Didn't SteamOS mostly become a hedge against Microsoft locking away first class access to their platform for Games from competitors like Valve and the titles they help sell/publish for developers?
In the years since SteamOS first launched the open source hardware drivers have also gotten much better. Valve continues to contribute towards better OpenGL and Vulkan (note, mis-spelled as a distinct brand name) support (new features, hitting current contemporary hardware support levels/etc).
The biggest issue that I can see is the number of development studios that only target Windows (currently 90+% of the market share); it tends to be better if they use a development kit that also allows for builds on other platforms, but if the developer doesn't have basic automated testing on either a Mac or Linux/BSD/etc system of some other kind then platform specific bugs easily slip in.
90+% without Linux? That's not how it feels like, though. Maybe it's because the genres I play are accidentally also the genres with the most Linux support, but it feels like beside the odd game here and there or an older incarnation of a game I like, everything is on Linux. And with most engines providing the easy choice to add another OS to build for, the numbers should be growing even more.
I also don't think it feels like 90+% these days. My own steam profile says 144/323 games support Linux, plus I've got some number more Linux capable games from bundles that I never imported into Steam but either played on desktop or in some cases the mobile version. (Then there's the "classic" Linux games from the dark ages that I really liked a lot like Battle for Wesnoth, Tremulous, and BZFlag.)
Of course of the steam ones, I've encountered several that either flat out didn't work on my system (probably the hardest part of shipping linux binaries is testing for many distros and setups, SteamOS/Ubuntu gets you most of the way there though) or they worked poorly. So most of my gaming is still in Windows 7 but I don't really use that OS for other stuff. Of those 144, I've played maybe half of them (my backlog is huge..) but mostly not on Linux. Either because they didn't support it when I played (Half-Life games as probably the oldest example) or I tried and it didn't work [well] or I just didn't even bother trying.
Even for the engines that don't support one-click Linux (or Mac) builds, it's not an insurmountable effort to support Linux, and if the game was already targeting multiple consoles+PC they've probably abstracted most platform specific stuff already. So it's just an economic issue. A lot of the technological issues have been mitigated or gone away.
This is what Carmack has to say on the subject of graphics APIs.
"John Carmack: Its still OpenGL, although we obviously use a D3D-ish API [on the Xbox 360], and CG on the PS3. Its interesting how little of the technology cares what API you're using and what generation of the technology you're on. You've got a small handful of files that care about what API they're on, and millions of lines of code that are agnostic to the platform that they're on."
And for every Carmack out there that can dream up whole new render pipelines in their sleep, there are thousands and thousands of games that start with a pre-made engine that in turn i deeply wedded to the DirectX platform.
Game engines, much like web browser engines, have become complex beasts. Hell, even as far back ad Wolfenstein 3D i suspect ID Software made as much or more licensing their engine to third parties as they did selling actual games.
SteamOS is based on Debian last I heard. Not that many games for Linux and at least 90 percent of video game companies target Windows. Until Linux gets more support, SteamOS will not be viable enough for people to buy.
Anyone remember the Indrema that was Linux based? It had some of the same problems.
Consoles routinely launch new machines with not-previously-existing operating systems and do fine. So "there aren't enough pre-existing Linux games" is not explanatory.
That's true. The issue is that no linux API has ever existed that can guarantee binary compatibility with at least $x (measured in millions) machines (for $40/game games, like the big consoles), or $x00 (millions) machines (for $5/game games, like phones).
To overcome the "no pre-existing software" consoles use two things:
1) First party studios/Second party publishing arm (even MS releases more 1st/2nd party games than Valve by far and Microsoft has the weakest 1st party out of three) .
2) Powerful 3d party developer support/QA (even Nintendo has more people in dev support than entire Valve's staff and N has the weakest 3d party support out of three).
A small company cannot possibly launch a brand new platform the same way consoles launch.
> Until Linux gets more support, SteamOS will not be viable enough for people to buy.
I don't see why this argument is reasonable. Windows is currently costing you about $100. Linux is currently costing you exactly $0. Linux has come so far, I doubt there is any area where it doesn't make you at least 80% satisfied. So the choice would save you about $80 of value, two triple A games in a Steam sale.
