For a company that is wholly reliant on Linux to sell quite a few of their products, the lack of support from Nvidia continues to be impressive.
Intel is actively trying to kill off lowend and mid-range GPUs in the desktop, and has locked them out in the mobile space, leaving Nvidia with just the high end GPU market and their ventures into the ARM SOC space.
By not treating Linux end user devices as first class citizens, I don't see how Nvidia intends to have a viable business long term. Who wants a tablet, phone or desktop GPU without drivers?
A company that is wholly reliant on Linux to sell quite a few of their products?
Gaming enthusiasts aren't running Linux. The HPC/GPU compute market don't care about Wayland. Android devices aren't using Wayland.
What part of their target market runs Linux?
What portion of the desktop market runs Linux?
If it wasn't for Linux, I'd be in a very different profession, so I'm a huge fan. But I don't understand where you guys that act like the Linux desktop is a big enough market that this will make a material affect on nvidia's bottom line is so weird to me.
I'm in the automobile industry and most of the In Vehicle Infotainment systems started using Linux some time ago, and for security reasons you want to use Wayland. Those quite big numbers, not as big as mobile but sure bigger than Desktop Linux.
Not really because then you would need to implement composition from different processes yourself, instead many are using Qt Wayland https://wiki.qt.io/QtWayland to implement it. One of those implementations is https://doc.qt.io/QtApplicationManager/ which uses wayland to compose the UI from 3rd party applications, which run in a sandbox, inside of the System UI.
I see, on my naivety regarding your domain, I thought there would be just some kind of UI process/thread doing all the GUI stuff on the Infotainment UI.
Since I am only familiar with BMW attempts at JavaFX (described at JavaONE) and from what I am aware most embedded GUIs aren't X compatible, although there used to exist TinyX.
So is Qt kind of winning all the new GUI projects in that domain?
It's a bit weird, many try HTML5, some try Android, many do Qt but some run something with help of Weston. It's difficult to say for me who is winning because I'm not involved in enough different projects I feel.
Not to pile on, but every few years I decide to see if Linux is ready for gaming. I use it every day to get my work done, so why not double dip and not have to deal with Windows?
However it's never been good enough. At first there just were no drivers, so I couldn't get acceptable performance from any game. Then came drivers, but I had to fight for days to get them to work on handful of games and even then performance was sub par compared to Windows. Last time drivers installed without any problems, but still from the games I tend to play only couple had native Linux support (in this case, they could be downloaded straight from Steam), but still most required Wine and that sucked the performance right out. Then there were bunch of games that didn't even work with Wine.
I'm first to admit that I do not have high end gaming PC, but 16GB of middle of the road RAM, i5 2500K, and nVidia GTX 970 serve me well on Windows side. Frankly, I see no reason to spend more money to get the same performance on Linux.
If you can get acceptable performance from your Linux PC while gaming then more power to you, but from my point of view we are still long way and I mean really long way from gaming being "viable" on Linux.
Portal, Portal2, Tomb Raider, Kerbal space program, Shadow of Mordor, Rocket League, The Talos Principle are some of the AAA games that I've played on my linux mint box in the last few years. I am just an occasional gamer and I do not have a gaming machine (amd fx8350, nvidia gtx770, 1GBram) but all these games worked flawlessly. No wine, no glitches no driver problems. I don't know if it would have been "good enough" for professional gamers (I admit I play on "high details" instead of "ultra high details" mode)
The terms gamer and gaming community are used by people to lump anyone who plays video games into a small digestible nugget. There is no single gaming community and the sheer size of each of the countless gaming communities dwarf most other hobby or enthusiast communities.
Most people promoting gaming on Linux are so out of touch that they think having Tux Racer playable makes a Linux desktop computer a first or even second class gaming platform. Having The Witcher 3 or Tomb Raider on Linux is a novelty or a way to win brownie points.
I'm not a serious gamer by any means, my highest played game only logs 403 hours of playtime, I've see people with ten times that. With that in mind, none of the games in my top 10 list in terms of hours (~1800 hours) played are available on Linux.
I generally agree with this sentiment. The way I see it, if you have to make concessions for a platform, then that platform is not really a very viable option. If I want to play X, but X is not available for my platform, telling me "Well Y is supported, and it's also a AAA game!" That doesn't really help me any, because I don't want _Y_, I want _X_.
I feel like a large part of the group making who believe Linux is gaming ready doesn't understand that angle.
