Valve will officially let you install SteamOS on other handhelds as soon as this April
That’s before the Legion Go S arrives.
You're viewing a single thread.
PC when?
The handhelds are just PCs, there is no distinction.
Are there handhelds with nvidia GPUs? That might be the big distinction.
Valve's added some Nvidia support to steamos, but Nvidia has not played well with Linux in general in my experience. Sounds like that's gotten a lot better recently though.
Yeah, nvidia started open sourcing their kernel interface. But it only works with new GPUs and not the millions of 1080s people still have. Plus the license of nvidia's closed source drivers (which are still needed) forbid them from being included in an OS image.
Small projects like Bazzite might get away with including them anyways but I bet Valve will have to make a deal with nvidia.
edit: With nvidia releasing their cloud gaming client on SteamOS I just now realised that that is probably a direct result of this kind of discussions.
Nvidia is fine and has always been fine, it just hasn't been ideal. I've used Nvidia GPUs on Linux for >10 years, and it has worked well.
The main issues are (and have always been):
If you use a release based distro and don't need to be on the bleeding edge (describes pretty much everyone), Nvidia is fine.
I switched to AMD a couple years ago because they offered better value and I needed an upgrade anyway, Linux compatibility was a nice value add.
Back before AMD bought ATI (and during the transition) nvidia was actually much better than ATI. For my previous desktop I specifically bought an nvidia card for the better Linux support.
I also wasn't so worried about my laptop having an nvidia GPU either. And then it kept crashing after being suspended to RAM. Stopped after switching to the integrated Intel graphics. Sadly nVidia doesn't care about old or laptop graphics.
Absolutely! I remember having ATI embedded graphics on my motherboard and it was more annoying than Nvidia's drivers. Nvidia didn't really change since then, it's just that AMD submitted their driver to the kernel, so newer software tends to work better with it.
dGPUs on laptops have always sucked on Linux, this isn't new, nor is it necessarily a problem specific to Nvidia. Graphics switching on Linux just isn't smooth, which is why I haven't bought a laptop with a dGPU since switching to Linux. I hear it works, I just don't see the point. Get a cheap laptop with an AMD APU and you can play casual games on it, and then build a cheap desktop PC with the savings and your experience will be much better. It'll probably cost a bit more at the start, but your laptop will last much longer and you can upgrade the desktop more cheaply.
Oh true, but it is based on arch so I figure it should be possible to install the nvidia drivers. They may get wiped each system update though
Just install Bazzite this way you have the actual Linux flexibility and an OS made for PC.
I highly recommend not using SteamOS on your PC, unless it's literally used as a console. Use any major distro instead.
I agree with your sentiment, but I disagree with your conclusion of using any major distro. If you've ever had to fix a corrupted package manager database caused by an in-place distro upgrade or had to install third-party package repositories to get access to up-to-date software, you'll understand where I'm coming from.
Beginners should start with something that either has bells and whistles included for out-of-the-box gaming, or comes with an easy way to un-fuck itself when you end up breaking something. It doesn't need to be Nix, but it probably shouldn't be Debian (which has a slow release cadence) or Ubuntu (because fuck Canonical and their "my way or the highway" approach to doing desktop OSes).
Debian is totally fine, why do you need a rapid update cycle? Everything you need is packaged with Steam. If for some reason you need something newer, you can always use whatever release is in testing at the time (use that release name, not "testing" itself) and you'll get newer packages with minimal risk of stability issues (a lot of people run testing).
There's really nothing special about newer packages for gaming. Once it's working, Debian will keep it that way.
I personally use openSUSE Tumbleweed because I like newer packages for other reasons (I use it for software development) and hate release upgrades because they take forever, but tons of people use stable distros without issue.
If you want some bells and whistles out of the box, I hear Bazzite is good. But any distro will work fine with Steam, and I'd assume Heroic and other launchers should also work fine on any distro they're packaged for.
Drivers can be an issue with recent hardware on Debian due to said slow release cadence. May not work as well on recent hardware.
In that specific case, yeah, maybe try a different distro. SteamOS will still be a worse option since Valve doesn't have any published update cadence.
But still stick to a major distro, like Fedora or Linux Mint. It's unlikely you'll actually run into issues on Debian though...
