no visual change
nvidia driver on zorin was 580 - it takes the value doesnât prompt error still no visual change and this is under x11 at that.
nvidia driver 595 on my install on two different devices desktop and laptop - still no error or issue with environment variable but again no visual change.
__GL_SHARPEN_ENABLE=1 works without issue here on MATE (x11) with the 595.58.03 drivers and RTX 3080. Arch Linux with cachyos 7.0 kernel.
tested with games running vkd3d-proton with proton experimental, as well as glxgears and vkcube.
game called Witchfire as an example
without sharpening:
with __GL_SHARPEN_ENABLE=1
Youâve posted this list multiple times now. Has anything changed, or are you engaging with any of the responses youâve gotten?
A list without context is very unhelpful.
It would be nice, if you could at least link to topics related to the issues that you listed.
Was having this error reliably in Elden Ring but it appears entirely resolved with default proton, max settings and max raytracing in new driver version 595.71.05
I think Iâve reported this one a while ago but received no answer and itâs still a problem on 595.71.05, so here it is again: applications fail to start on the iGPU when running in dGPU MUX mode. For example:
> vkcube --gpu_number 1
Selected WSI platform: wayland
Selected GPU 1: AMD Radeon 680M (RADV REMBRANDT), type: IntegratedGpu
[destroyed object]: error 7: importing the supplied dmabufs failed
(hangs without anything opening)
> vkcube --gpu_number 1 --wsi xcb
Selected GPU 1: AMD Radeon 680M (RADV REMBRANDT), type: IntegratedGpu
Segmentation fault (core dumped) vkcube --gpu_number 1 --wsi xcb
nvidia-bug-report.log.gz (897.7 KB)
Specific to AMD iGPU? Just tested on Intel, works fine.
Could be - I donât have a laptop with an intel iGPU/nvidia dGPU combo to test myself. Please note the âdGPU MUX modeâ part - everything is okay when Iâm using the iGPU as the primary GPU (aka the typical PRIME setup), but the issue arises when the dGPU is set as the primary one.
595.71.05 Ray Tracing option causing Wuthering Waves freeze/crash at 80% loading.
Itâs not the new driverâs fault since I have the same issue with my laptop by using the 595.58.03 driver.
So something is maybe broken in the v3.3 patch of the game for the linux users. (Iâm using cachyos but it happens on other distros as well)
When disabling raytracing manually in the INI file
Engine.ini localized in the
/home/<username>/.local/share/Steam/steamapps/common/Wuthering Waves/Client/Saved/Config/WindowsNoEditor/
by changing the r.RayTracing.LoadConfig value from 1 to 0, the game can run :
[/Script/Engine.RendererSettings]
r.RayTracing.LoadConfig=0
Just wanted to leave a feedback that 595 driver made Starfield unplayable on Linux Mint. I can start the game, I can walk up to a Ship Technician NPC and talk to him but when I try to enter ship builder the game freezes. It worked fine with driver 590.
It looks like version 595.71.05 introduces two new regressions that donât occur in earlier versions.
Reverting to 595.58.03 resolves both issues.
Unfortunately seeing freezing with 595.71.05 and CTX SWITCH TIMEOUT errors with Starfieldâs CreationKit.
NVRM: Xid (PCI:0000:01:00): 109, pid=44101, name=CreationKit.exe, channel 0x00000023, errorString CTX SWITCH TIMEOUT, Info 0x1fc02f
The problem is also tracked on version 595.71.05 on Debian 13
nvidia-bug-report.log.gz (660.4 KB)
While I understand that this list has some usefulness to have an overview of the current state of the driver, I donât think itâs worth posting this on every minor update.
Especially because:
- You didnât add the exact version number this list applies to. So why repeat on every minor bump?
- You didnât provide reference links for each claim which could otherwise help users understand the problem better and find a work-around.
- You didnât break it up into categories (e.g., model-specific, gaming, desktop), instead the order seems quite random.
