I vehemently disagree with sentiment expressed in the article. Sure, on the surface the work done Valve does significantly improve the state of Linux gaming. But this is not Linux gaming. This is Windows gaming under Linux. We need to think about the long-term effects, about maintaining a healthy ecosystem. I cannot imagine how delegating Linux and Vulkan to be a long-term emulator of DX12 and Win is supposed to be healthy. What about platform autonomy? Shouldn't we strive towards a future where Vulkan is a strong, independent API with its own voice and not just an interface for emulating DX12? And one where game developers actively want to support Linux as first class citizen and not just make Windows games and putting all the burden of compatibility on the maintainers of the emulation layer.
The worst outcome of this is that the market share and quality of Vulkan implementations will diminish. WebGPU is a strong upcoming API contender on the mobile side, which IMO has a good chance of displacing Vulkan for many applications. And if DXVK works that well on Linux, more developers might end up choosing DX12 and only supporting Windows. Giving up autonomy and becoming a subordinate of a different technology stack is never a good idea. I mean, if this is the way how the Linux community wants to play, then why not make it official and just scrap Vulkan and have a DX-compatible API in place with IHV drivers?
The Linux Desktop community has had 20 years to make Linux Desktop a reasonable, if not attractive, platform for gaming and game development and they have consistently and utterly failed at it. Valve did the only thing that was going to work.
That is true. Proton is a pragmatic solution. But it hardly a satisfactory solution, and I worry that it is not sustainable solution either. I do not really have a horse in the run personally, as I am not a Linux user (my platform of choice is macOS), but as a GPU programming enthusiast I dislike anything that gives Microsoft more power.
And of course, let us not mistake Valve's intensions. They couldn't care less about Linux users. They are simply leveraging Linux as a low-cost platform to get into the console business.
It doesn't have to be forever. The biggest problem with Linux gaming was simply the games you want to play don't run and the ones you did want to play the, makers had no incentive to distribute a linux build. This bridges that gap, removes a lot of friction for game makers to support linux and with the steam deck it could give them millions of new users to target. It's a massive step in the right direction, IMO.
While these moves are incredibly beneficial for Valve, I applaud them for doing it in a way that benefits the community as a whole and not just themselves.
> but as a GPU programming enthusiast I dislike anything that gives Microsoft more power.
I don't see how this gives Microsoft more power, if anything it takes it from them. People are not dependent on Windows to run Windows games, and developers can target Wine directly if they wish.
> And of course, let us not mistake Valve's intensions. They couldn't care less about Linux users. They are simply leveraging Linux as a low-cost platform to get into the console business.
I've been watching Valve since the beginning and that's not quite how I read it. I do think they're pretty neutral on Linux, that much is true, but in general Valve has consistently fought for openness in gaming. Pretty much anyone can sell on Steam, Valve will let you bypass the 30% cut and sell Steam keys on other stores if you want, they created the OpenVR SDK while Facebook did everything it could to close theirs, etc. Their moves on Linux aren't about having a console, it's about having an alternative to Windows so Microsoft has less control over their (and the rest of PC gaming's) future.
Much more than that, in my mind, is linux-native steam installations.
I can now play many games directly on my linux laptop without tinkering with anything. It's fantastic and has done more to make me optimistic about linux as a desktop / gaming platform than anything else.
I have reluctantly reached this exact conclusion. Sure, I wish Linux were a growing target for developers, but after living through Loki/Icculus ports and Wine/Crossover/Transgaming/Proton, the only practical approach is the latter, it seems. I've made my peace with it.
If Valve makes it possible to play many/most games on Linux then an influx of Linux converts may incentivise more native development.
Certainly the current market has not created those incentives, whereas a massive backcatalog w/ a (potentially) popular portable device may move that ball forward if it means there's a built in market for it going forward.
Recent release of the Windows 11 with requirements to have at least 8th gen CPU might help actually. My 6th gen is perfectly fine for a lot of games, and thanks to Valve I have quite a good library on Ubuntu.
Same was said for Vista, Windows 8, Windows 10 and probably many other times. Microsoft is not that stupid - they are doing these changes only after they are confident that they can get away with it and can easily change their strategy if needed.
I think that's somewhat the point of what Steam is doing. Positioning themselves to exist without MS in the worst case while applying pressure that may curb MS's most user hostile or competition-limiting (in the app ecosystem) initiatives. Especially at a time of chip shortages and price premiums, users are more likely to hesitate buying a new computer just to have the latest win version when their existing hardware is still fine.
I think this is especially the case with gamers where and older CPU is much less likely to be a bottleneck in performance.
Someone better educated than me can probably help me out here, but I think you're either misunderstanding what some things are, or mixing up your terminology.
Proton is the emulation layer created by Valve.
Vulkan is a cross-platform graphics API.