And I don't know what your choice of games are, but while there may be really 90% of steam product listings without Windows support most people also wouldn't buy/play about 90% of the products. If you look at the interesting games the feeling is more like 80% of what we need is there as well. And the rest will certainly follow once the user demand for Linux is there.
So, please continue to think what the real reason for not switching is. The providers would certainly follow. Most already have a Linux deployment pipeline ready and just wait for users to ask for it.
We're talking about gaming here - Steam machines had to compete against consoles as well. And they were more expensive, ran significantly less games than PS4 or XB and were slower to boot.
Consoles don't need someone to configure them and run system administratoon tasks on them. A video game console is less likely to break the OS and have a kernel panic or something.
Ok a better argument would be that GPU companies put driver priority on Windows drivers and don't do too much for Linux.
Some popular games like Civilization VI have been ported to Linux but they run better on Windows and faster for some reason. I suspect the Windows drivers are better optimized. I triple boot Linux Mint, Windows 7 Pro, and Windows 10 Pro. Linux Mint seems to play the Civ 6 game slower than 10 or 7.
Depending on your hardware (of course), many AAA games run fantastically well on Linux. I'm playing Mad Max for example, and it runs beautifully on a GTX960M, great framerate and no glitches that I can see.
The number of games in my steam library that have Linux/Mac support now is over 100 versus a few years ago where it was maybe 10 games. My gaming PC died a week ago, and I started looking into getting a new pc. But the thing is I don’t feel forced to do a Windows machine to have a gaming PC. I don’t play many AAA games, and am considering going with a Linux build.
I think the healthier ecosystem for non-windows games is what Value was going for. And it hasn’t stopped shifting towards that direction.
Valve is a lot like a small version of Alphabet in that there are no adults around to make sure people are dilligently finishing and polishing products.
I say this as an developer myself. After a 20 year career of working on other people's businesses and struggling to put my own things together, the whole, "we'll make amazing things if we just get away from management" thing is a gigantic, unproductive trap of thinking. Few programmers have the necessary self-awareness of their own limitations to stick to a problem all the way through to the end.
I think most people do not want to get away from good management. (Not to restrictive, not breathing down your neck, allocating time to do things right, some freedom, giving direction/focus to your company and projects, ...)
I work in a big corp's software science department and this is really how it feels like. Lots of awesome ideas, but everything is just 40%-60% done. And you can't really blame budget or upper management for this, since in our industry it's totally fine to come 10 years late to the party. In this market segment you may still be the first.
You should have lunch with the non-science department guys and see where they're at. I would bet it's pretty similar, not everything but lots of things only 40-60% done, PMs and some devs have lots of awesome ideas that can't all squeeze into the next planning horizon. The difference is things at least get shipped in that state and worked on incrementally with lots of flags to hide the roughest edges, ideally at some point things are worked on enough and declared 90+% "done" and only worked on in maintenance mode but there's always new things shipping half-baked.
I'm pretty sure that Valve rushed it, botched it and now it's waiting to have something substantial before giving it another go. They have been putting a lot of work on improving Linux as a gaming platform, so I'm fairly confident it isn't just something to keep MS at bay.
Disclaimer: I can't stand the Windows hegemony, period, much less their dominance of PC gaming (and I'm sure Gabe felt similarly).
Valve will need to dump a lot of money into driver optimization for 3D hardware in order to make this work. Windows 3D driver optimization, like it or not, is top-notch. I believe early demos of games on SteamOS were running at something like half the framerate (or less) vs. exact same game running on Windows on the same hardware.
There is also the matter of "is it a PC, or is it a console?" since the inertia of people's brains doesn't really have a defined space to put something in-between into.
Is this still the case? I hear from many people that at least on Ubuntu that's no problem anymore. Are the graphics cards still overheating? Any experience in that area?
And yes, Valve could've done things better, but it's not that bad, that you couldn't do anything with at. As said, at least on Ubuntu with Steam gaming should be fine. And if we got one thing from Steam it's that more and more games are deployed for Linux as well.
Really, why hasn't it lifted off now? Is it really that users are so stupid that they don't see a $0 offer is better than a $100 offer for an OS? Hard to believe.
Is that still the case? Last I checked the graphics support was there. Exactly in this time frame since SteamOS announcement they pushed NVidia and friends to provide better Linux support as well.