It's a chicken and egg problem. Gamers don't want to adopt a platform that will hinder them and studios don't want to support a platform no one is using.
It was in the past, but it's not stalled anymore, since you can observe the gradual growth. I.e. more games are coming out for Linux, which in turn makes more people use it, which in turn brings more games. So it's improving.
The two most played games on steam have a decent linux port.
At this moment they make up nearly 1.3 million concurrent players. The rest of the top 10 games that do not have a linux port accumulate close to 0.7 million concurrent players.
For a lot of players, CS:GO or Dota 2 are the only games they play regularly, so linux is a viable option for them.
For those who play CS:GO and Dota 2 competitively, Linux isn't a viable option because CEVO and ESEA don't have clients for Linux. That's ignoring all of the bugs and quirky behavior that come with the platform. ALSA has a 10-100ms delay and for a game like CS:GO that relies so heavily on audio feedback, that's inexcusable. There are also issues with mouse input, lag, and random frame drops on Linux that have lingered for years.
You'll see all kinds of posts showing Min/Max FPS and how Linux beats Windows at this game or that but you rarely see any mention of Average FPS because they're usually pretty awful.
I never said anything about PulseAudio, I said ALSA introduced delays. JACK uses ALSA the same as PulseAudio.
In any case, it's the game developer's choice and not mine. Yet another problem Linux has for developers is that there isn't always a clear choice and the flavor of the day may not be the best choice long term. That's why frameworks like Qt and Unity are so prevalent because they abstract a lot of the crap away.
Rocket League started out that way, but I'm not sure its still that way. It started small, but is now one of the most popular online games and competitive e-sports - to the point where actual soccer clubs are supposedly setting up e-sports teams for it. Psyonix is building an empire around it.
> but is now one of the most popular online games and competitive e-sports
Are there any numbers, I'm actually curious. I don't regularly see it on the Twitch.tv front page, so my impression was it was a bit behind the big names in esports (LoL, DotA, CS:GO, SC2, Hearthstone, Overwatch).
That's like calling a small Indie film a Blockbuster and dismissing objections as a matter of opinion.
The term AAA is analogous to Blockbuster in the film industry. It's a reference to both budget and expected unit sales. Blockbuster movies have budgets starting in the hundreds of millions where as Indie films budges are usually in the single to double digit millions. Video games are no different in that respect.
No, it's like saying that good art doesn't equal to whether big publisher funds it or not, and millions in budget can result in trash result.
That's why I don't like the term AAA, it's way too confusing, conflating unrelated and often orthogonal issues (budget, publisher funding vs independent development, quality and so on).
Anyway, there are big budget games coming out for Linux, so above is really besides the point.
It has nothing to do with the quality or content. There are plenty of Hollywood Blockbusters that are awful, most in fact, but they still cost a lot of money to make.
Budget is a good indicator of audience numbers regardless of quality.
I mean artistic quality, as in good art in general. And why should we care about awful games? I buy games which are interesting to me, not based on their budget size.
I understand what you're saying but the vast majority of people playing games do not share your sentiment and game studios go where the money is. Games like Braid, while beautiful, innovative, and memorable, don't bring in the money of the annually released cookie cutter FPS games like CoD or Battlefield that are shallower than spilled coffee on your desk.
You can say that there are plenty of good games on Linux and everyone should switch but the reality is that the games people want to play are either unavailable or lacking in some way that makes Linux on the desktop a non starter. C64 and Amiga probably have a larger base than Linux.
> Games like Braid, while beautiful, innovative, and memorable, don't bring in the money of the annually released cookie cutter FPS games like CoD or Battlefield that are shallower than spilled coffee on your desk.
In way I don't mind that Linux gamers are getting better games on average, because of less mass market mediocrity, but more interesting (usually crowdfunded) releases that are reaching Linux gamers today. However it ties to the next point, of the market size:
> You can say that there are plenty of good games on Linux and everyone should switch but the reality is that the games people want to play are either unavailable or lacking in some way that makes Linux on the desktop a non starter.
That doesn't describe the situation well. Linux gaming is for sure a much smaller market than Windows gaming, but not only it's already viable (as many developers who sell Linux games demonstrate), it's gradually growing. So it's surely not a "non starter" anymore.
However, since it's smaller, having too many games can actually cause a problem, even if it sounds funny. Simply because it will be overcrowded developers wise, and they'll have harder time selling Linux games. So I'd say actually current amount of games healthily corresponds to the rate of Linux gaming growth.