Also, kernel upgrades. Unless the user knows about and specifically opts to use Debian backports, they're going to be on the same kernel version until the next stable Debian release. It's not the end of the world to leave performance on the table, but some people are picky about getting their money out of their hardware.
Using backports and upgrading to a newer kernel is fine for someone familiar with Linux and confident enough to tinker and make at-your-own-risk changes. Having to do that can be offputting for newcomers, coming across as intimidating or unnecessarily complicated.
A newer kernel does not automatically offer more performance. In fact it could be the opposite if it includes workarounds for Intel's latest CPU security fuck-ups.
It's not so much the lack of a rapid update cycle as much as it's the pinned kernel version alongside the years-long pace of Debian's stable upgrade cycle.
That would be fine if the kernel didn't see much improvement over ~2 years of development, but there's constantly new stuff being added or optimized with every kernel release. It's just not much of a friendly introduction to Linux gaming for a newcomer to either have to pick between missing out on recent improvements, or diving into the intimidating realm of fiddling with packages and backported kernels—especially if they're not coming from a tech savvy background.
Most Linux users, including gamers, don't really benefit from improvements to Linux since most of it is drivers for hardware they don't have. Most userland software can be installed via flatpak or PPA (or other form of additional repository for your distro) if you really need something newer. But my understanding is that people (esp gamers) get annoyed more by stuff changing than missing out on new stuff.
The whole point of recommending a stable distro is to give the best chance of the person finding the help they need, as well as things not breaking randomly, and you get that with stable release distros. If the user knows enough to disregard that, they know what distro would be a better fit anyway.
I see your point, but i don't really agree that it doesn’t benefit gamers. In the 10 versions of kernels released since Bullseye released, we've seen improvements like the EEVDF allocator, sched_ext, the beginnings of ntsync, and an optimization to MDS mitigations for Intel processors. In a gaming-oriented distro, these would (ideally) be configured out of the box for the best gaming experience. Using a stable LTS distro, in contrast, would require manual tweaking and experimentation to achieve the same result.
But my understanding is that people (esp gamers) get annoyed more by stuff changing than missing out on new stuff.
They get annoyed when stuff visibly changes, like the desktop UI, userspace GUI programs, or noticable performance regressions.
Anecdotally, I have rarely seen the typical gamer complaining about or even noticing when something changes in technical stuff that they aren't directly interacting with. Nintendo actually does a good job creating situations where you can observe that behavior, funnily enough. When they release a new console with a different UI, non-casual gamers vocally bitch about it being worse than the previous generation. But when it comes to updates, the complaints are pretty much all about how it only changes the bad word list, doesn't have x in 2024, or how every update is just more "stability." Meanwhile, they have successfully done major rewrites and changes behind the scenes without anyone but the CFW and modding scene actually noticing it.
The whole point of recommending a stable distro is to give the best chance of the person finding the help they need, as well as things not breaking randomly, and you get that with stable release distros.
I agree that a stable distro will be more stable, but I don't agree that a stable distro is the best chance to get them help as a gaming newcomer. For newcomers in general, sure. But for gaming, it would be better to direct them towards a distro primarily focused on gaming, where they'll have a likeminded community. A popular stable distro will have more community resources available overall, but most of that is just going to end up becoming noise that makes it harder for them to find a solution for game-related problems.
Bullseye
Bookworm?
EEVDF allocator, sched_ext
Schedulers really don't matter much for gaming workloads.
Ntsync is a bit more exciting, I'll give you that. Looks like it's in testing, so you can update to trixie to get it. Running on the testing branch is usually fine, and it'll be pretty up to date until release freeze time, at which point it's best to sit tight until the new testing branch settles down (a point release or two is usually plenty).
manual tweaking
Tweaking is usually not worth it. The most I really do is up sending like the mmap limit if a game is struggling.
comparatively niche game-related problems.
I'm not really sure what types of issues you're referring to. Either you'll have OS questions, in which case the generic help is ideal, or you'll have game specific issues, in which case most distros will be extremely similar (e.g. proton db for game specific workarounds).
For people coming to Linux, I'll usually recommend sticking with Steam verified or playable games as well, since those should largely just work. Yeah, you might be able to eek out a few FPS or, more likely, framerate stability with some tweaks, but that's honestly not worth it to most.
