I’ve been an IT professional for 20 years now, but I’ve mainly dealt with Windows. I’ve worked with Linux servers through out the years, but never had Linux as a daily driver. And I decided it was time to change. I only had 2 requirements. One, I need to be able to use my Nvidia 3080 ti for local LLM and I need to be able to RDP with multiple screens to my work laptop running Windows 10.
My hope was to be able to get this all working and create some articles on how I did it to hopefully inspire/guide others. Unfortunately, I was not successful.
I started out with Ubuntu 22.04 and I could not get the live CD to boot. After some searching, I figured out I had to go in a turn off ACPI in boot loader. After that I was able to install Ubuntu side by side with Windows 11, but the boot loader errored out at the end of the install and Ubuntu would not boot.
Okay, back into Windows to download the boot loader fixer and boot to that. Alright, I’m finally able to get into Ubuntu, but I only have 1 of my 4 monitors working. Install the NVIDIA-SMI and reboot. All my monitors work now, but my network card is now broken.
Follow instructions on my phone to reinstall the linux-modules-extra package. Back into Windows to download that because, you know, no network connections. Reinstall the package, it doesn’t work. Go into advanced recovery, try restoring packages, nothing is working. I can either get my monitors to work or my network card. Never both at the same time.
I give up and decide it’s time to try out Fedora. The install process is much smoother. I boot up 3 of 4 monitors work. I find a great post on installing Nvidia drivers and CUDA. After doing that and rebooting, I have all 4 monitors and networking, woohoo!
Now, let’s test RDP. Install FreeRDP run with /multimon, and the screen for each remote window is shifted 1/3 of the way to the left. Strange. Do a little looking online, find an Issue on GitHub about how it is based on the primary monitor. Long story short, I can’t use multiple monitor RDP because I have different resolution monitors and they are stacked 2x2 instead of all in a row. Trust me I tried every combination I could think of.
Someone suggested using the nightly build because they have been working on this issue. Okay, I try that out and it fails to install because of a missing dependency. Apparently, there is a pull request from December to fix this on Fedora installs, but it hasn’t been merged. So, I would need to compile that specific branch myself.
At this point, I’m just so sick of every little thing being a huge struggle, I reboot and go back into Windows. I still have Fedora on there, but who would have thought something that sounds as simple as wanting to RDP across 4 monitors would be so damn difficult.
I’m not saying any of this to bag on Linux. It’s more of a discussion topic on, yes, I agree that there needs to be more adoption on Linux, but if someone with 20 years of IT experience gets this feed up with it, imagine how your average user would feel.
Of course if anyone has any recommendation on getting my RDP working, I’m all ears on that too.
I read the first paragraph and saw your prerequisites included working with nvidia.
That is a non-starter, right there. You can blame Linux for a whole lot of little flaws, but most of the blame should go to your hardware vendor for providing shitty support for Linux.
Isn’t most of the AI training work in the world done on Linux using Nvidia GPUs (in the cloud)? I guess it’s a different use case…
Probably dedicated vector/tensor coprocessors these days - which don’t have to work with your monitor layout or desktop setup!
And it also sucks in the cloud. Depending on the scenario there might not be many alternatives, though. CUDA is pretty much the standard in machine learning.
ROCm has hints of adoption, but it’s only just getting started.
Having spent the weekend trying to get it working on WSL2 for lulz, I can honestly say it’s just not there yet. Most of the issue is that AMD cards aren’t exposed properly through WSL, but it was worth a shot.
Sure. But by the amount of adoption CUDA has, and the amount of GPUs / AI accelerators NVidia pumps out and into the datacenters of the world… AMD better hurry (and deliver an excellent product/ecosystem) or they won’t be part of the AI boom.
thats guiless.
x11 and wayland support (what matters for DESKTOP use) are complete garbage
Popos has out-of-the-box nvidia support that works great
Works with CUDA and RDPing on a 2x2 monitor grid?
System76 (who makes popos) has their own CUDA repo for their NVIDIA implementation, but I don’t think it’s installed by default. So there’s a tweaked version to work on popos, but I’ve never tried it. From some cursory googling, it doesn’t seem to be too complicated to set up.
deleted by creator
Yeah it’s really weird. I have done what OP has done for a while now on an Nvidia GPU and Pop!_OS with KDE and have had 0 issues. I don’t use a 2x2 grid though. Can that really be the issue?
