I think some folks are missing the really neat thing about this: remote play.
Steam supports playing games remotely on another computer. So your desktop is doing the heavy lifting, but you can view the game on your TV in the livingroom. Now you can be playing the game on your Chromebook as well.
Then the next step is that Google Cloud offers you an on demand high end gaming machine, pay by the minute, which runs Steam on Windows x86 and has an instant VPN to your Chromebook.
I set this up successfully once on an AWS machine and a very low end laptop running Ubuntu, but I couldn't get the hardware accelerated video encoding to work right. 3 seconds per frame made the game hilariously unplayable, but it did function.
> Then the next step is that Google Cloud offers you an on demand high end gaming machine, pay by the minute, which runs Steam on Windows x86 and has an instant VPN to your Chromebook.
I find it difficult to believe they'd offer this, given that they already have Stadia.
I do see how this could be more appealing for consumers, but aside from competing with themselves, I also would be curious about the economics. Considering how expensive GPUs are, is it even possible the prices for an on-demand gaming service could ever make that much sense? With Stadia, they are making you buy the games through Stadia itself, where they would get a cut of the revenue, which probably helps subsidize the actual cost of the service a bit more.
There's also other very irritating issues. For example, a litany of AAA game companies do not allow their games to be streamed via GeForce Now, even though it would be running your own copy that you bought from your Steam or other library. I assume they would fight to earn additional money on game streaming, even though the money simply isn't there.
This is unfortunate, but perhaps it's OK. Lower end devices will be getting more and more powerful and eventually I'm sure even reasonably priced Chromebooks will be able to play games decently. Given this launch only applies to newer Intel processor based Chromebooks, that day is not today... but still, it certainly could be on the horizon.
Google never really tried with stadia. It's just another dead Google product, now changing into some sort of B2B thing. Shrug.
Meanwhile Nvidia kept improving geforce now and it's actually really good now, running on everything from operating systems to browsers with no noticeable lag. Cheap too, like $15 a mo or so, and works with most of your Steam library. Where it doesn't, usually means that the publisher wanted to get off for some reason.
Xbox Game Pass is another great steaming option (via xcloud), Netflix like subscription service that lets you stream games.
There's also Shadow and Luna and such. Many companies are doing this well, just not Google.
Google is even worse than Apple in what concerns talking to game developers, they always focus on the Play Store and KPIs, and seldom focus on the tools.
Even when they really try (see last week Games Development Summit), it is mostly about high level stuff, like PMs talking to PMs, handwaving to lightweight docs and github repos.
Naturally very few game devs ever get enthusiastic from whatever comes out of Google.
Great point. Meanwhile Steam and the other platforms spent years and years making great tooling and ecosystems to court developers. Stadia just feels like an executive's half assed demo...
I don't think Google really has a game plan (no pun intended). They never gave Stadia the dev and marketing resources it needed so it just died on the vine. Their "pivot" is more like a surrender, I think, an act of desperation where they hope other companies will pay them for their tech that never really took off with consumers.
Thing is, porting to Stadia is so much work vs flicking a switch in Steam to publish to GeForce Now. And Stadia can't cross play or cross save with anything else by default, making it a very lonely ecosystem.
That is why they are now writing a Windows emulation layer for Stadia, which I think won't convince anyone that isn't yet bought into their Stadia sales pitch. See last week Stadia sessions for developers.
It feels to me like the only logical economic future for game streaming is as a value add for a game subscription service (so like, xgp and xcloud as a logical pairing).
$100 or so a year for a game steaming service is a heck lot cheaper than a gaming PC and a graphics card update every few years. And you can play anywhere, on any device, where you have a fast connection. It was a godsend for me before I could afford a gaming PC (or find one, during the pandemic)
> $100 or so a year for a game steaming service is a heck lot cheaper than a gaming PC and a graphics card update every few years.
You are not going to compete with high end hardware/graphics using Stadia, not sure if that was solved but for quite some time they just upscaled to the promised resolution instead of actually rendering it.
Everyone casual will be kicked to the next set of consumer hardware by Microsoft(Windows 11 incompatible) or Google (required native codec support for some android apps) either way.
That $100 or so (definitely way more) is going to go on top of the hardware upgrade treadmill, not as alternative to it.
Stadia was dead on arrival. Working with Valve has the potential of being really really smart, but I’m not sure this develops into anything significant.
Every time I've tried Steam Play, even over hardwired gigabit, it's been laggy, buggy, and crash prone. There was always display switching problems, controllers and keyboards not being detected correctly, graphical issues, etc. Just a PITA.