> In way I don't mind that Linux gamers are getting better games on average
Who says they are? There's plenty of crap available for Linux. I mentioned Braid because it's a very popular and very good Indie game available on both platforms. I surely wasn't implying that Linux only receives ports of the finest of games.
Also what makes a game "good" is subjective. From the perspective of hours played, Braid isn't very good.
> That doesn't describe the situation well. Linux gaming is for sure a much smaller market than Windows gaming, but not only it's already viable
It is viable for some small developers and for larger developer who are already using tools ported to Linux it can make sense to add support however for the majority of game studios making popular (not necessarily good or bad games) it doesn't make sense.
Crowdfunded games are more commonly released for Linux than publisher funded games, and they more often represent not the mass market, but something interesting that creators want to pursue. So for me it translates into better games on average on Linux. But as you said, it's subjective.
The market however is viable for everybody, but as I said, if everybody from the Windows gaming market will start releasing games for Linux now, it can actually hurt it, rather than improve it. Gradual growth is better, and it's already happening. Tools wise, the situation is getting much better as well, with all major engines gaining Vulkan support. Common middleware like Umbra 3D also often supports Linux these days.
Nothing you said is wrong, but it doesn't change the fact that quality doesn't sell. Once a game is "good enough" it often comes down to marketing, and the studios putting out for Linux do not control a majority of the marketing dollars.
> Portal, Portal2, Tomb Raider, Kerbal space program, Shadow of Mordor, Rocket League, The Talos Principle are some of the AAA games that I've played on my linux mint box in the last few years.
Here are the release dates with linux support date.
Tomb Raider: 2013, 2016
Middle-earth: Shadow of Mordor: 2014, 2015
The Talos Principle: 2014, same day
Here are some other random selection:
Borderlands 2: 2012, 2014
Europa Universalis 4: 2013, same day
Pillars of Eternity: 2015, same day
Civilization 5: 2010, 2014
Civilization 6: 2016, 2017
Factorio: not yet released, already supports both
There is a clear trend towards more games with linux support and less lag between windows release and linux release. My impression is that the tipping point for newer games is already in favour of linux support than not linux support.
If you want to argue against linux gaming, I would argue with slightly older games that didn't receive a linux port and maybe never will.
That's one of those xckd which are missing the point, and it's not uncommon for the author to have pointless sarcasm. That's why I mostly don't find his comics interesting.
Along with what others have said, this is also due more to Valve's push for SteamOS to get greater adoption, and their push to have developers of engines like Source and Unreal prep for linux too. This has extremely little to do with the Linux projects individually.
Steam has plenty of games for linux... but few of them are the kind of AAA titles that benefit from chunky GPUs. I use steam both on windows and linux, so see the store listing for each; linux 'is there' for gaming, but just doesn't have the offerings to drive desktop GPU demand.
Yup Linux definitely would have "enough" games for even a heavy gamer like myself, but it wouldn't have the freedom of ANY game that I enjoy using Windows, tragically.
Yes, for sure there are plenty of games for you to play, if you are mostly into indie type games. It might even be good for mostly story driven single player games. I'm probably out lier since I exclusively play multi player games, because for me video games are just replacement for out board game nights which we can no longer maintain constantly since my friends and I live now across the country.
Only "native" Linux games I've played are Counter Strike: Global Offensive, Dota 2, and Insurgency. Beyond thous only game I've gotten to run smoothly enough that I could even consider playing them is Minecraft (which obviously runs even on a potato). But even with these I had some problems. CS:GO's FPS didn't quite keep above 120 all of the time and Dota 2 had few graphical glitches.
Most of the games I play on Linux very much benefit from a beefy GPU. I'm currently playing Hitman (2016), which may not be the most graphically intense game ever, but it certainly looks great with the right hardware. I could run it on my laptop with on-board Intel GPU, but it would look like crap and lag.
We do get our share of AAA titles, just not as many as on Windows. As it is, there are more appealing titles than I can ever play without cutting into work, chores, or (non-gaming) leisure time.
Most indie games benefit as well though; even though these do tend to work on less well-endowed GPU's, they simply look better with a heftier graphics card. I've got a desktop computer with beefy GPU exactly because I enjoy gaming; and it runs Linux, exclusively.
I'm in the same boat. The last AAA titles I played on Linux were Deus Ex: Mankind Divided and Alien Isolation - high end, graphics intensive games. I've been meaning to get around to Hitman - I downloaded the demo a few weeks back and it seems to run fine but not had time to get started with it properly.
