Install Steam
login
|
language
简体中文 (Simplified Chinese)
繁體中文 (Traditional Chinese)
日本語 (Japanese)
한국어 (Korean)
ไทย (Thai)
Български (Bulgarian)
Čeština (Czech)
Dansk (Danish)
Deutsch (German)
Español - España (Spanish - Spain)
Español - Latinoamérica (Spanish - Latin America)
Ελληνικά (Greek)
Français (French)
Italiano (Italian)
Bahasa Indonesia (Indonesian)
Magyar (Hungarian)
Nederlands (Dutch)
Norsk (Norwegian)
Polski (Polish)
Português (Portuguese - Portugal)
Português - Brasil (Portuguese - Brazil)
Română (Romanian)
Русский (Russian)
Suomi (Finnish)
Svenska (Swedish)
Türkçe (Turkish)
Tiếng Việt (Vietnamese)
Українська (Ukrainian)
Report a translation problem
With many good Linux distros to choose from there really is no need or point to SteamOS Desktop Edition for Desktops/Laptops
Most of the drivers that you need are included in the Linux kernel itself. When installing Steam on Linux, there are no drivers that have to be manually installed. The drivers are installed during the install of the OS, and are upgraded with each kernel upgrade (Edit: and any dependencies and services/daemons are installed asdeps with the Steam desktop application package from the distro's repos). The only time drivers have ever been a problem for me were when I installed a distro that only shipped FOSS by default and required non-free software repositories to be enabled explicitly. Debian and Fedora used to be this way, and even on Linux Mint several years ago when I first started, I had to ask the community for help with installing a driver for wifi. But on distros that include non-free software out of the box like Arch Linux, which is what SteamOS is based on, I have never needed to manually install drivers.
[Edit: I know that Nvidia drivers have been an exception to this point for a long time because they have been notoriously bad on Linux, but the situation with Nvidia support has improved greatly and is continuing to improve. In fact, I think that having a desktop SteamOS might drive even more rapid improvement of Nvidia on Linux.]
The reason it seems like SteamOS lacks driver support is because Valve modified Arch Linux and designed SteamOS specifically for the Steam Deck, so it expects native Steam Deck hardware and isn't prepared for anything else. The SteamOS image even lacks support for standard display dimensions by default. It expects the native display to have the dimensions of a Steam Deck. You can see this in LTT's video at timestamp 12:14 where he mentions that games launch at the Steam Deck's native 1280x800 res instead of the display's true native res, and requires manual intervention to fix this. This isn't because Linux doesn't support standard desktop resolutions OOTB, it's because Valve designed SteamOS explicitly for the Steam Deck and didn't expect users to be installing it on standard desktop hardware.
I agree that there is no strict need for a SteamOS desktop edition, but I respectfully disagree that there is no point. The point would be to create an experience that is inviting to users who are seeking to get away from Windows. Many of them are already apprehensive about trying Linux because of the misguided and misunderstood image that many of them have about Linux in general. I argue that the point of having an option that is officially developed, maintained, and supported by Valve is to provide an option that is more inviting to those users because although they're hesitant about trying Linux, at least they trust Valve.
But what's in it for Valve? How will they get their money back?
It's basically a live service that will continuously cost them money and will create new problems they never had.
That's an excellent point, and I don't have a good answer for that problem. The best answer I have at the moment is to do something similar to what Canonical does with Ubuntu. Ubuntu is free and access to community support is free, and sometimes Canonical developers participate in forum discussions, but users have the option to pay for dedicated professional support directly from Canonical. Perhaps a similar model would work for SteamOS, perhaps not. Definitely an issue that would need further discussion to find an appropriate solution.
Edit: It also occurs to me that when you purchase a Steam Deck, you're purchasing the hardware AND the software. It might also be an option to put desktop SteamOS behind a paywall the way Red Hat does with RHEL, but at a lower price than the full cost of a Steam Deck since you would only be purchasing the software, not the hardware. To be clear, I DON'T think this is a good solution considering the Steam desktop application is already provided free of charge (and SteamOS is really just Arch Linux configured to be a bootloader for the Steam desktop app in BPM), but I think we would be naive to blindly assume that Valve has not or would not consider this option for distribution of SteamOS desktop installer images.
That's exactly my point. SteamOS is not designed to be installed on a laptop or desktop. The recovery image that Valve provides assumes that the user is installing on a Steam Deck or (now) other supported hardware such as the ROG Ally and the Lenovo Legion Go. The settings for those devices are essentially hard-coded into the recovery image, so even when you attempt to install on a laptop or desktop, the recovery image still assumes (albeit incorrectly) that you're installing on a supported handheld device, and behaves as such. SteamOS itself still has support for other display dimensions/configurations, but that is intended to be configured after installation for use cases where users are connecting their handheld device to an external display via HDMI.
