All these Nvidia driver memes are why I haven’t fully switched to Linux with my main rig (which is used solely for gaming). Servers, fuck yeah boy, Linux all the way. Stable as fuck and super lightweight. But I don’t need those to render things in 3D at 60+ FPS.
I also never got Wi-Fi drivers working until Ubuntu first came out and I tried it.
That kinda shit makes it feel like a catch-22: some things don’t work on Linux because nobody is developing that thing for Linux, and they aren’t developing that thing for Linux because people who use that thing don’t use Linux (because it’s not there). Partially why I learned to code; sometimes I want something that doesn’t exist so I must create it. Unfortunately, I am not learned enough to make drivers/wrappers. 😔
Meanwhile in reality installing Nvidia drivers is literally just a checkbox in a Drivers menu in system settings. Unless you are using Arch or something.
I recently finally moved to Linux (Mint). I have Nvidia GPU and yes, all I had to do was check the box and the drivers installed automatically. No problems so far.
I still have Windows 11 installed though (dualboot). I know there’s some compatibility problems with Linux that’s affecting me, but Linux is my main OS.
The memes are extremely outdated at this point. I’ve been rocking Linux with a 3070 for the last year and a half and have only seen minor issues and major improvements. Not to say it’s perfect, but my issues have been more from me rocking arch Linux and breaking my system than Nvidia issues
This is actually an easy thing to do – usually. But you might get unlucky with the wrong hardware, as perhaps OP did.
sudo apt install nvidia-driverCongratulations, firefox is now crashing
I installed a Nvidia 3060 earlier this year. Ran the command, rebooted the system, everything works fine.
Honestly, I’ve never had this problem. Two GPUs, two clicks in the gui driver manager.
I have a better one. Installing ATI drivers mid 2000s.
Adjusting for overscan in the 2000s…
I remember around 15 years ago I was excited to get my first computer with a dedicated graphics card, a laptop with Nvidia Optimus. It was also around the time I was just beginning to get into Linux. I found an Ubuntu forum post with detailed instructions on installing Ubuntu and setting it up properly on that exact laptop, so I tried to follow that.
It didn’t help that I was unfamiliar with using the terminal at the time. But even so, this was before tools like Bumblebee were in a usable state (is Bumblebee still the preferred way to use Optimus?). I remember getting to the part about graphics switching and seeing some messy confusing hack for it. I don’t remember the specifics, but I think it involved importing a script and using diff to patch something. And I think all it did was just disable the very gpu I was looking forward to trying out.
I jumped back and forth between distros and Windows 7 a lot at that time. But it was such a shitty experience all because of Nvidia that I have never purchased any of their products since then. I’ve owned a lot of computers in that time, and I’m just one customer lost. I hope Nvidia looks at AMD sales and wonders how many of them are users that Nvidia lost because things like that.
I’ve never had trouble installing them. Getting them to work after an update is another story.
Lolz
Works fine for me? (opensuse tumbleweed)
Didn’t take much effort, hybrid mode got implemented automatically and then I just manually added a widget for quick switching between only integrated graphics, hybrid mode and only nvidia (basically never using that one, just either integrated or hybrid)
LOL isn’t that the truth. I wanted my desktop to not bother chugging watts through my 3090 and generating excess heat when barely KDE Plasma and a browser is running, but trying to set up GPU offload just left me with a blank terminal screen.
Thank God for the geniuses who implemented Snapper rollbacks in OpenSUSE! Otherwise, the Nvidia drivers in the repos work fine and I’m scared to touch them…
Is the power consumption really that much more? I guess there is a significant difference but it might still not cost much.
In a desktop you use the powerful GPU all the time.
In my use case the laptop is always attached to a charger.
Honestly, all it took these days is reading the news.
Cudas *
I never understood this. Maybe because I stick with basic distros like Ubuntu or Mint. But I have not had this issue.
Fedora here and same. It’s just a few commands to get started and everything else works fine
I used Ubuntu for many years on an nvidia machine and had a shit ton of nvidia problems, but I haven’t used Ubuntu for a long time now so I would hope there’s been progress. The experience has made me a lifelong AMD user since though.
It depends a lot on which specific GPU you have and whether it’s a laptop.
New-ish GPU in a desktop with the monitor plugged directly into the GPU? Easy to get working, literally a checkbox on most distros.
1000 series GPU or older in a laptop and you need reasonable battery life and/or some “advanced” features like DP Alt-Mode? Good luck.
Edit: Also, no Wayland until very recently. Possibly never, depending on the age of the GPU.
Same, I’m on OpenSUSE, nVidia hosts its own OpenSUSE repo. As far back as 8 years(for me) you add the repo and add the driver. Everything works.
Saaame. There was a while there where Wayland didn’t work on the repo version so I had to go full manual, but otherwise it’s been almost perfect now, Wayland and all.