Yes, it's still the case. The Linux support is far better than it was a few years ago, but not even remotely comparable - NVidia (and I think AMD too, but I have a NVidia card) provide a new Windows driver with specific optimizations for almost every AA to AAA game. For example: The last Linux driver in the "short lived" branch came out at 09.05., the last Windows driver 29.06
Can't speak for nvidia, but AMD has been improving rapidly since they released full open source drivers. Ubuntu 16.04 won't work well with newer AMD cards, but 17.04 works extremely well and there are further improvements coming in future.
Anyone wanting to play games on linux should be buying AMD at this stage.
GNU/Linux and BSDs aren't a platform for gaming, it is quite easy to see on their conferences.
One goes to WWDC and BUILD, even Google IO, there are always a few tracks related to game development, libraries and debugging tooling.
GNU/Linux and BSDs are all about file systems, kernel drivers, license advocacy.
When graphics get talked about, it is usually about X, Wayland or how to get vendors to release specs, hardly motivating for professional game developers.
I eventually came to realize that regarding gaming, the OS X/Windows comunities are closer to what Amiga and Atari ST were, than GNU/Linux and BSDs will ever be.
The thing that happened with Steam Machines is the same thing that happened with Linux-based netbooks.
In the netbook case, vendors needed something that would run on lower spec hardware, and would be cheap enough to not be a major component of the machine's price. Windows Vista was to resource heavy, and expensive. But then Microsoft responded by re-releasing XP, for something like $10, but only for netbooks, so a couple of the advantages of Linux had evaporated.
Steam Machines were a response to Windows 8, and the direction that Microsoft was heading. After Steam on Linux became a reality, Microsoft responded by making Windows 10 more palatable than Win 8.
You also had Intel move in with Atom, and the two of them combined basically put down a stringent definition of what a netbook was supposed to be.
This in large part to avoid Netbooks eating into their lucrative ultra-portable platform that corporations kept buying for their road warriors.
The fear being that corporations would switch to Netbooks with a mobile data connection and simply access corporate computing resources using Citrix or similar.
Note btw that this was much the same idea that Google had when they first started pushing Chromebooks, to the point that one of their big launch demos were the ability to run Citrix.
Never mind that this also mothballed any plans for Android beyond phones. As seen with how even the 3.x "tablet" UI, that was perfectly adapted for landscape use, was gradually rolled back.
39 comments
[ 7.8 ms ] story [ 165 ms ] threadIn the years since SteamOS first launched the open source hardware drivers have also gotten much better. Valve continues to contribute towards better OpenGL and Vulkan (note, mis-spelled as a distinct brand name) support (new features, hitting current contemporary hardware support levels/etc).
The biggest issue that I can see is the number of development studios that only target Windows (currently 90+% of the market share); it tends to be better if they use a development kit that also allows for builds on other platforms, but if the developer doesn't have basic automated testing on either a Mac or Linux/BSD/etc system of some other kind then platform specific bugs easily slip in.
Of course of the steam ones, I've encountered several that either flat out didn't work on my system (probably the hardest part of shipping linux binaries is testing for many distros and setups, SteamOS/Ubuntu gets you most of the way there though) or they worked poorly. So most of my gaming is still in Windows 7 but I don't really use that OS for other stuff. Of those 144, I've played maybe half of them (my backlog is huge..) but mostly not on Linux. Either because they didn't support it when I played (Half-Life games as probably the oldest example) or I tried and it didn't work [well] or I just didn't even bother trying.
Even for the engines that don't support one-click Linux (or Mac) builds, it's not an insurmountable effort to support Linux, and if the game was already targeting multiple consoles+PC they've probably abstracted most platform specific stuff already. So it's just an economic issue. A lot of the technological issues have been mitigated or gone away.
http://store.steampowered.com/hwsurvey/
Scroll down, click on "OS Version" (you might need to enable some steam domains in your script blocker)
Windows has about 96%, Mac a little under 3 percent, Linux about 0.75%.
"John Carmack: Its still OpenGL, although we obviously use a D3D-ish API [on the Xbox 360], and CG on the PS3. Its interesting how little of the technology cares what API you're using and what generation of the technology you're on. You've got a small handful of files that care about what API they're on, and millions of lines of code that are agnostic to the platform that they're on."
http://fd.fabiensanglard.net/doom3/pdfs/johnc-interviews.pdf
Game engines, much like web browser engines, have become complex beasts. Hell, even as far back ad Wolfenstein 3D i suspect ID Software made as much or more licensing their engine to third parties as they did selling actual games.