I also enjoyed Thimbleweed Park and I see the (excellent, btw) Wonderboy remake was released for Linux this week. Us Linux gamers are incredibly well catered for these days!
Without high end gaming PC, it would be tougher on Linux, as you noted since in some cases performance can be lower, and translation layers like Wine or wrappers can introduce extra overhead, which higher end hardware helps to mitigate.
However, it's more than good enough for many games already, and Linux gaming keeps improving and growing. And not all games need super performance.
I can get around 40-50 fps in the Witcher 3 in Wine on Linux on max settings (minus hairworks), using AMD RX 480 at 1920x1200. Is that acceptable performance? It's somewhat lower than it would get on Windows on the same hardware, but it's pretty playable. More critical are rendering bugs that Wine developers are still working on.
For sure, 40 fps is playable, but at that point you might as well just buy a console and play the games with that with no hassle and run Linux for other purposes. It will probably be cheaper in the long run.
However I've spent money on 144Hz monitors and for enough performance that the games I play run on at least constant 120 fps. Obviously it's not necessary if you are in it just for a story, but as mentioned in another post I exclusively play multiplayer games with my friends and if you haven't experienced playing games (or even just using a computer) with 144Hz (running at or higher frame rate) it's like a night and day.
> but at that point you might as well just buy a console and play the games with that with no hassle and run Linux for other purposes.
I stay away from DRMed platforms, so incumbent consoles are no go for me.
About higher framerates - it should be OK, once AMD will merge FreeSync support in amdgpu driver upstream. My monitor is 60Hz anyway, but to have good experience without tearing on high frequency monitors, FreeSync can help. Today, to use it you'd need to install customized kernel from AMD.
As I said earlier we are still long, long way away from Linux gaming being a thing and longer it takes for Linux to come to parity now the further Windows will race ahead if you believe all in the VR hype that's going on.
>I stay away from DRMed platforms
I am not ideologically bound, I care about performance. If Linux is enough for you then more power to you, I wish it could be enough for me too.
Vulkan is already available, which removes any advantage that Windows might have had in the past beacuse of DX being ahead of OpenGL. VR related development in Vulkan drivers / Linux graphics stack is now active as well.
If you care about performance, you'd stay away from consoles anyway. They are way too limited in comparison.
So I don't agree that Linux is much behind technically. Smaller market share is more of an issue.
I've heard about vulkan for quite some time now, still haven't seen it used widely. Currently I do not see a future where Linux would be on equal footing with Windows when gaming is concerned. Linux would have to become way easier to use and have something that was strictly better than Windows. Free software and other ideological stuff won't make gamers to switch. What Linux needs is image and video editing suites to rival and surpass Photoshop, AfterEffects, Premiere, Final Cut, etc. Some streaming technology that is strictly better than anything available on Windows (so I guess OBS should drop Windows support or something) and then games would need to run at least on par. Maybe then some of the influential people (youtube personalities, twitch streamers) would switch over and bring their fans with them.
As long as Linux is just sort of viable there won't be wide adoption, because why would you change everything about your computing life for something that's +/-0% gain. Even at 10% better experience you might have to consider hard if changing over is worth the effort. Linux needs to magically become superior in many ways for people to hope on board in large numbers.
> I've heard about vulkan for quite some time now, still haven't seen it used widely.
It's recent, so its usage is gradually growing. Unity engine now supports it, with Unreal and Cry on the way. Once major engines will have it, it will implicitly be used in much larger numbers.
I have no idea how game development works, so would this mean that you can just compile Linux and Windows binaries from the same source code? Or is it just easier to make it multi platform? Also is there any thing like with Xbox vs Playstation were you have to do some platform specific tricks to get the performance up or is that way further in the line for optimisation?
It would allow using higher performance design where it's supported. Walled garden platforms like PS and Xbox will require duplicating the work anyway, thanks to MS and Sony being lock-in jerks in general. It wastes engine developers' efforts for no good reason, but there isn't anything they can do about it.
>I stay away from DRMed platforms, so incumbent consoles are no go for me.
Yes, but we were talking about Linux's popularity as a gaming platform, and why it's not more widespread, and that's not an issue the huge majority of gamers share -- so they'd rather use a console.
Consoles vs PC has nothing to do with Linux popularity. Linux is underrepresented in gaming because of smaller market share, which in turn is caused by sickening dominance of MS, and their efforts to push lock-in on developers (DX and such). I.e. still the vast majority of PCs are sold with Windows pre-installed. That's the root of most of these problems.
