Slashdot Mirror
Worrying Aspects of Linux Gaming
jones_supa writes: Former Valve engineer Rich Geldreich has written up a blog post about the state of Linux Gaming. It's an interesting read, that's for sure. When talking about recent bigger game ports, his take is that the developers doing these ports just aren't doing their best to optimize these releases for Linux and/or OpenGL. He points out how it took significant resources from Valve to properly optimize Source engine for Linux, but that other game studios are not walking the last mile. About drivers, he asks "Valve is still paying LunarG to find and fix silly perf bugs in Intel's slow open source driver. Surely this can't be a sustainable way of developing a working driver?" He ends his post by agreeing with a Slashdot comment where someone is basically saying that SteamOS is done, and that we will never get our hands on the Steam Controller.
Please move to gaming. It's the final blockade to me switching all the family and friends to Linux.
The old people were happy getting away from win 8 and back to a desktop that works, but the younger ones still can't accept Ubuntu.
Linux can and will in my opinion finally crush the other two os's - they are both fixated on the walled garden with "apps" to feed them.
Save us Linux, you are our only hope //the holograph figure turns and sees something, then fiddles with the holograph controls//
_ _ _ Go for the eyes Boo! GO FOR THE EYES!
I can't help but imagine you as a preacher in a Baptist congregation.
"I have SEEN the FACT-ORY LINE!"
I'm a guy who ordered a copy of Redhat all the way from Maldives back in 1996 (the shipping of which cost a bomb then), because it promised a new way to power our computers. From '96 till about 2007 I have exclusively used Linux in all my work. However, I've always had to keep a high-powered PC just for my games. With all the promise of different types of Wines and opengl implementations, games simply did not look as good or work as seamlessly (with few exceptions) as they did in Windows.
Since 2007, I have been using MacOS primarily for all my work, replacing all my Linux machines. Despite using Redhat, Turbo Linux, Slackware with Enlightenment, SUSE and Ubuntu, no Linux seemed to have the seamless productivity options boasted by the more mature MacOS or Windows applications, and some of these applications did not work proper with any of the Wines.
I think Linux as a desktop OS never really happened. I've mostly used it as my coding environment, and when I needed to author a document I swivel the chair and wake up the Windows (and these days MacOS) machine. All the various X, opengl and windowing implementations are just making applications ported to (or even originally developed in) Linux acquire quirks that aren't there in Windows and MacOS. Maybe instead of complaining about games developers, all the vendors should get together and conjure up a more unified Linux standard.
Dry Land does exist, I've seen it!
As someone who prefers almost any other OS other than Windows for my main, I still have problems believing that AAA gaming developers will make the big move to support an OS and framework that only covers a minuscule percentage of users.
I used to really be into running games under wine for Linux and OS X (osx86 *cough*), to the point where I would apply patches and do custom wine builds to get my favorite games running. I eventually just let go of it after 8 years and decided to always keep a Windows install ready for games and nothing else. That's probably why I haven't had much interest in SteamOS. It's a wonderful idea and I support it, but you need to win over the big players. The big players will most likely find SteamOS support to be financial waste. I'm glad that the amount of Linux games on Steam continues to rise, but like TFA says, it's just not worth the time and effort to optimize for Linux and in many cases not even try at all. There are many Windows applications that use a cross-platform framework that has wonderful support in X11, but why won't the company release a Linux or *nix version? Time, money, and less profits based on the amount of active desktop Linux users.
As the way things are going, SteamOS will be a great platform for indie games, that's for sure. But Ubisoft? Rockstar? EA? Activision Blizzard? I don't see that happening in the foreseeable future.
Perhaps I'm just getting old or something, but I've actually started to move closer to consoles, even with my nice PC setup that was the latest and greatest in 2013 which I keep around. The interest with indie game developers porting their PC games to PS4 makes me feel I made the right choice with getting one over an xbone. I wish the wave of Linux gamers receive the support they need to defeat this obstacle, but with its small percentage and the fact that it comes down to money and manpower to port and optimize games, it will unfortunately take some time for this to become a reality. If it happens, expect me to be there and ready to make the move from Windows when it comes to gaming.
Game developers are not Linux advocates. It is not their role to invest their time and money to displace Windows and Mac OS X with Linux.
Game developers create and sell games. Whether it is a Linux, Mac or Windows sale is irrelevant. A gamer who prefers Linux, but keeps Windows around for games, is already a customer. Letting that gamer move from Windows to Linux does not pay for Linux development, you are replacing a Windows sale with a Linux sale, there is no new money.
It is not game developers who are holding back Linux gaming. It is Linux enthusiasts who play Windows games that hold back Linux gaming.
Sure, the ports might not be optimised perfectly, but I don't care. You know why? Because I don't have to reboot to play them. Being able to use the desktop I like and still have games, even if not perfect, is way better than having to use a desktop I dislike or reboot every time I want to fire up a game.