If you want to tweak, you're probably also the type to want the "advanced" distros like Arch and will ignore my advice anyway. And that's fine, I use something else as well, but I'm going to stick to a more predictable experience for new users.
Sorry, yeah. Bookworm. I'm terrible at remembering codenames.
Schedulers really don't matter much for gaming workloads.
sched_ext is a bit more interesting. It's not a scheduler with a new set of knobs to turn, but actually more of a BPF-powered scheduler framework. Being able to swap out the scheduler depending on the game's needs could be a huge win, and there's already a couple of schedulers like LAVD (phoronix article) that are designed for improving the Linux gaming experience.
Tweaking is usually not worth it. The most I really do is up sending like the mmap limit if a game is struggling.
Trying to squeeze every bit of performance out, yeah. I don't want to waste my time trying to get a 1 FPS improvement either.
My concerns with tweaking are more along the lines of things like this happening. User reads about how to get better FPS, sees that some system package is really old, then tries to replace it with a newer one without knowing the risks.
That guy actually asked around and was convinced to just use Flatpak instead. But, suppose they didn't ask around and instead followed a guide or YouTube video, breaking something in the process. The typical self-proclaimed "gamer" isn't any more technologically savy than your average consumer, and probably isn't going to spend the time learning how to fix their mistakes. It's a more likely that they are just going to walk away with the impression that Linux sucks, and tell everyone else about how bad it was. And unfortunately, first impressions matter a lot when convincing other people to try things like ditching Windows for Linux. Avoiding stable distros is mitigation for that. If the package is already up to date or doesn't require going off the happy path to update, it leads to less bad choices.
My overall perspective is that if you give gamers a gift-wrapped install that immediately just works, they won't have a reason to improve or fix things until they're actually ready to. If 19/20 of their games work out of the box, they'll play those 19 games until they find the courage to try and learn how to make game #20 work. Or, maybe someone else will fix it for them first.
If 0 out of 20 games work and they need to follow a guide telling them to run various commands to install stuff and get started, they're going to be overconfident and underprepared when they step outside of the guardrails and do at-your-own-risk stuff they find online.
Either you'll have OS questions, in which case the generic help is ideal, or you'll have game specific issues, in which case most distros will be extremely similar (e.g. proton db for game specific workarounds).
I meant game-specific issues, yeah.
Even with ProtonDB, I don't think the typical coming-from-Windows gamer is going to end up having a simple experience.
With Windows, practically everyone is running the same system software. If some game is broken, they Google "windows 11 Cyberpunk crash" and get a bunch of suggestions and solutions relevant to them. I don't like Windows, but it does have the fact that there's a massive community of gamers running the same "distro" (or in this case, Windows version) going for it.
Even with ProtonDB, you have a bunch of people running different distros, different graphics driver implementations for a single graphics card vendor, and just a bunch of software fragmentation... A stable distro would probably be fine for newcomers if was tailored for gaming and had a critical mass of users, I'm sure. But in the current landscape, and for the reasons I listed in rant above, I simply don't think it would be a good idea to put someone on a stable distro when they're going to be surrounded with solutions from people on other distros describing how they fixed the problem by updating something or another.
For people coming to Linux, I'll usually recommend sticking with Steam verified or playable games as well, since those should largely just work.
Absolutely. If it Just Works™, it leads to a better first impressions.
gift-wrapped install
The ones I'm aware of are:
Bazzite is going to work differently from pretty much everything else, so I can only really recommend it if it's being treated like a console. Otherwise there's just way too much learning to figure out how to properly make changes.
Garuda being Arch based is a liability. I don't recommend any Arch spins until you're comfortable with both Arch and Linux, which certainly doesn't describe your average new Linux user. They're going to mess around with the AUR and they're going to break their system.
Nobara is probably the best option here, but it's maintained by one person who promises to keep working on it. It's also heavily modified, so getting help from the Fedora community may be iffy.
For Debian, you can get 90% of the benefit of the above by installing whatever release is in testing (trixie in this case, I don't recommend testing directly). Switch only if you run into issues nobody else seems to have.
Use what you want of course.
PCs got SteamOS way the hell back in 2013.
But it's the old version. 2.0 not 3.0.
Yes, Ted. That's the joke.