Anyone remember the Indrema that was Linux based? It had some of the same problems.
A small company cannot possibly launch a brand new platform the same way consoles launch.
It's just, their corporate culture makes them very conservative about spending money. SpaceX they ain't.
I don't see why this argument is reasonable. Windows is currently costing you about $100. Linux is currently costing you exactly $0. Linux has come so far, I doubt there is any area where it doesn't make you at least 80% satisfied. So the choice would save you about $80 of value, two triple A games in a Steam sale.
And I don't know what your choice of games are, but while there may be really 90% of steam product listings without Windows support most people also wouldn't buy/play about 90% of the products. If you look at the interesting games the feeling is more like 80% of what we need is there as well. And the rest will certainly follow once the user demand for Linux is there.
So, please continue to think what the real reason for not switching is. The providers would certainly follow. Most already have a Linux deployment pipeline ready and just wait for users to ask for it.
Some popular games like Civilization VI have been ported to Linux but they run better on Windows and faster for some reason. I suspect the Windows drivers are better optimized. I triple boot Linux Mint, Windows 7 Pro, and Windows 10 Pro. Linux Mint seems to play the Civ 6 game slower than 10 or 7.
I think the healthier ecosystem for non-windows games is what Value was going for. And it hasn’t stopped shifting towards that direction.
I say this as an developer myself. After a 20 year career of working on other people's businesses and struggling to put my own things together, the whole, "we'll make amazing things if we just get away from management" thing is a gigantic, unproductive trap of thinking. Few programmers have the necessary self-awareness of their own limitations to stick to a problem all the way through to the end.
Valve will need to dump a lot of money into driver optimization for 3D hardware in order to make this work. Windows 3D driver optimization, like it or not, is top-notch. I believe early demos of games on SteamOS were running at something like half the framerate (or less) vs. exact same game running on Windows on the same hardware.
There is also the matter of "is it a PC, or is it a console?" since the inertia of people's brains doesn't really have a defined space to put something in-between into.
Is this still the case? I hear from many people that at least on Ubuntu that's no problem anymore. Are the graphics cards still overheating? Any experience in that area?
And yes, Valve could've done things better, but it's not that bad, that you couldn't do anything with at. As said, at least on Ubuntu with Steam gaming should be fine. And if we got one thing from Steam it's that more and more games are deployed for Linux as well.
Really, why hasn't it lifted off now? Is it really that users are so stupid that they don't see a $0 offer is better than a $100 offer for an OS? Hard to believe.
Anyone wanting to play games on linux should be buying AMD at this stage.
One goes to WWDC and BUILD, even Google IO, there are always a few tracks related to game development, libraries and debugging tooling.
GNU/Linux and BSDs are all about file systems, kernel drivers, license advocacy.
When graphics get talked about, it is usually about X, Wayland or how to get vendors to release specs, hardly motivating for professional game developers.
I eventually came to realize that regarding gaming, the OS X/Windows comunities are closer to what Amiga and Atari ST were, than GNU/Linux and BSDs will ever be.
My huge library of GOG, HumbleBundle and Steam games, some of which are AAA titles, begs to differ.
In the netbook case, vendors needed something that would run on lower spec hardware, and would be cheap enough to not be a major component of the machine's price. Windows Vista was to resource heavy, and expensive. But then Microsoft responded by re-releasing XP, for something like $10, but only for netbooks, so a couple of the advantages of Linux had evaporated.
Steam Machines were a response to Windows 8, and the direction that Microsoft was heading. After Steam on Linux became a reality, Microsoft responded by making Windows 10 more palatable than Win 8.
This in large part to avoid Netbooks eating into their lucrative ultra-portable platform that corporations kept buying for their road warriors.
The fear being that corporations would switch to Netbooks with a mobile data connection and simply access corporate computing resources using Citrix or similar.
Note btw that this was much the same idea that Google had when they first started pushing Chromebooks, to the point that one of their big launch demos were the ability to run Citrix.
Never mind that this also mothballed any plans for Android beyond phones. As seen with how even the 3.x "tablet" UI, that was perfectly adapted for landscape use, was gradually rolled back.