Same reason I deal with wine, except the Linux ports generally "just work" which is also worth losing a bit of optimisation.
I still can't do the consoles. I bought one from every generation except the latest xbone / ps4 iteration. I just can't do it. I find the controllers bad for anything except fighting games and I generally like more complex games. Games that lend themselves to planning over days and often tabbing to a browser for insights.
I hate however having to boot to windows to play games. It drives me nuts. So I have a couple of linux native games I play but I mainly stream them from a windows machine via the steam client. It "just works", so my everyday machine is a dell latitude in a docking station running linux mint and I have an over the top gaming rig running windows in the garage. WOL and autostart steam. then a shutdown script. done.
The OpenGL API is fundamentally opposed to an efficient implementation. It allows developers to do fundamentally inefficient things (like dramatically changing configurations at the last second, before rendering, requiring the driver to recompile/reoptimise shaders and/or reverify states) immediately before rendering. Furthermore, it doesn't allow developers to do fundamentally efficient things (i.e. giving the driver a heads up about exactly what state/shader combinations it's going to use, so that they can be made ready at compile/launch time).
Good points. But while the API may not coerce you into writing performant code like (perhaps) the alternatives, it does not make it impossible or practically unobtainable. I will readily admit that I know very little about 3D programming, shaders and the like.
However, modern games are all built upon some form of game engine that in turn is typically used in multiple games. Few game developers write to the API anyway, so if the few (relative to number of games they support) game engines were optimized, wouldn't this difference go away?
Which brings us back to the developers of the game engines: If the developers of the game engines would invest the effort to create engines with high performance on Linux, multiple games would benefit from it immediately. OTOH, if the game engine developers *do not* optimize their code for Linux, there is very, very little actual game developers can do about it, short of creating their own game engine. Which is a monumental task.
Reading slashdot one-liner: (irm http://rss.slashdot.org/Slashdot/slashdot).rdf.item | fl title,desc*
Back when it was announced. No small number of people voice some choice words with me at the time about how Valve supposedly knew what they are doing better than I possibly could.
Honestly, I really wished, and admittedly even dared to hope that I would actually be wrong.
File under 'M' for 'Manic ranting'
Chicken meet egg. Egg meet chicken.
Both of you meet your chaperone, Valve, who are actually doing something to solve the problem of nobody bothering to port to Linux because "there are no other games on it", and thus nobody bothering to optimise for games "because nobody is porting to Linux".
More has happened in Linux gaming in the last couple of years thanks, almost exclusively, to the push from Valve than has happened in all the years before.
Something like a third of my 800-game Steam library runs on Linux now. That's bloody amazing. And they are all double-click-and-it-just-runs from the Steam client.
Those publishers too lazy to do this - are you telling me that they don't spot bugs in nVidia drivers and report them on Windows? Are you saying they don't spend a lot of their time working around bugs in drivers? Because for damn sure I've seen a lot of big releases have to patch like mad on day one when they hit all the ATI and nVidia and Intel bugs, and get bad performance reviews on certain chipsets etc.
Valve are DOING SOMETHING ABOUT IT. Whatever you perceive the current state to be, that's something to be applauded. And, to my eye, they've done a damn good job and not once have bitched about Linux beyond "look at this odd performance bug we found where a manufacturer never bothered to turn the optimisation on for Linux machines".
Epic's Unreal Engine 4 and the Unity engine both have Linux versions already. So does Valve's (obviously). EA's Frostbite engine has an OpenGL version, so that's part of the way. There's no market. Not a significant one, anyhow. Most people that are in the market to buy games either have a console, handheld, or a Windows/OSX PC. The vast majority. Then you've got the people like me, who dual boot all of their systems (so we're already customers, anyhow).
Then, over in a tiny little corner, you've got the Linux users with a gamer-grade PC, no OS but Linux, no console, and pockets lined with cash earmarked for games if only publishers'd release them on their OS of choice! Except it's "user", not "users". Yeah, that one guy standing in the corner. That's the market: people that want to buy games, want more than the (mostly Indie) games that have been released for Linux already, but won't (or can't) switch to a platform that has a larger selection.
It is pitch black. You are likely to be eaten by a grue.
Look, if you're going to keep pointing at me I'll have to bite your finger off.
You only need to look at the stats - Metal on an iPad Air will manage to run 3000 draw calls per frame, OpenGL, only 200.
In all fairness, Metal is an optimized platform-specific API, like AMD Mantle. It is certainly true that OpenGL is slower, but it provides a common HAL for many platforms, making porting applications easier.
Just tried Warthunder on Linux 2 days ago and was shocked to see that it simply worked like no other game on Linux ever before.