I agree. The majority of my issues come down to the manufacturers. I even updated my BIOS to see if it would help with the ACPI issues, but no luck. Motherboard is 3 years old, so it’s not like I’m trying this on brand new hardware either.
deleted by creator
Its not totally wrong honestly. Nvidia is kind of bad and you can get a used AMD GPU for $100 bucks.
If you are using Nvidia use Linux mint or Pop os.
deleted by creator
Microsoft is free to publish minimum requirements for Windows (TPM 2.0 for Windows 11, for instance), but you don’t have that in Linux. You are free to throw it at any hardware you want, and it will mostly work out of the box.
But that depends on companies and volunteers working on the hardware support. Intel and AMD provide good support for their hardware. NVidia does not. You should act accordingly, either buying supported hardware or sticking to software that supports your hardware (Windows or Mac).
deleted by creator
Having a nVidia GPU does not stop you from running Linux, it just makes it more painful depending on what you’re trying to achieve due to nVidia’s poor Linux support.
I merely suggest that one should use the appropriate tool for the job or endure the consequences. Blaming the tools achieves nothing.
deleted by creator
“Nvidia, fuck you!” - Linus Torvalds
Nvidia is by far the most popular dedicated GPU manufacturer out there. If distros can’t figure out how to make it “just work” then Linux will never take off outside of the nerd market.
It isn’t something that is in the distro vendors control. Nvidia do not disclose programming info for their chipsets. They distribute an unreliable proprietry driver that is obfuscated to hell so that noone can help out fixing their problems.
If you use an AMD card it will probably work fine in Windows and Linux. If you use an Nvidia card you are choosing to run windows or have a bad time in Linux.
It doesn’t matter whose at fault at the end of the day. If it doesn’t just work™️ the average joe will never use it.
The open source vs closed source drivers is a whole separate issue too, and only makes things worse for the noobs.
I’m not the PR department for desktop Linux for everyone man.
People who only have Windows experience see an Nvidia card that is premium priced product with a premium experience and think that this will translate to a Linux environment, it does not. I’ve been using Linux for like 27 years now and that was my opinion until a couple of years ago.
Hopefully the folks that might read this thread ( like the OP 20 year IT veteran ) can take away that Nvidia cards in linux are the troublesome / subpar choice and are only going to get worse going forwards ( because of the Wayland migration that Nvidia are ignoring ).
That’s a great explanation. Knowing next to nothing about Nvidia and Linux, the original comment made it sound as though Linux is just wildly incompetent.
Your comment makes it sound like Nvidia is the graphics version of apple.
Oh yeah. That video of Linus Torvalds giving Nvidia the finger linked elsewhere in this thread was the result of a ton of frustration around them hiding programming info. They also popularised a dodgy system of LGPL’ing a shim which acted as the licence go-between the kernel driver API ( drivers are supposed to be GPL’d ) and their proprietary obfuscated code.
Despite that, I’m not really that anti them as a company. For me, the pragmatic reality is that spending a few hundred bucks on a Radeon is so much better than wasting hours performing arcane acts of fault finding and trial and error.
The problem is Nvidia’s drivers, not the distros.
You may as well be saying distros really need to get their shit together on releasing Photoshop for Linux
If someone with no experience installs Linux on their machine, and has to spend 20 hours fixing all of the problems they’re not going to stick with Linux. It doesn’t matter which distro it is, they’re just going to say Linux sucks and never use it again.
There’s a pretty big difference between trying to run software for X OS on Y OS, and trying to just make your computer do basic tasks. The average person doesn’t know that Nvidia are a bunch of assholes, nor do they care.
I know.
But there’s nothing that can realistically be done about it until Nvidia stops being dickheads.
Distros can’t constantly hop about putting out fires that Nvidia starts, and neglect the other work they need to do.
Even when they do that, it doesn’t work anyway. It’s still buggy, systems still break. It really is only Nvidia who can fix their shit drivers, unless the nouveau team make an alternative that’s superior to Nvidia’s proprietary drivers.
And nah, there’s no difference between my Nvidia/Photoshop example. None whatsoever.