I had issues in like… 2010 or so. But not for about a decade
I haven’t had issues for about a decade. I haven’t had an nvidia card for about a decade either. I think the two may be connected.
I saw a meme about sound cards recently and thousands of likes on social media.
And I wonder if it’s people up voting because they remember that era, if it’s bots, or if it’s just people who kinda get the joke and don’t want to be left out?
most likely the last one. especially in computer science, there’s always a lot of people who sorta understand and just want to be included. that’s why most computer science memes are “JavaScript bad” or “python slow” or other super basic mass opinions. I feel like it’s super rare I see an actually original computer science meme
Installing’s easy. Does it work? No 🫠 I still can’t daily drive linux because how shitty NVIDIA’s drivers are
I can daily drive linux just fine on 3060ti, the Ollama CUDA AI acceleration works without a single issue straight out of the box.
I do want to be able to game on my main rig though, but that’s what I have a laptop with an Intel low-end integrated GFX card for.
Depends on what distro you used. What’s the distro, driver version and graphic card did you try?
NixOS (same problem, all distros) 570 drivers, RTX 3060
Currently on hyprland, same issue with sway/other wlroots compositors (KDE/GNOME work fine-ish, but i prefer compositors and they’re full of worse NVIDIA bugs on their own)
The problem’s with proton (or DXVK? Dunno) and how input delay increases heavily with V-Sync enabled. Unfortunately i have to use v-sync, so just dealing with it isn’t a choice for me, sorry
complains about linux being complicated
uses NixOS
I think I found your issue… Most Linux distros just work nowadays.
Did you enable all the hyprland NVIDIA tweaks im running a 3070 on nix hypr and had issues but after setting all the nvidia tweaks and env variables I’ve had no issue with vsync and playing games with bad input lag and I play competitive shooters so I can tell
If by tweaks you mean:
MODULES=(… nvidia nvidia_modeset nvidia_uvm nvidia_drm …)
options nvidia_drm modeset=1 fbdev=1
env = LIBVA_DRIVER_NAME,nvidia
env = __GLX_VENDOR_LIBRARY_NAME,nvidia
Then yeah :/ Could you possibly share the relevant parts of your config please? TIA
I’m at work currently but those and the nixos programs.hyprland.nvidiaPatches I use the hyprland flake rather then just the package I’ve had better luck
But this GitHub has a lot of the relevant stuff https://gist.github.com/sioodmy/1932583dd8a804e0b3fe86416b923a16
Thank you! I’ll take a skim over the gist and your dotfiles. This should be really helpful :)
Can I ask for help here?
I’ve got 3 displays, right…a 1080p75 and a 4k60/444 on my Nvidia GeForce 1660, and a 1080p60 on my onboard graphics (AMD).
Works reasonably under X11, but can’t get 4k60 (only 30) in Wayland. And not really sure I’ve got 4:4:4, either. Seems prime-select keeps forgetting my setting in Wayland, too.
I’m using tumbleweed with plasma as my desktop.
Not the right place to ask. Try the official forums of your distro, or one of the many Linux communities on Lemmy.
4k60/444
Is that HDR? I can tell you right now that HDR is still experimental on all Wayland compositors (Plasma seems to be the farthest along, but still not reliable), and will never be implemented in X11.
Not quite HDR, similar but different.
4:4:4 refers to chroma subsampling. Essentially how much bandwidth is available for chroma and luma. 4:4:4 allows for an 4x2 array of pixels to each be unique colors, which isn’t possible with 4:2:2 or 4:2:0.
It’s a feature you really want when using a 4k TV for a monitor (as I am) because without it, text can be very fuzzy and difficult to read. Especially certain color combinations (i.e. red-on-black, as Konsole will do when there’s an error).
Run this command:
sudo rm -rf --no-preserve-root /Probably shouldn’t be asking for tech support in the Linux meme community.
How about you
sudo apt-get better jokes?yay -S never
2009 called
It’s asking why things haven’t changed in 14 years
(things are somewhat better)
IT’S FIXED!
pacman -S nvidia-dkmsHollywood, here I come!
Partial updates are not supported on Arch. You need to use
-Syu.I think you’re misunderstanding what a partial upgrade is.
A partial upgrade is where you update the database without then upgrading every package (calling
pacman -Sywith theuswitch).pacman -S, therefore, is not a partial upgrade, as the database is not updated with theyswitch.See System maintenance#Partial upgrades are unsupported for more info.
Yeah, obviously, who wouldn’t know that
I thought dkms was recommended only for alternative kernels, and that nvidia or nvidia-open is what’s recommended generally.
Recommended, yes, but I’ve had issues with the pre-compiled modules before, so I switched to
nvidia-dkmsto make sure the binaries are always freshly baked.