Your point might still stand, but it's worth debating. Without Proton, gaming on Linux is limited to native Linux games. With it, the story changes. Someone who likes PC gaming might be willing to turn Linux into their daily driver, game on Linux (with a mix of Proton and Vulkan), and now the market starts to support developers making more native Linux games, where they otherwise might not.
I was not mentioning Proton explicitly. But the core of my point is that Proton de-facto delegates Vulkan to be a second class citizen whose purpose it is to serve as a backend of implementing a Microsoft-compatible technology stack. With Proton working well, the incentive to develop native Linux applications and games goes to wards zero — why would you even bother if you can just develop and test for Windows and let the Proton maintainers sort out the rest?
Currently, there are some games that use Vulkan to target both Windows and Linux. Proton encourages the developers to use the Microsoft technology stack instead. Apple has their own Metal, so Vulkan is dead in the water there anyway. Where does this leave Vulkan? Android? That is likely to move to WebGPU as I mentioned. GPGPU? That is totally dominated by CUDA and other vendor-specific APIs. Linux pro applications (Blender)?
If there is a native Vulkan backend there is little reason for an API emulation layer with an identical before-and-after to not just forward the calls on as much as possible - especially with an API which makes its state a lot more explicit and keeps less internal state.
With that said, as a game dev, Vulkan - along with DirectX and OpenGL - are second-class citizens already. The large bulk of gamedev is concerned with content, gameplay systems, and UI, with graphics being on the long tail of things that can make a title look truly unique and great, but the prevalence of Unity and Unreal make the actual rendering pipeline a thing of secondary - hugely important, but not primary - concern to most titles.
This comment is loaded with some pretty unrealistic conclusions. Firstly, Vulkan support is becoming extremely common even on Windows-exclusive titles.
> Apple has their own Metal, so Vulkan is dead in the water there anyway.
How is Apple at all relevant in a thread about games? Apple has proactively shut out gaming in the past few years. Metal is almost exclusively used to draw fancy GUIs.
Furthermore, DXVK is an open source (strictly) project. Who honestly cares who came up with the API? People even run DXVK on Windows because it often offers superior performance.
There are some very big studios getting behind Vulkan: Embark, id.
Could you expand on what you are trying to argue? It seems to me that Proton is the immediate reward while waiting for the platform to grow to get native ports would be the delayed reward. Yet "In follow-up studies, the researchers found that children who were able to wait longer for the preferred rewards tended to have better life outcomes" would support the caution against Proton if you were to generalize from children to platform strategies (not saying that you should).
If I may, I suspect your angle on this might be slightly askew at least from my perspective. Almost all the major game engines are now supporting a native vulkan/linux renderer. I'm running in godot 4 right now. If anything, watching things like glorious eggrolls proton and the dxvk repo [1] I think the people working at the direct translation level are causing more eyes on vulkan than it would otherwise be getting.
My actual concern for proton/wine on linux is security related. The binary game space is a security nightmare, and enabling the windows side to run with very little compartmentalization is going to be a security disaster.
Are you saying it's the case that running a game on Proton on Linux is a superior experience to running it natively on Vulkan?
That's surprising to me, but yes, if it's better than a native graphics API, it's probably better for everyone if developers stick to DirectX.
On the other hand, I don't think you addressed my point (with my assumption that games run better natively on Linux in Vulkan). If gamers are just going to skip Linux entirely without Proton, how is that better for Linux gaming, than if gamers migrate to Linux using Proton for an acceptable emulated experience, then there are more Linux gamers, and developing for Vulkan has some incentive it didn't have before.
> If gamers are just going to skip Linux entirely without Proton
Gaming on Linux was a thing before Proton and so far, it has not brought a giant jump in Linux users - e.g. Steam hwsurver still has Linux users hovering at about 1% [0]. Maybe that will change but so far this argument falls flat.
OP lauds their WINE-like extensions, but there is much more they are doing.
Not least of which: Making STEAM run perfectly in linux so that linux-native games have a seamless installation / management platform for linux-native machines.
Then of course steamdeck, which brings linux to those who would not otherwise try it and further incentivizing developers to target linux native.
We need boots on the ground. We need people installing Linux rather than making excuses about why they can't.
Obviously what you describe is a problem. If Valve is going to spend the money getting games working on Linux no one, trying to make money, would port their game. It will hurt. It will hurt groups that specialize in porting Linux games.
Without the installs, whatever we do today is unsustainable.
The worst outcome of this is that the market share and quality of Vulkan implementations will diminish. WebGPU is a strong upcoming API contender on the mobile side, which IMO has a good chance of displacing Vulkan for many applications. And if DXVK works that well on Linux, more developers might end up choosing DX12 and only supporting Windows. Giving up autonomy and becoming a subordinate of a different technology stack is never a good idea. I mean, if this is the way how the Linux community wants to play, then why not make it official and just scrap Vulkan and have a DX-compatible API in place with IHV drivers?