Not really, I'm not interested in running Windows. I do run some Linux guest VMs on Linux host, mostly for building purposes, in order not to mess up the main system :) And passthrough itself is a interesting method of course. Now there is also SR-IOV coming which would allow using the same GPU for virtualization more effectively.
That's fair enough. I'm like you in that I currently dual-boot but most of my gaming time is spent playing Dota 2 so I rarely need Windows. SR-IOV seems to be very interesting!
I just stopped gaming on the PC, so no more need for windows :)
I do own a couple of gaming consoles, but now that I have less spare time, my gaming preferences have definitely shifted toward casual games I can pick up and play for 20 minutes and then forget about for a week (so I'm a Nintendo fan these days).
This is the wrong mindset, and if you take it, Linux will never be better by your criteria.
Linux is not a 1:1 replacement for Windows. It is a separate platform. In the same way you can't expect to play all PS4 games on Windows, you can't expect to play all Windows games on Linux. Retune your expectations.
Just means I won't be playing on Linux since if not all most of the good games live on Windows. If I take your perspective, that Linux is separate platform, like console vs PC. Then the battle is already lost and there is no point for me to care in the slightest.
Many, many great games are available on Linux, and the platform has a lot of compelling reasons outside of games to choose it over Windows. And unlike any other gaming platform, Linux has software like Wine to expand its game library into those of other platforms. And unlike the typical console vs PC wars, Linux is still a PC and has all of the benefits of using one. But it sounds like you're just looking to justify your confirmation bias, and so be it.
The console wars are not won by the games available on both, but by the exclusives.
Unless Linux has exclusives (which it doesn't afaik) then it has lost the battle completely and utterly. Windows is the default choice for gaming PCs and unless something drives a significant change that will continue to be the case as linux is clearly a second class citizen.
Exclusives are increasingly dying out. For independent developers, there is no point to limit their distribution to some specific platform. It only decreases their profits.
Most games I play do not have Linux port, so by definition I would be spending most if not all of my time in Wine, which drastically lowers performance. All examples people have thrown at me have been single player story driven games which do not interest me in the least.
I only play multi player games with my friends, since single player games rarely keep my interest (last single player campaign/game I finished was Portal 2 since it was so short)
Then there's always "will this even work with wine" question. Like take any Blizzard game, the downloader/launcher simply does not work on Wine (or at least it did not when I last tried it). When a new game comes out that I want to play, I don't want to think if I can play it, I want know I can if I choose to. This is even bigger concern since most interesting games now days are not AAA titles, but various smaller games, mods, and early access titles. Game we've spent most time playing last month is easily Playerunknown's Battlegrounds and with quick look at Wine DB it's status is "garbage"
If I look Linux as separate platform then there is no hope I would be using it to play games.
I have had performance decreases and increases with Wine. I would not call the change drastic. The larger issue with wine is that not every game works.
I encourage you to measure both and see what the performance looks like in both places, then see which one gives you better options for improving that if performance really matters.
This is exactly the stupid mindset I'm talking about. You can play different games! It's okay, really, to play different games! Other games are fun too!
Here's an incomplete list of multiplayer games that run natively on Linux, sourced from my own steam library and the popular linux games on steam:
- Rocket League
- CS:GO (and CS:Source)
- Rust
- Garry's Mod
- Terraria
- Insurgency
- Borderlands 2 (+ pre-sequel)
- Civ V
- Left 4 Dead (1 & 2)
- Dota 2
- Torchlight II
- Team Fortress 2
- Battleblock Theater
- Natural Selection 2
- PAYDAY 2
You like multiplayer games? Great! There's loads of them on Linux, and they're all super fun. You don't need to play the same games as Windows to have fun.
You're missing my point. Does Linux have fewer games? Yes. But Linux is absolutely a viable gaming platform and the other benefits to using Linux over Windows are extremely compelling.
You're missing his: that he feels (or wishes to explore the possibility) that it's a strictly inferior platform because it's not "extremely compelling" unless you've already bought in. You do not get to tell him that those benefits are "extremely compelling". That is a relative thing and you don't get to tell him what he finds compelling.