As for support for more GPUs and CPUs, that is configured at the kernel level. Even though SteamOS is based on Arch Linux, it still uses a custom kernel that was customized with handheld hardware in mind. The Linux kernel is modular, so certain pieces can be added or removed according to the intended or expected use case. For example, if SteamOS is intended to only ever be installed on Steam Decks, there is no need for the kernel to have support for other CPUs and GPUs that are not installed on the Steam Deck, so they could remove those kernel modules to reduce the size of the kernel to reduce bloat. Trimming the fat, so to speak.
Obviously some of those kernel modules still exist because SteamOS can be installed on devices with other CPUs, but GPU support is a slightly different matter. The Linux kernel does have support for Nvidia GPUs, but there are other Nvidia drivers that often need to be installed after the OS has been installed. However, the build versions of those drivers need to be compatible with the build version of Steam and Proton that ship with a given distro. These build versions don't always line up, so manual intervention is required to get Nvidia drivers working correctly. This has been one of the hang-ups with running Linux on Nvidia hardware in general, not just gaming. Again, the situation here has improved and is still improving, but you're absolutely correct that getting things on Linux to work on Nvidia hardware can be a PITA. It's an unfortunate reality, but again, a lot of work has been done, especially by the Linux community contibuting to the open-source Nvidia drivers.
This is factually incorrect. While there are some core drivers compiled directly into the linux kernel; the majority of drivers are compiled as kernel modules, not directly in the kernel itself. You are making a lot of unsupported statements of fact that are not true or are simply assumptions being made by you without basis; such as the following.
This is wholly assumption on your part as to what Valve has done with the SteamOS 3 kernel. Also, you keep conflating kernel modules with the kernel itself.
SteamOS 3 is a fork of Arch as you've noted; everything about SteamOS other than some portions of things such as the Steam client, and some of the functionality for Game mode, quick resume, and mode switching is already open source. They are contributing directly to the open source features they are leveraging for SteamOS 3 and they are developing gamescope as open source. They are already doing what you are trying to suggest.
Lastly, Valve has already said that their intention is to eventually release SteamOS 3 as a generally available OS with an installer to be installed on typical commodity hardware. They are just operating on Valve time as per normal. Their first order of business is refining the experience for SteamDeck, then to extend it to similar 3rd party platforms, as well as building it out for Deckard (their next gen VR HMD). All of those things from Valve's perspective take precedent over building an installer and validating for general hardware use.
I appreciate this feedback. I admit that there are things that I misunderstand about the nuances of this topic, and indeed some of what I said is speculative. I hope you don't mind if I pick your brain a little bit here about the things I've gotten wrong. I'm still searching and doing my own independent research to learn more about this topic, so please be patient with me.
I also want to point out that the intent of my post was not to dive head-first into the technicalities of building SteamOS for the desktop, but rather to make a case for why I think it should exist. Naturally this means that technical aspects of this topic will arise, and I'm completely open to criticism and correction.
Based on the searching that I've already done up until this point, everything that I've found about SteamOS being released for the desktop has revolved around the same two points: 1. SteamOS on desktop is pointless, just install Linux and use the native Steam client, and 2. Valve has made no official statements regarding releasing SteamOS for desktops. I actually have a tab open right now doing more research on this topic as we speak, so I'll search for information about this as well, but if you have any weblinks at the ready that you can share, I'd love to check them out.
You said that it's factually incorrect that most of the drivers you need are included in the Linux kernel. You're right, I must have conflated two different facts here. I'm aware that certain hardware divers are configured as modules to the kernel, not necessarily directly included in the upstream kernel itself. The fact that they exist as modules in the kernel is what I meant by being included in the kernel. I alluded to this later on when I said that Valve "could remove [certain hardware] kernel modules to reduce bloat". Since posting that comment I've learned that Valve almost certainly has not done this. In any case, you're right that I incorrectly conflated core kenel code and kernel modules.
You're also right that I assumed that Valve built a custom kernel for SteamOS. I made this assumption based on the fact that certain hardware support is unavailable on SteamOS. Does this not indicate a custom kernel? Am I again misunderstanding the difference between the kernel and kernel modules? What did I get wrong here?
I was unaware that so much of SteamOS is open-source. Until now, I was under the impression that SteamOS was closed-source. I apologize for the error. If Valve’s work is open-source, then it sounds like they’re already inviting the community to take part, which helps distribute the development effort. I wasn’t fully aware of that, so I appreciate the insight. This can only be a good thing, right?