GeForce Now works much better and doesn't require you to have your own GPU. Do need fast internet though.
I use Remote Play regularly (bed-bound lots due to medical issues) and never had a problem, but have found it’s finicky on your setup. Eg it never really worked on my ISP-provided modem/router but once I got a UDM (for unrelated reasons) it worked flawlessly and has ever since.
The fact I can lie in bed with laptop plugged into bedroom TV and use a PS4 controller to play games running on my gaming PC in another room is truly awesome.
And I believe you. I wish I knew what the variables were. I really wanted to like Remote Play and even got special hardware for it (Steam Link, Shield TV, etc.) and still couldn't get it to be reliable.
Yeah as alphager said, it’s a router by Ubiquiti that solved a lot of my networking issues. I wish I knew what exactly it was that made Remote Play work, but that was just one of the many things that didn’t work before then I realised later suddenly worked with my new router.
You can do remote play on a Chromebook right now with Moonlight! Works great (for me). Some other alternatives include Parsec and Rainway, and probably a few others I'm missing.
AFIACT, it's a "no but yes" sort of answer. Linux games are free to build binaries for any architecture they wish, but any game relying on Proton because it's a Windows binary will require x86.
Good news is, any of us in here are going to buy whatever we want and aren't limited to Chromebooks. Its everyone else that is screwed by the virtual duopoly of Windows and !Windows.
There's Box86 and Box64 which run x86/ARM64 binaries on ARM64 fairly well. People have gotten games like Portal 2 running on a PinePhone Pro: https://www.youtube.com/watch?v=yPr0Aw3xZrA
That's not entirely clear to me. Windows on Arm has some weird sort of compatibility layer that could also exist here. It's a different environment but I'm able to run (some) x86 windows programs on an Arm Windows virtual machine, unmodified, under Parallels on an M1 MacBook. Boggles my mind how it all works together, but it does.
I remember when Chrome OS first came out. I have an old Chromebox I got at Google I/O over a decade ago too. And I remember all of the narrative about it being better because it was just the web browser, and the web was all you needed.
Chrome OS has firmly educated me on how to read marketing copy about things the product doesn't have. "Tell people they don't need it until you catch up to your competitors."
Ten years more from now if Chrome OS still exists, I expect it to simply be another macOS or Windows clone with more or less all of the same features and trappings.
Wait - does this mean that if Steam ships on Chromebook, and Steam can be made to run arbitrary software via proton, that Chromebooks have become regular laptops?
It is worth noting that ChromeOS runs Linux software in a neat way to isolate it from the host OS.
It runs it in Debian-based KVM VMs, and it uses virtio-wayland[1] to pass the applications through to the host compositor, making them feel as if they are not running in a VM.
This architecture means that ChromeOS can preserve its readonly system partition, delta updates, and fast boot times while being able to run any Linux app. Very similar to Qubes, actually.
Does virtio-wayland support accelerated graphics? It's not clear to me how that would work, since accelerated graphics drivers typically include hardware-specific userland code that has to run in the application.
Since it is running in a VM and the host needs the GPU as well, it likely uses virtio-gpu to do paravirtualization. The latest iteration involves emulating a Vulkan device, serializing the commands, and passing them through to the host:
The irony here is that this is a ChromeOS-specific issue, not really related to typical desktop Linux.
Of course, it can and hopefully will be improved by improvements to desktop Linux. I haven't kept up, but ChromeOS will probably be able to adopt newer open source desktop technologies like Wayland and Pipewire/XDG portals in order to get a better Linux multimedia experience, if they care to.
To be clear, chromebooks have for a few years been able to be regular labtops since crouton [1] allowed installing linux distros on them. (And could even use wine to run windows games on the x86 chromebooks.)
Or go a step further and flash a modified UEFI firmware on it, then natively run a Linux distro. My HP Chromebook from 2013 is rocking still, almost a decade after release. I run Gallium on it, for optimal compatibility with all the hardware.
Regardless, last time I was in Walmart Chromebooks where the only laptop available. For better or worse (mostly worse IMO) as far as a large portion of the population is concerned "regular laptops" are specialized equipment that they won't have.
Maybe Chrome OS Flex will eventually support ARM Macs [0]? That would be cool for sure. I installed CloudReady for someone last year and it saved him from buying a new computer, and saved me from worrying about ongoing OS updates.
Disclosure: I work at Google and am speaking only for myself. I do not work on anything related to Chrome OS.