This arrogant prescriptivism is what you did last time, too. You did this last time this came up by trying to slag me because I didn't want to waste hours of my life debugging weird edge-case issues with Linux as a desktop operating system (and, funny story, Ubuntu 16.04 hard-locks on my desktop when I plug in a bog-standard USB sound card, that's cool, and has DVI timing and desync problems between my monitor and the graphics card--this is fiiiine, everything is fiiiine). You have decided it's compelling for you. That does not mean stating "it's compelling" makes it so for a single, solitary other soul.
Or don't care and use something else. I now run a literally problem-free Hackintosh dual-boot and life is great. With hardware better supported on a literal hack system than on Linux. (Didn't buy hardware intending for a Hackintosh, either; I just happen to tend towards Gigabyte motherboards and Intel processors and everything shook out great.)
Linux isn't "my software", so why would I buy hardware for it--especially when nothing else has a problem with it? Your OS is not special no matter how much you yell at people on HN about it. Your OS doesn't matter to anyone else no matter how much you yell at people on HN about it. Attempting to put the onus of "due diligence" on the people you want to do something is just straight-up nuts--and, tbh, the same rhetorical evasion that's come out of the Linux zealot camp for decades now because it's easier than acknowledging problems and that the things you like aren't universally appropriate.
And the thing is, you actively are providing great rhetorical reasons to not listen to people who do have a sense of user empathy--which you lack, it seems, as you would rather browbeat people into using it than understand why they don't and maybe make it a better value proposition--and want people to use Linux. You're literally, precisely acting against the interests you purport to hold!
I'm not in any kind of hole here. Maybe you're projecting your experiences with other Linux users on me. I'm clearly explaining my viewpoint and reasonably giving answers to critiques of it. If Linux is not for you, then so be it, but I disagree with your reasons and I am free to express that.
Re-reading this discussion only shows one of us being unreasonable and abrasive. I don't care to engage you any further unless you change your attitude.
Every time you post about desktop Linux here the starting point is "people should be using this because I like it." Criticisms are deflected with "well, don't do that"--guess what? People are gonna "do that" because they do it with literally everything else. "Do some due diligence on this thing I want you to use" is as unreasonable as it gets, champ.
But it's me being unreasonable. And abusive--I laughed aloud, dude.
The implication here is that should you find the advantages of Linux compelling, you would factor that into your consumer choices. "I bought a hundred gallons of desiel but I couldn't get it working in my sedan so I'll just stick to the truck"
- Can enjoy a smug sense of satisfaction as Windows users are hit by ransomware
- Does not spy on you
- Runs great on everything, including >10 year old hardware
It's also not driven by corporate interests. The #1 goal of Microsoft is to make money, and they'll put that before the needs of the user. Things like Windows 10 forced upgrades, spyware, ads in Windows explorer, etc - none of this would fly on Linux because Linux is made by people who just want the best possible operating system, not the one that monetizes uses the most effectively.
Linux is an amazing server platform and mostly sucks as a desktop and I'm okay with that. I've been using Linux professionally for well over a decade and I've used a Linux desktop a handful of times. I don't expect that to ever change.
In my opinion, that's complete bollocks. Windows is awful as a desktop platform, especially for a developer. The dev tooling sucks compared to Unix tools. The package manager is brand new, has very few packages, works poorly, and 99% of the time you install something it won't use it. You can't dig into the tools to fix any issues you find. The terminal emulator - and the shells - suck.
As a user, it depends on what you're looking for. I bet most end users would have no problems on a Linux desktop, especially one like Cinnamon or KDE Plasma.
I have no idea what you're talking about, Linux on the desktop is great.
Anything up until Windows 10 - I'd have agreed with you. But Windows 10 is a whole different ball game. PowerShell is finally where it needs to be, native Docker, they have an official Ubuntu sub-system with various tools.
Honestly, this whole "WinDoze is lame" attitude is so 2008. Windows 10 is a great platform. And I say this as an Arch Linux user who just wasted 2 hours dicking around with DBus and NetworkManager because there are still 3-year-old unfixed bugs regarding Cisco VPN connections.
I will not use any OS that serves me ads and I have to pay for it. I will not use an OS that spies on me. I will not use any OS that is not open source so I can see what it is today should I need/want to (and I will gladly investigate it for others if you can convince me... $).
I really don't know why people like windows beyond having their applications (That really is enough). It is limited, the tools they make are clearly more for checking off boxed on some marketing list somewhere because they rarely work right and often limited in their core performance. Even their tools like their expensive compiler and database does work as well for my worklaods as their free and open source counterparts.
