Nvidia 595.45 are a mess for me, how to handle it?

Issues & Assistance
1

I’m relatively new to this situation with cachy, I never had to directly handle an nvidia driver regression, the 595.45 made most of my games unplayable, I used a snapshot to rollback the system to 590.48 drivers, now I can’t update to avoid problems.

Problem: it’s most likely going to take a long while to fix the issues and the 590.48 are great for me, how do I make it so that the rest of the system gets the updates it needs and meanwhile avoiding the 595.45? no risk of kernel drivers mismatch?
How do you guys handle this kind of situations on cachyOS? thank you.

2

downgrade can be used to download the old versions and it can also add the packages to pacman ignore config so they are skipped in upgrade

3

595’s are some of the best drivers I’ve seen so far. What’s your issues?

4

Crashes in some loading screens for some games (like the recent toxic commando) or performance issues with space marines 2 (which is not recent at all). I’m glad it’s working for you but for me doesn’t seems the case. 590 = no issues.

I don’t know if this can help identify the problem, but this is the error I get:
Error

5

I’m also getting constat issues with series 595 on Witcher 3 (VK_ERROR_DEVICE_LOST).

6

I did this to downgrade:
sudo downgrade nvidia-utils lib32-nvidia-utils opencl-nvidia nvidia-open-dkms lib32-opencl-nvidia
Now everything seems to work as intended… this doesnt risk to cause some mismatches between the kernel and the drivers in future updates?

7

Personally I have reversed to stable branch drivers

sudo pacman -S linux-cachyos-headers nvidia-580xx-dkms nvidia-580xx-utils lib32-nvidia-580xx-utils nvidia-580xx-settings opencl-nvidia-580xx lib32-opencl-nvidia-580xx

All seems to work for me.

8

I’ve followed your pacman command for now because I can’t see how to select the 590-open drivers. Different drivers have the same name in different repos so I would need to read up on selecting them by repo.

But 580 seems to have mostly fixed the issue I was having, which was that I found my SSD was constantly above 80 degrees Celcius because the error logs were getting hammered by issues relating to 595. I was getting the following message approximately 100x every second and 25mb/s was being sent to disk.

kwin_wayland[1444]: 0x500: GL_INVALID_ENUM error generated. Invalid <face>.
kwin_wayland[1444]: Invalid framebuffer status:  "GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT"
9

I’m a 4050 mobile user with proprietary drivers and 595.58.03 introduced an issue where it’d randomly freeze and I have to hard restart and it seems that it’s carried on to 595.71.05, the only difference there is that instead of a freeze, I get a blank screen and I can enter TTY.

Today I can still downgrade to 590.48.01 but now I have to use the LTS kernel to be able to use that driver.

Is there a tracker for this? I don’t see any other threads with people complaining and I’m afraid the issue might not get fixed

10

610.43 solved my issue