No crashes, runs in fullscreen mode well and yet I can switch workspaces without breaking it. Sound works well, incredibly fast on my old machine. Just amazing.
For me, this is a big milestone, because I am so used to Linux games ('bigger' ones) not working properly - especially on release day.
What I'd really want to see happening is that somebody would finally manage to be successful by consecrating on actual game content worth spending time on.
You know, I played my first computer games some 35 years ago - it's actually scary to think about those numbers; games like 'Colossal Cave' on a Cyber computer (http://en.wikipedia.org/wiki/Colossal_Cave_Adventure) or the first platform hoppers (character based on CP/M). What is really scare, though, is that content-wise nothing has ever moved since then. I don't give a toss about whether Linux has the very best driver for the latest ultra-, hyper-, super graphics card out there, because the games are still the same, old, tired re-run. It's like a $1000 gift card for MacDonalds - yeah, it's worth $1000, but on the other hand, it's for MacDonald's.
You missed one possibility, people who are enthusiastic gamers and want to game on a platform that is flat out compiled and optimised for playing games. That's not Linux today, but...if any platform can get there, it's Linux.
Windows is pretty megalithic. It's there to support general purpose computing and tries to be useful for everyone, and that's great. I work with Windows every day. I develop code for Windows. But I also run Linux and I can see a possible future where dual booting into a Linux optimised to play games at full tilt, low input lag, pedal to metal, no wasted cycles or memory is a reality.
Look into the work John Carmack recently did to make the Samsung Galaxy 4 phone capable of working as an Oculus Rift device. He cut through layers of crap, got access to the metal in some places and squeezed that phone for all it's worth to make it capable of displaying VR content.
Linux is flexible, multi-purpose and yet still capable of being fine tuned to a very specific workload. Steam OS has the potential to give us the stripped down, bare backing, fast and lean OS we need to get...a few more FPS and lower input lag :D
Oh yeh, if you think that's not too important, try out the VR gear that's out there now, and you'll start to see why input lag and high performance are likely to be key factors in the next few years.
All those moments will be lost in time, like tears in rain.
My Windows 7 laptop for work crashes 2 to 3 times a week with a reboot and a popup that says there was a "BSOD Error" After months of diagnostics by the help desk people and days of down time the best they can say is it looks like a "Video Driver issue" but there are no updated drivers. Thus I have to live with it rebooting and crashing tell microsoft/intel/someone releases a new driver.
Compared to my Linux Desktop at the house
07:42:39 up 316 days, 19:56, 7 users, load average: 0.97, 1.07, 1.20
Windows is still crap in my book and still "Blue Screens" on a regular basses they just call it a BSOD error and use a pop up to tell you instead of a blue screen. Hoping no one will ever realize that BSOD == Blue Screen Of Death
My Daughter has a Windows 8 laptop that I had to get her for school, in the last year we have had to re-install it three times. It gets an update that causes it to continually reboot with an error about the update failing and it rebooting. After the third time she brought it to me and asked me to wipe the crap and put linux on it. Why? Because in 10 year of her running a linux laptop it never crashed on her, yet the first windows laptop she had crashed every couple of weeks and the new one crashes to the point of re-install several times a year.
Don't give me the crap that windows is stable or good. Would you put up with a car that broke down twice a week? or even 3 times a year? Then why put up with windows doing it?
Companies barely optimize for Windows at this point (have you seen the minimum requirement for assassin's creed unity?).
Heck, some games have slowdowns on -consoles-.
And you expected them to optimize the Linux version?
Baby steps here cowboy.
I agree entirely, and that's what I was referencing when I mentioned that it could compete on performance. A full realization of SteamOS would be huge for linux, but consensus seems to be it is dead in the water. I really wish it'd take off because none of the other console makers have been daring enough to work touchpad-style input into a controller, which is the one thing that could bring RTSes and other strategy games to the console successfully. It would also make FPSes bearable to anyone whose ever used a mouse. I realize there would be a tearning curve, but the analog joystick input is inherently limited and the circle touchpad breaks that input glass ceiling currently in place.
I really wish it'd take off because none of the other console makers have been daring enough to work touchpad-style input into a controller, which is the one thing that could bring RTSes and other strategy games to the console successfully
The PS4 controller has a touchpad. The PS4, like the PS3 and PS2 before it, has USB ports for a reason.
But who says RTS's need mice? The first ever RTS was a Sega Genesis title! The only reason they use mice is that they're "designed" that way. It's quite possible to design them so that they don't need them. You really don't need pinpoint accuracy since you're lassoing units.
Heck I've played the PSone port of Red Alert. While it has PSone mouse support, it quite playable without it.
It would also make FPSes bearable to anyone whose ever used a mouse.
Depends on the FPS, slower paced more "tactical" FPS's are fine, faster paced ones benefit more from mice. It just so happens that bunny-hopping/skating/headshot centric FPS's tend to dominate on the PC, so PC gamers have a systemic bias.