It wasn't that long ago we saw the remotely exploitable flaw in Windows Defender that allowed arbitrary execution of code in a security product! Or more innocuous but no less annoying, the guy with the umpteen core ryzen getting blocked by some random mutex deep inside windows so his build was being throttled and he couldn't use his own CPU the way he wanted. It buggy, its lower quality, but they succeeded in having many make their software for it exclusively so it is
Then we can start talking about the ethical issues with it, like giving money to a monopoly that has tried to bribe standardization bodies and has been convicted several times of anticompetitive practices.
I've used Linux on the desktop since about 1999. It's not "great"; it has a lot of problems. But it is indeed better (IMO) than the alternatives, which are MacOSX (proprietary, tied to limited and expensive hardware), and Windows (a complete horror show, esp. now with Win10--spyware, ads, a ridiculously horrible UI when you click "start", and of course the total lack of decent utilities and shell and the horrifically bad-to-nonexistent package management you mention).
The main problems with desktop Linux as I see it are 1) crappy DEs (namely Gnome3, the most popular and pushed hard by Red Hat) and fragmentation in the DE space, video driver problems, and power-management problems (particularly on laptops, esp. with suspend/resume). My laptop also has a lot of trouble going from docked (w/ dual monitors) to undocked and back. This stuff actually works pretty well on Windows, but on Linux it's a real PITA and frequently requires reboots IME, and that's on a not-new Dell business laptop that's generally well-supported.
So... Linux is the worst desktop, except all others.
I agree with most of your gripes, but fragmentation is more an more just becoming a reality. UIs won't look the same anymore, it just doesn't make sense to. So people will experiment with different things and try to find something better, the alternative is to stagnate with accepting what we have.
Fragmentation, can be restated as "Community diversity" and it ain't going away because its a huge strength. Because while you think gnome3 sucks if Canonical had their way everyone would be using it.
UIs don't need to look the same, and there's no reason they should. Choice is a good thing. The problem, however, is a lack of interoperability: Gtk apps look like shit in KDE and vice-versa. There's no good reason for this; it's possible to make it so apps with different toolkits blend in seamlessly, but the Gnome devs are infamous for being completely unhelpful here and having a "our way or the highway" approach to everything.
>Because while you think gnome3 sucks if Canonical had their way everyone would be using it.
Not true; if Canonical had their way, everyone would be using Unity. Honestly, I don't begrudge them making Unity; it's good to have some diversity, but I do begrudge them switching to Gnome3 recently and abandoning Unity altogether. A lot of people liked Unity and its concepts. Instead of throwing in the towel and becoming yet another me-too Gnome3 distro, they could have adopted KDE, and then created a custom KDE theme or version of Plasma which looks and works like Unity and allows them to continue to explore those concepts which leveraging the existing KDE underpinnings and the superior Qt toolkit. The real problem IMO is the horrible Gtk toolkit: it's in C but bolts on a hacky OOP concept, it's undocumented, and it's constantly changing. I know of at least one commercial project which had to abandon it and switch to Qt because of the constant API deprecation, and there's at least one FOSS DE (Lxde I think) which also switched from Gtk to Qt for the same reasons, and got a performance boost in the process IIRC.
>the alternative is to stagnate with accepting what we have.
I can't even count the number of comments I've read from people who are sick and tired of the constant change and churn in Linux UIs. Lots of people were perfectly happy with Gnome2, but then it was deprecated and they were pushed to Gnome3 and it was terrible and they hated it. As a result, not 1 but 2 new DEs were created, just to satisfy people who wanted something more stable: Cinnamon and MATE. It's a complete mess. A lot of people were similarly pissed when KDE4.0 was released and they couldn't use KDE3 any more because the distros all deprecated it, and the new version was filled with huge bugs and didn't work. At some point, you need to aim for product maturity instead of constant churn. I've been using ssh for many, many years and I don't see that being deprecated any time soon and having to switch to some other standard, but somehow we're supposed to accept this as normal for DEs?
3 months ago I would have agreed but they have announced that for their next LTS 18.04 they are dropping unity in favor of Gnome3.
I agree that improved interop would be good, but I am grateful for how well it does work. Copy/paste and other common shortcuts and features work. But, I use KDE and it sucks when I click something and it tries to open nautilus and clobbers my wallpaper. A "killall nautilus" later and everything is fine, but its annoying and a non-power user couldn't do this. The new the changed right click menu on the desktop would be really weird for the uninitiated too.
