Steam On Linux Marketshare Hits New Multi-Year High, AMD Powering ~40% Of Linux Gaming Systems

[spykes]

...

It needs to start somewhere, nobody would buy their device if the current games catalog was not compatible with it. So yes I'm happy it runs windows games, thanks to that I already have better graphic drivers, and we will soon have a better scheduler for AMD APUs as well, so yes it already improves Linux platforms.

[birdie]

Cheers and happy gaming.

As a person who closely follows Linux kernel development I have to admit that regressions hit the Linux kernel far more often and harder than any Windows issues which are normally blown out of proportions and affect a very small group of people.

Case in point: Linux 5.15 has again broken multiple systems with ... AMD graphics. Broken to the point they are completely unusable:

https://bugzilla.kernel.org/show_bug.cgi?id=214901

https://bugzilla.kernel.org/show_bug.cgi?id=214859

https://bugzilla.kernel.org/show_bug.cgi?id=214853

And such critical bugs affect each kernel release. I don't remember the last time AMD, Intel and NVIDIA have released outright broken driver releases for Windows. And don't get me started but "Distros don't include point releases anyways". That's bullocks. I once had to spend 20 hours debugging a critical issue affecting the Linux kernel after the next stable version of it had already been released, and a version after was already in development. All the rolling distros at that point already included the broken kernel including Fedora and Arch.

I'm sorry, but Windows overall is a much more stable platform despite the fact that fans of Open Source love to claim the opposite. OEMs not only have their own QA/QC for Windows systems, all the drivers for Windows go through WHQL, so there are two certification processes involved.

In Linux on the other hand, patches are included based on "I'm fixing this problem, please merge". What hardware has actually been tested? What regression tests have been run? Nothing.

Again the Steam handheld will not address any of the inherent Linux issues but surely will improve Wine and DXVK.

Cheers and happy gaming.

[oiaohm]

Looks like I was out of the loop, sorry.

It remains to be seen how many people will use the handheld in this mode. Won't it require an external monitor, keyboard and mouse to be used this way? It surely doesn't look to contain a built-in keyboard.

Not at all. There is a full 7 inch touch screen in the steam deck so it can do a full on screen keyboard and the two touch pads can be mouse. Do remember KDE plasma without many modifications has been run on the pinephone.

I would say if you were doing a lot of typing you would most likely want external keyboard. Yes this could be wired keyboard or bluetooth keyboard.

Do remember there are quite a few KDE games that are not in the steam store but will work quite well on a steam deck. So maybe you are back at the KDE desktop to play KDE games.

Will Steam run a normal unmodified Linux distribution? Considering that the Steam runtime for Linux comes with pretty much everything but the kitchen sink it doesn't seem likely. Native Linux Steam games are tested and built against specific library versions and Valve prefers not to deal with new versions which could result in serious regressions and hard to catch bugs.

https://www.steamdeck.com/en/tech

The reality its in the specifications.
Steam OS 3.0 will be a lightly modified arch Linux with KDE plasma.

Steam application has it only steam runtime sandbox stuff.
https://gitlab.steamos.cloud/steamrt...ressure-vessel That is basically valve own version of flatpak. So the same solution valve uses on any normal Linux desktop distribution for giving games specific library versions is what will be used on the steamdeck. Birdie why would valve bother reinventing the wheel here.

Will it use the vanilla Linux kernel? Also seems rather unlikely.

Valve has stated they wish to use a mainline kernel.org as soon as possible. So they are working on upstreaming everything this include the major push on futex2 work. So this is not unlikely. Steamdeck when demoed to people like Linus tech tips was using a mainline kernel. Of course if futex2 is merged there will most likely be zero third party patches of any interested to valve for the Steamdeck so just use mainline.

Will it be subject to the same update policy as normal desktop distros? We'll soon find out.

Valve is planning to just use arch updates so we will see how that goes.

The reality here is Valve is really using what is fairly much a normal Linux distribution being arch and a normal Linux desktop being KDE. On a console. Of course this does not mean it will exactly be a desktop.

birdie really do you think Valve wants to waste man hours remaking what already exists and works good enough.