I didn't feel the need for a mouse in SOCOM.... but I felt Timesplitters (faster paced) would have benfitted from it.
But it doesn't matter, since FPS's on consoles, can support mice if the developer chooses to do so. If they do, some people gravitate to hybrid setups, using the analog stick for movement, but the mouse for aiming. That works VERY well in the games that support it.
but the analog joystick input is inherently limited
digital WASD is inherently limited.
It is kind of weird reading all of the negative comments about the steam Linux experience because mine has been largely positive. After spending about a decade screwing around with wine to get various games to play often with limited functionality or weird graphics bugs, playing natively through steam has been amazing. I don't know what the Civ V experience was like on windows I guess but it looks great even on my 2 year old laptop. Beyond the indie stuff, I find it absolutely amazing that any AAA titles are getting ported to linux *at all*. Could it be better? Probably. Seems to me that is going to take time as the Linux gaming user base needs to expand to the point where it makes financial sense to go that extra mile. I'm willing to be patient and buy the ports as they come in, because that's likely the only way it is going to get better.
There is no engineering solution to this particular problem. The only solution is a market one. When customers buy games on Linux desktop at the rate of Windows desktop the game industry and hardware stack developers will care enough to put their A team on it (better yet they will hire more developers downstream to work on it). I don't believe this will happen and here is why:
We are in the middle of a platform shift today. The PC desktop is in decline and the players are fighting for a shrinking market. Mobile is saturating the market. A successful Linux gaming market people don't want to talk about is Android. Google provided a compelling alternative to the Apple ecosystem. Many hardware vendors had a limited to no market in the Apple ecosystem, Google provided an more open hardware ecosystem with developer credibility. Hardware vendors are now squeezing every bit of performance out of the mobile hardware today.
The Linux desktop does not have the same opportunity now, we kind of blew it a decade ago when PCs were relevant. We then blew it again when Netbooks were on the rise (started as Linux only at first), and then blew it again during the Windows 8 debacle (Chromebooks are our only success story here (similar to Android in this respect)). With all the in fighting about compositors, windows managers, incompatible kernel ABI, etc there is no compelling story. There was no real market to demonstrate who the winners and losers were and drive developer resources, so here we are in 2014 still arguing about stuff like Wayland, Mir, X, etc. The "free" free market leads developers to argue about dumb shit like GUI tool kits, syntax indenting, and init systems. Hardware vendors don't give a shit because they can't sell units based on these things. We are not solving problems that would grow their bottom line and thus they have no interest in growing our bottom line. This is simple economics.
Many Linux desktop diehards have moved to MacOS X which is has competent desktop, with a rich market of customers willing to play, and software library that is brimming with quality apps, with a UNIXey environment underneath. Apple demonstrated what we could have done if we had gotten our act together and the market has rewarded them.
Epic's Unreal Engine 4 and the Unity engine both have Linux versions already. So does Valve's (obviously). EA's Frostbite engine has an OpenGL version, so that's part of the way. There's no market. Not a significant one, anyhow. Most people that are in the market to buy games either have a console, handheld, or a Windows/OSX PC. The vast majority. Then you've got the people like me, who dual boot all of their systems (so we're already customers, anyhow).
Of course there aren't a lot of gamers who run pure Linux systems, because there aren't enough high quality games to be a gamer on a pure Linux system.
But if SteamOS becomes more popular then some Linux gamers who dual boot start going pure Linux (and playing more games because rebooting is less of a hassle). And some current pure Linux users who do a little gaming, and have been satisfied with basic Linux games, start buying modern games through valve.
It's not a huge market but there's enough that I can see them being interested, particularly since it gives Valve a chance to corner the new market.
I stole this Sig
Why, oh why do I always click, "Post Anonymously"? Seems I get far more +5s as an AC than as meself. The mods stop at +3 when I'm posting under me own name!
</lament>
Interesting fact: It depends on where you look at your posting and whether or not you have "Excellent" Karma.
There are two ways to look at your posting: From the article's comment section and from your own comment history on your profile.
If you're an AC, your comment gets a nifty 0 moderation. People need to upmod you 5 times until it's at +5.
... ... but that one is only visible on the article's comment section. In your own history, it will appear as a normal, regular +1 posting.
If you're logged in, you get an immediate +1 that is visible to you and everyone. People only need to upmod you 4 times until it's a +5.
If you have "Excellent" Karma, you get another +1 putting your posting at +2.
Since people stop moderating when a comment reaches +5, your own history will never show them as more than +4. ... even if the article lists it at +4.
If then someone downmods it only once shortly before the thread is archived (and moderation gets closed), your posting will sit at +3 in your history forever
You see, you don't even need Slashdot Beta for the posting system not to make any sense. :-D