- If you would track reference links, it would be easier to know if something is âmaybe fixedâ, or âfixedâ, or ânot fixedâ
- A section showing what has been fixed since the previous post of this message would be helpful, and if that section is empty, simply donât post it.
- Many claims simply state a fact without giving a clue of the âwhyâ (e.g. DX12 performance).
- Many claims state brokenness without giving a context of when (e.g., HDR is not broken for me).
- I think wording like âoftenâ or âa lotâ should be avoided, it is subjective.
Some points in the list could benefit from more precise wording:
NVIDIA drivers (and thatâs probably true for every dedicated GPU) donât have a concept of shared memory. While we know what is meant, I think a technically correct statement is still important to not misguide users who want to inspect or try to fix the problem. Technically, thereâs always dedicated device memory, and host memory. This even reflects in the Vulkan API. The driver has always been able to use host memory (what is probably meant by users when writing âshared memoryâ).
Whatâs true instead is that the NVIDIA driver doesnât support eviction to the extent which is needed to transparently swap allocations between host and device memory. And that eviction is also meant when some people write âshared memoryâ (blurring the line of what is supported, and what isnât, see above).
So the term is mixing two completely distinct concepts (accessing host memory, and evicting device memory), which in combination allow swapping between host and device memory (which is the third meaning that âshared memoryâ probably addresses).
So please, track these functions individually. âshared memoryâ is just the wrong term. I think what most people refer to as âshared memoryâ is actually a combination of host memory access and eviction behavior.
Improving these points may increase a positive reception of this list, and reduce the âtroll flaggingâ.
Iâm not asking to stop posting this - I think it could actually be very valuable if structured a bit more like a tracked issue list rather than a static list of complaints.
HTH.
Giving an abstract/simplified example, I thought of something like this:
Driver: 595.71.05
Model-specific:
- Secondary GPU issues (multi-GPU setups) â [link]
- VRR flickering on certain displays â [link]
Desktop:
- Gnome-shell / GDM freeze on login â [link], workaround: switch to TTY and relog
- HDR slider can crash driver â [link]
Gaming:
- Random black screens under high load â [link]
- DX12 performance penalty (likely related to descriptor heap / resource handling) â [link]
- Ray tracing crashes (XID 109 in some titles) â [link]
Memory / VRAM:
- High desktop VRAM usage (>2GB) â [link]
- Unaccounted VRAM in nvidia-smi â [link]
- Crashes when VRAM is full (status unclear) â [link]
Display:
- VRR flickering / backlight strobing â [link]
- HDR issues (context-dependent) â [link]
Other:
- Shader cache not persisting correctly â [link]
- Wake from sleep issues â [link]
Fixed (since previous version):
- [short list or empty â skip repost if empty]
Failed to upgrade kernel 6.12.85+deb13-amd64 from previous version 6.12.74+deb13+1-amd64
Debian 13.4 Trixie Cinnamon + NVIDIA 595.71.05 (installed from a â*runâ file downloaded from nvidia.com).
After running the commands âapt updateâ and âapt-get dist-upgradeâ, the following message appears:
/etc/kernel/postinst.d/dkms:
Sign command: /lib/modules/6.12.85+deb13-amd64/build/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pubAutoinstall of module nvidia/595.71.05 for kernel 6.12.85+deb13-amd64 (x86_64)
Building module(s)âŠ(bad exit status: 2)
Failed command:
âmakeâ -j20 NV_EXCLUDE_BUILD_MODULES=ââ KERNEL_UNAME=6.12.85+deb13-amd64 modulesError! Bad return status for module build on kernel: 6.12.85+deb13-amd64 (x86_64)
Consult /var/lib/dkms/nvidia/595.71.05/build/make.log for more information.
Finally:
Error! One or more modules failed to install during autoinstall.
Refer to previous errors for more information.
run-parts: /etc/kernel/postinst.d/dkms exited with return code 1
After reverting to version 595.58.03, the problem disappeared.
Kernel 6.12.85+deb13-amd64 installed correctly.
Best regards.