Many like to believe that Android is also kind of "Linux" however the existence of Android has had zero effect on Linux popularity on the desktop. Many Android phones can be used as desktops - I don't know anyone who actually use them this way.

This is different because the steamdeck is going to be using same kind of Linux people do use on desktops. So valve work in places to fund fixing up final parts of KDE like kde kio for system files by polkit so dolphin file manager can work right will help all KDE users. Valve upstreaming code into the main Linux kernel as well.

Screaming that Steam handheld is going to be a Linux desktop is premature at best, insincere at worst as no one here really knows what it will encompass.

No we already know the software stack valve has demoed it. So on the steamdeck is basically a stock standard desktop Linux with steam store starting in big picture mode.

Now being hand held console sized you may not wish to use this as a desktop computer.

KDE is able to configured for touch.

[Mez']

Imagine a person wants to install Linux on his device. Actually don't, what you've offered right now is simply borderline insanity. There should have never been a situation like this in the first place. You don't go around scouring for the "proper" Windows or MacOS release - you just install, download and install drivers in Windows mostly for your GPU and you're all set. Everything works, nothing breaks, no regressions to speak of.

I favor choice. Always.
I can pick whatever type of upgrade process suits me best in an active and conscious decision depending on the use cases, rather than passively being subject to whatever decision has been made in my name with no idea of what I actually prefer for what device. On the contrary, I'm very glad this situation exists in the first place. I can have my critical mind do its work rather than being a MacOS user with no thought of his own and following the cult blindly because that's how it's been pushed down my throat.

Plus, everything works for me, so that applies to Linux as well in my books.

[LeJimster]

We're gonna see unfathomably high Linux share on Steam. I'm talking 2.5, even 3%.

If the Steam Deck sells millions of units it will go well above 3%. However, with supply shortages.. It's going to hurt adoption in the short term.

[birdie]

I favor choice. Always.
I can pick whatever type of upgrade process suits me best in an active and conscious decision depending on the use cases, rather than passively being subject to whatever decision has been made in my name with no idea of what I actually prefer for what device. On the contrary, I'm very glad this situation exists in the first place. I can have my critical mind do its work rather than being a MacOS user with no thought of his own and following the cult blindly because that's how it's been pushed down my throat.

Plus, everything works for me, so that applies to Linux as well in my books.

The "WFM" mentality won't propel Linux anywhere specially among the people with no IT background. I don't quite understand what it is that you're proud of exactly. You may as well use FreeBSD, HaikuOS or anything open under the sun and call it a day.

I want a Linux something (distro whatnot) which I can recommend to all my friends and not be afraid they will be stuck staring at the boot screen showing hieroglyphths - a situation which is all so familiar to Linux users, not knowing what to do to fix the issue. Yes, you can totally break Windows to achieve the same level of "success" but in Linux it comes naturally due to the fact that no one is really responsible for anything except of course RHEL but no one here really considers this distro to be a desktop OS. Debian stable is also extremely stable and equally unsuitable.

[spykes]

I'm sorry, but Windows overall is a much more stable platform despite the fact that fans of Open Source love to claim the opposite. OEMs not only have their own QA/QC for Windows systems, all the drivers for Windows go through WHQL, so there are two certification processes involved.

And despite that they are still failing to provide a correct handling of AMD CPU even after their first patch:

https://www.tomshardware.com/uk/news/amd-ryzen-windows-11-patch-testing-gaming-benchmarks-L3-cache-bug

"Much more stable platform" my ass

[chocolate]

As a person who closely follows Linux kernel development I have to admit that regressions hit the Linux kernel far more often and harder than any Windows issues which are normally blown out of proportions and affect a very small group of people.

Case in point: Linux 5.15 has again broken multiple systems with ... AMD graphics. Broken to the point they are completely unusable:

https://bugzilla.kernel.org/show_bug.cgi?id=214901

https://bugzilla.kernel.org/show_bug.cgi?id=214859

https://bugzilla.kernel.org/show_bug.cgi?id=214853

And such critical bugs affect each kernel release. I don't remember the last time AMD, Intel and NVIDIA have released outright broken driver releases for Windows. And don't get me started but "Distros don't include point releases anyways". That's bullocks. I once had to spend 20 hours debugging a critical issue affecting the Linux kernel after the next stable version of it had already been released, and a version after was already in development. All the rolling distros at that point already included the broken kernel including Fedora and Arch.