And of course I don't mean to suggest that Valve should prioritize developing SteamOS for the desktop over other projects. I completely understand that other projects take precedence. I just didn't realize that a SteamOS desktop project was even on the docket. Again, everything I've read so far has suggested that Valve has no plans at all to bring SteamOS to the desktop. That paired with the large number of comments I've seen suggesting that a desktop SteamOS would be pointless and a waste of time prompted me to share my own thoughts on the matter since I haven't seen anyone take the same stance. Regardless of the things that I've gotten factually incorrect, I still think that making SteamOS for the desktop at some point is certainly not a waste of time and would provide a positive experience for many users, which is the point that I was trying to make. Do you agree?
Again, I'm currently doing more of my own research here, but if you have any resources that you can provide, I would greatly appreciate it.
Edit: For anyone reading who is interested, I just found the github for the SteamOS kernel[github.com].
Edit: It seems I also misunderstood what "atomic" means. I was incorrect to say that SteamOS is atomic. It is not. Furthermore, though I didn't mention immutability, I unknowningly and incorrectly conflated (once again) the two terms. It would be more correct to call SteamOS immutable. The message that I was trying to convey was that SteamOS has safeguards in place that help prevent users from breaking it.
If you're interested in a "try before you buy" experience, you could look into using VirtualBox to install something like Linux Mint or Ubuntu in a virtual machine to dip your toes into the water. A virtual machine will allow you to install Linux without damaging Windows or losing any data, and if you break the virtual machine, you can just delete it and try again. Gaming isn't likely to work very well in a vm without configuring some things. Unfortunately, I have no idea how to configure a vm for gaming because I've never even attempted it. But you should at least be able to install Steam, and it should even let you install games to the vm provided you gave the vm enough storage space. It would also be a good way to test out Wine and Lutris which you can use to install the native launchers for games if you don't want to run them on Steam.
Just be aware that even with distros that are designed to be friendly to new Linux users, there will still be a learning curve. Even after living in Linux full time for about 5 years now, there are still a lot of things that I don't understand, as you can see from PopinFRESH's comment above.
Linux is not Windows[linux.oneandoneis2.org], but frankly you wouldn't want it to be. If it were the same as Windows, then it would be pointless to switch. So you'll have to learn some new things, just as you would if you were to switch from Windows to Mac, or Android to iOS.
Be patient, learn at your own pace, don't be afraid to ask the community questions, and don't be afraid to possibly sound like a noob, or even outright stupid. I'm not, as you can clearly tell. xD Linux isn't as hard to use as a lot of people think it is, and it doesn't always require you to jump into a terminal (Command Prompt/PowerShell) and mess with the system. Distros like Linux Mint and Ubuntu provide a lot of great GUI tools. If nothing else, messing with virtual machines will help you prepare for if/when a proper SteamOS desktop OS is officially released.
Live disks are a great option as well as they allow you to test a distro running on real hardware instead of virtual hardware. However, the reason I didn't mention live disks is twofold.
Firstly, they're not persistent by default, and often don't usually have an option to enable persistence. Nothing the user tests will be permanent, turning the live disk into Groundhog Day Linux and forcing the user to perfom the same tasks every time they boot. A live disk is great for exploring the UI and seeing how the distro runs on hardware, but that's really kind of a one-time test.
Secondly, in many cases the live versions of distros are intended to be used as installers, as is the case with Ubuntu and Linux Mint. A curious user may decide to check out the installer and could potentially stay a little too deep into the process. Once partitioning begins, it's game over for whatever was installed on the target disk. Bear in mind that many users trying Linux for the first time might not even know what partitioning is, so they may not realize what they're doing until it's too late.
I don't discourage live disks by any means, but I do urge caution. With respect, in my opinion VMs are a safer first step. Users can still explore the live environment, and they can even experiment with using the installer within the VM, which is significantly safer than playing with the installer in a live environment booted on bare metal. If they bork the install (as I did embarrassingly many times when I started), they can just delete the VM and start over. Also, once installed in a VM the system is persistent. Users can make changes, close the VM, then come back later and pick up where they left off.
Using a VM also allows the user to boot the .iso without having to sacrifice a USB stick to burn the image. Ventoy also technically allows this, but even then, Ventoy itself must be burned to the USB stick to work, which has the same effect of occupying a USB stick. A minor issue, but still something to consider for users who may not have a USB to spare.
There are pros and cons to both, but my personal preference is for using a VM.
Sure thing.