With the exception of a few missing things like full USB pass-through to crostini containers, chromeOS with crostini works about as good as my fedora laptop. I like its window management defaults better then gnome as well. Also you get auto-updated and supported firmware from google and secure boot by default.
Currently running a framework laptop with fedora just because I like the hardware better then most chromebooks and I can very cheaply throw in a fast 2TB ssd rather then spend hundreds extra to upgrade a chromebook to 512GB.
If you have a remote box, probably just remote SSH + web IDE (potentially running on your remote box). For full local, you'd do it through the Crostini layer, which is basically Debian.
I tired chromeOS a few times as a dev machine and was always disappointed. Thing is, OSX doesn’t really get in my way using just a terminal, web browser, and spotlight. Installing stuff can be annoying at times compared to Linux, but it was even worse on chrome (basically didn’t work). I ended up ssh’d to a remote vps on chromeOS to get things done.
Not really. Here are the advantages of ChromeOS as I see them:
Fast, unobtrusive, and regular software updates
Better touchpad and touchscreen support than any other Linux
Perfect integration between local files and files on Google Drive
A bare-bones userspace that is not bristling with security vulnerabilities
Software including the kernel is peak-optimized with profile-guided feedback, specifically and separately for each model of Chromebook, unlike most Linux distros that ship barely-optimized builds without PGO and targeting k8-generic.
But I suppose it can be useful for arguing that Linux gaming market is expanding, so developers who like to excuse lack of Linux releases with market size will have less excuses now.
At this stage, native Linux releases are becoming less and less likely because of Value's efforts to improve gaming on Linux, as paradoxical as that seems. Proton is getting better and better with each release, and in many cases, is providing better performance than running native code on Windows.
Why bother with official Linux support when you can just check that your game runs via Proton and push the burden of support onto Valve and the Proton community?
If Proton just becomes the "game runtime" for Linux, that's OK! It's still open source. There may come a time when Proton compatibility becomes as important or even more important than Windows compatibility for game developers. Not only might game developers start explicitly supporting Proton, they could even contribute fixes upstream, unlike with Windows.
Ultimately if it becomes popular enough, Proton could execute the embrace, extend, extinguish strategy on Windows itself. We'll know it succeeded if one day Microsoft gives up NT kernel development in favor of shipping Linux/Proton, just the way they ultimately gave up on Trident and started shipping Blink/Chromium. Today Windows is a relatively small and shrinking percentage of Microsoft's revenue so maybe it's not out of the question in a decade or two.
I'd argue that Proton is better than native Linux support in the long term, hear me out.
Linux userspace API has no promise of long term compatibility, and in fact there's a lot of churn, especially nowadays as we approach the Year of the Linux Desktop and technologies come and go as they're improved.
The Windows API instead is known for its long term compatibility. Microsoft goes out of their way to ensure applications keep running a decade later, and using that as a base for gaming is a win-win, as developers can target two operating systems with one API, and gamers have more guarantees their game will still be playable on Ubuntu 2030 edition.
The last few times I played native Linux games I had to fish for old and unsupported libssl and libjpeg libraries that my distribution doesn't ship anymore. I can blame the port, but nowadays I just try the Proton version first.
>The last few times I played native Linux games I had to fish for old and unsupported libssl and libjpeg libraries that my distribution doesn't ship anymore.
The so-called steam runtime can be targeted by devs to prevent this.
It's not about Windows long term support, it's about Wine long term support (which is much better than Windows own).
So I agree with your point that long term Wine offers better support than Linux native ABIs.
I doubt Windows ABIs are better than Linux native ones on their own long term wise (i.e. without Wine).
That said, it would be cool for someone to develop Wine-like wrapping of historic Linux ABIs into modern ones so you could have the same preservation effect.
There was for example such project for older SDL over new one.
The steam linux runtime is open-source with a license that reads much like the MIT license, so you need not run anything Steam proprietary if you don't like Steam.
I don't see why it has to become less likely for those who care about performance (which major games usually do). Translating DX12 or DX11 into Vulkan has performance overhead, even if you manage to run the game better than on Windows itself (for example Cyberpunk 2077 performs on Linux better than on Windows now). But native releases can perform even better.
And it's probably not always the case that Windows games perform better on Linux through translation than on Windows.
Besides, Wine has to play constant catch up to all kind of NIH stuff that MS will produce. Direct storage is the new one that doesn't have any translation implementation yet. So value of native releases won't disappear.