I'm sorry, but Windows overall is a much more stable platform despite the fact that fans of Open Source love to claim the opposite. OEMs not only have their own QA/QC for Windows systems, all the drivers for Windows go through WHQL, so there are two certification processes involved.

In Linux on the other hand, patches are included based on "I'm fixing this problem, please merge". What hardware has actually been tested? What regression tests have been run? Nothing.

+1 for the info, that's... bad. Luckily haven't been affected by that on my little gaming machine (RX 5700, yes I still remember you were among the early adopters and opened bug reports for RDNA).
Fair enough, at least we can extrapolate that regression testing should be a top priority for Linux & first-party driver developers, especially given the monolithic nature of the kernel.

But my line of thought is still the same: if I'm making comparisons, then I can't be reading about Windows breakages on Softpedia for years and be convinced that the overall experience would be more stable than mine is on Linux, simply due to the existence of a driver certification programme; after all, MS ain't testing every rig out there for regressions either, as proven by fatal updates in the wild. But, again, Linux adopting solid engineering practices to catch such regressions/breakages would be more than welcome.

Now, those games aren't gonna play themselves, are they? Just tried Microsoft's Age of Empires II: DE the other day, works alright via Proton-GE. Quite the nostalgia attack.

[birdie]

And despite that they are still failing to provide a correct handling of AMD CPU even after their first patch:
https://www.tomshardware.com/uk/news...s-L3-cache-bug

"Much more stable platform" my ass

GCC has performance regressions almost each release. The Linux kernel also regresses in terms of performance quite often - you can find lots of benchmark results here on Phoronix. I'm not sure what it is that you're proud of. This performance regression in Windows is the only one on my memory for the past 20 years. Lastly, I am talking about criticial bugs preventing users from using Linux at all, and you've found a performance regression which affects a small number of users running Ryzen 5000 CPUs and Windows 11 which was released a month ago, and then it wasn't that serious: https://www.tomshardware.com/news/am...s-L3-cache-bug

The Windows 11 patch undoubtedly corrected some issues with L3 cache latency and bandwidth, but the profound differences in synthetic measurements simply didn't carry over to most of our real-world gaming tests.

Imagine every distro upgrading to a new major Linux kernel version as soon as it gets released and all the pain that will follow. This is Linux for you.

I've got this discussion a hundred of times already and I'm bloody tired. Any Windows issues are blown out of proportions, while criticial Linux bugs are swept under the carpet: "but you can use an older kernel". Yeah, right, meanwhile people using Linux have no idea what to do unless you're smart enough to understand what's going on and fix the issue. If it can be fixed. I've seen tons of threads recently where people complain, "I've tried six different Linux distros and all of them boot into the black screen". This is simply unheard of in Windows.

[chocolate]

Linux in the Steam HW survey running ... Windows games. If it's not pathetic and utterly insulting, then I don't know what that is. Let's celebrate it!

I don't know about insulting, but certainly not pathetic. Wine & co. will go down in history as a great software preservation platform that also enables Windows software to run on CPU architectures other than those supported by Windows. Legacy software isn't getting updated and is, or is becoming, increasingly less compatible with current and future versions of Windows. I have now forgotten which ones, but I swear to you I have encountered some games on Steam where people opened discussions about the game not starting on Windows 10 at all or requiring workarounds. I know there are success stories about GUI programs written for Windows 98, not changing one bit to this day, and still functioning on Windows 10, but I guess games just aren't that simple and have to account for multiple moving grounds such as sound and input.
Right now, the market is targeting the "PC" intended as a Windows host. We need it to shift toward an OS-agnostic set of APIs & technologies that can let publishers justify the existence of builds that span multiple CPU architectures and operating systems with close to zero additional development cost (so, I'm leaving QA & support costs out to focus on the technology aspect).
Then, if there's interest in Linux as a host to publish "native" games on, and the games aren't native native, but Proton-native, then maybe it becomes insulting. At the moment, I don't think offering Proton-native builds is insulting. Developers can take their time learning new stuff, and I don't have to stomach Windows, win-win.