PC World CES 2025 Interview with Valve's Pierre-Loup (Time Code 5:56)
Discussing form factor and expanding to other handhelds he notes that the goal is to bring SteamOS to general PCs.
Lawrence Yang and Pierre-Loup Griffais both mentioned this in interviews in the lead up to the original Steam Deck launch when discussing SteamOS 3 and what they learned with the original Steam Machines and SteamOS/SteamOS2 experiences.
Valve Steam Deck Booklet from the expanded launch bringing Steam Deck to Japan, South Korea, Taiwan, and Hong Kong. Scroll to page 30/31 (page 16 of the PDF).
Again, it's Valve and they work in Valve Time so "soon tm" for them is likely not what you or I might consider "soon". Regardless, they've stated multiple times that their intention is to continue SteamOS 3 development with the intention of eventually being released as a generally available OS with an Installer for users to install on any (assuming relatively current) PC hardware.
No, they did compile a custom kernel from the shipping kernel with Arch. I was pointing out that your assumption was their reasoning for doing so; e.g. that it was customized for "handheld hardware" in mind. While I am sure they did work with AMD among other core hardware component vendors to ensure core hardware support was stable in-kernel for bootstraping the system; I'd wager a large portion of their custom kernel was additions to support the main novel features being the quick sleep and resume as well as the swapping between display servers / window managers. In other words, it wasn't "customized for handheld hardware in mind" rather it was initially customized with specifically Steam Deck hardware and features in mind; which happens to be a handheld form factor. This should be more clear if you listen to the discussions from Valve regarding what Steam Deck is; it is a PC. Hook up a dock with a keyboard, mouse, and display and it is a "Desktop" PC; just not what you are envisioning as a "Desktop".
Earlier on the Steam Deck recovery image had some device hardware checks in its extraction routine; however, those have been removed and with the launch of the Legion Go S it is fairly easy to boot the recovery image on an AMD APU, or system with similar AMD GPU and install it. There are still obviously going to be jank in doing this currently; which is why many people simply suggest other alternatives such as Bazzite, ChimeraOS, etc. for a similar console-like experience if that is what you are after.
The reason Bazzite is able to be so close to SteamOS, even though it is RHEL derivative rather than using Arch, is because the development of technologies like proton, and gamescope Valve is doing as open-source.
Every time you say "SteamOS desktop project", replace it with "SteamOS 3 general availability for PC". It's not a separate thing; just like there isn't a separate "Arch Linux for Desktop" or "Arch Linux for Laptop" or "Arch Linux for Handheld Hardware".
I'd suggest that it doesn't matter what I think or what any internet rando thinks. It matters what Valve's designers think and what they've consistently said, including GabeN, suggests they think that SteamOS as a generally available PC OS adds value for PC gamers while also reducing risk to Valve. Valve started this whole thing with SteamOS and Steam Machines more than a decade ago to help them break dependence on Microsoft for their core business (e.g. Steam). Their work with codeweavers to fork Wine and build out Proton should be evidence that this motive hasn't changed. Also in regards to your notion that they can "leverage the community" because of open source. I'd argue Valve's approach is quite the opposite. They are very financially successful from Steam. Their goal is to add value for PC gamers so that PC gamers continue using Steam and Steam users continue to increase.
Valve is developing new open source projects and contributing back to other open source projects to improve PC gaming on Linux so that any Linux distribution can/will be improved and gaming on Linux in general will improve; thus expanding the addressable market for Steam while also reducing Valve's reliance on Microsoft. GabeN has reiterated in multiple interviews that this is not a near-term cost/benefit equation for them. Rather it is a long-term strategy to make Steam provide more value to PC gamers.
See the above links; and kudos for finding and correcting your misconception of atomic. Your OP was in fact TLDR and I only skimmed it to get the general gist of what you were trying to say and commented on the parts that stuck out to me. I didn't even see you mention this previously.
So in conclusion maybe someday, possibly even before Half-Life 3, SteamOS 3 (proving they can actually count to 3) will be released for general availability with a nice pretty hand-holding GUI installer that walks users through installing it on their PC. Until then, Valve does as Valve does and does it in Valve Time.
Don't start holding your breath yet
Seems logical the SteamOS desktop PC release will come at some time in the future AFTER Valve releases a Valve-brand SteamOS desktop console.
It was necessary for a handheld PC.
But there is no good reason to release it on desktop where you would have to undo their changes and turn it back into vanilla Arch Linux.
What SteamOS can be used for are potential Steam machine consoles (mini pc).
Otherwise just install any Linux distro with normal desktop environment with steam.