I play elden ring on windows.. with dxvk/vkd3d dll because they fixed the massive stuttering issue that this console port had on PC. Even on windows this translation layer can give better performance.
I’ve seen game devs advertise that their game works natively on Linux and is seen as a positive. My guess is if steam deck takes off there will be more games trying to build natively on Linux.
Besides, supporting a game means more than making a Windows build and expecting it to "just work" through Wine / Proton. Proper support requires some effort, even if it's translated. So they can as well support a native release if they even care about support.
I'm a Linux gamer, and when I see an indie game with native Linux build I avoid it. In my experience they always have problems, and I can't work out how to get Steam to give me the windows version with proton instead of the Linux version.
You and the dozens of people like you, but the Chromebook consumer market is huge.
Right now Steam is only targeting higher end Chromebooks that make up a small fraction of its market share, but I’m sure they’ll improve that over time.
This is whats frustrating to me. We're talking a small slice of most chromebooks out there.
I actually have used old chromebooks with crouton to install steam and play games. You have to pick "lightweight" games like Baldurs Gate, but it still opens up a lot of possibility.
So I had hoped maybe something could be brought to old hardware, but I guess not.
I think you start by targeting the platforms with more horsepower and can move down over time. Obviously they might just not do that work and stay on the high powered ones, but I don't think it's a given.
It's a small slice today, but today's high-end computers are tomorrow's cheap used machines, and future budget machines will be more powerful than present high-end ones.
When travelling, I have used GeForce Now to play Fortnite on an a pretty weak ARM Chromebook from a web browser.
I was able to turn up the graphical settings up much higher than what my regular desktop's ageing GTX 970 graphics card can get, while still maintaining a high frame rate.
Latency is low enough to have fun and contribute to the team. Wasn't a good experience over cellular, but it was very good on a wired connection.
I was very happy to have had a streaming option available at the time.
I am! Though I've been using Stadia + Chromebook when traveling. As for runnig games locally: I'm not sure if most Chromebooks have adequate GPUs to run recent AAA games, but I'm guessing that's a small portion of most people's Steam catalog.
I am a hardcore PC gamer.. but last Christmas I was at the in-law's place and we had a blast trying out a bunch of stadia pro offerings. So much so that I seriously considered buying FIFA on stadia. I just think there's too much of occasional lag on cloud gaming which makes it only suitable for casual gaming.
I was a "only have a Macbook Air" person for a while. You end up sticking to small indie things, but there's lots of games that work on the old integrated graphics stack, available on Steam.
You'd think they would want to double- and triple-down on their cloud gaming platform, Stadia, for Chromebooks. But there's also good reason to believe that platform won't be going the distance so perhaps this is another signal in that direction.
... does Steam provide any sort of services that manage the runtime of games or is the majority of the game still dependent on the users rig? remote GPU workloads/rendering, managing game-play coordination, etc?
If not, then it sounds like working with google (maybe others too) to get Steam running on ChromeOS is a fantastic idea that doesn't conflict with any of their own ambitions.
It's good for google bc you'll potentially attract more users to ChromeOS and once they're on ChromeOS, I'm sure they'll have more exposure to Stadia and other google services. Moreover, as game developers and platforms start to migrate a lot of their workloads to the cloud, Steam might be able to provide a more seamless way to do so (proton?)... again, potentially using google services or even the Stadia backend to do so.
It's good for Valve bc if furthers their independence from Windows. It may also lower the barrier and provide access to individuals that don't have the money or access to expense machines and OSes, especially if the compute intense operations can be offloaded to GPU services in the cloud.
A few years ago (around 2015-2016), I got Counter Strike: Global Offensive running on my x86 Chromebook using Chrubuntu. The experience was about as good as one would expect, but it was still a fun time
stadia's biggest problem is stadia's business model. It's not solveable unless they change it. Paying for the streaming service and full price for the game too is a no-win situation and won't reach widespread adoption when microsoft is offering streaming for a large collection of games in their monthly xbox live subscription.
Steam supports playing games remotely on another computer. So your desktop is doing the heavy lifting, but you can view the game on your TV in the livingroom. Now you can be playing the game on your Chromebook as well.
Then the next step is that Google Cloud offers you an on demand high end gaming machine, pay by the minute, which runs Steam on Windows x86 and has an instant VPN to your Chromebook.
I set this up successfully once on an AWS machine and a very low end laptop running Ubuntu, but I couldn't get the hardware accelerated video encoding to work right. 3 seconds per frame made the game hilariously unplayable, but it did function.