> I can't even count the number of comments I've read from people who are sick and tired of the constant change and churn in Linux UIs
It's not just Linux UIs Every UI is advancing; my friends and family are constantly asking me for help with win10 or whatever crazy mac/iOS/android shit they dig up. I explain I never used it, but I can form better google searches than them so I can figure out how to do what they need (I also don't mind this because I like learning this stuff too).
The root of this is how does a UI designer improve the UI while keeping it familiar enough for previous users? How do you keep power users and people on the cutting edge happy without losing the technophobic grandparents and those slow to learn?
>3 months ago I would have agreed but they have announced that for their next LTS 18.04 they are dropping unity in favor of Gnome3.
No, it's still true. They didn't want to drop Unity at all; they did it because they were forced to. Didn't you read Mark Shuttleworth's blog postings about it? He was clearly unhappy about things. They gave up on Unity and Mir because they weren't working out in the market and it was costing them too much money to pursue that direction. If Mark had his way, Unity would be the most popular Linux DE.
>I use KDE and it sucks when I click something and it tries to open nautilus and clobbers my wallpaper.
Huh? I don't know what to say to this. Nautilus is Gnome software, not KDE. Why would KDE try to open Nautilus for anything, unless you've explicitly set it up that way? Now if that's what you've done because you like Nautilus, the problem is probably with the Gnome/Gtk devs. As I said before, they're infamous for not being amenable to cross-toolkit efforts. I would encourage you to ask them for their advice on this problem; they'll probably tell you "switch to Gnome3". :-/ I use KDE too, and personally I avoid GTK software if I can help it.
>A "killall nautilus" later and everything is fine, but its annoying and a non-power user couldn't do this.
I agree entirely. It's garbage, and it shouldn't be this way. The Gnome devs are entirely to blame for this mess, along with all the distros that push Gnome3. They don't even want you to use a wallpaper with Gnome3; Jon McCann said so some time ago, saying that customization (even wallpapers) is bad because it makes different Gnome installations different.
>It's not just Linux UIs Every UI is advancing; my friends and family are constantly asking me for help with win10 or whatever crazy mac/iOS/android shit they dig up.
Oh I entirely agree; all the UI people across the industry have gone collectively nuts. But I expect better from FOSS, for multiple reasons: 1) FOSS isn't for-profit (well, I guess Red Hat is, maybe that's part of the problem), so they shouldn't need to push unnecessary churn to justify pushing customers to "upgrade", and 2) FOSS is usually very resource-limited compared to commercial software, so unnecessary churn is a waste of developer time that's better spent fixing bugs, improving docs, making things more stable, etc. If I want to use the latest fancy BS, it would stand to reason that I'd go for commercial proprietary SW, because those vendors are all in favor of making "new" shiny stuff to get me to spend more money to do the same thing. The same shouldn't also be true of FOSS, but for some reason it seems to be.
>I also don't mind this because I like learning this stuff too
I don't. I consider it all an unnecessary waste, and I really don't like these new UIs at all. Thankfully I'm still stuck on Win7 at work for the moment, and minimize my use of Windows whenever possible as I do most of my work in a Linux VM (and don't use it at home at all), and I never even see any Macs up close. I put up with Android because it's the best of 2 bad choices, but also IMO it's not that bad: it's a touch UI designed for a handheld touchscreen device and it's OK at that, unlike PC stuff that they seem to be trying to tablet-ify.
>How do you keep power users and people on the cutting edge happy without losing the technophobic grandparents and those slow to learn?
Whatever it is, the answer isn't trying to tablet-ify PCs, or using garish and nauseating color schemes. Everyone was doing just fine with the WinXp-7 style UI for the most part, and why do we care about grandparents? Today's grandparents were using Win95 and XP, so they're perfectly used to that kind of UI. Grandparents today are not starting to use computers for the first time; that was 10-20 years ago. Anyone left now who isn't familiar with PCs to some extent isn't going to be alive for very long, and isn't interested in learning PCs now. And most older people with simple needs would be perfectly happy with an iPad or other tablet anyway, which is why the PC market is no longer growing.
> Nautilus is Gnome software, not KDE. Why would KDE try to open Nautilus for anything
It wasn't KDE software opening nautilus. It was random 3rd party software that decided to launch nautilus instead of asking what it should have launched.
> Everyone was doing just fine with the WinXp-7
Sounds like someone who never worked on a help desk. There is someone new using a computer for the first time everyday. I would be confident this was true even if it were just grandparents, but there are poor subsistence farmers settling into cities and luddites turned everyday. I think you are letting your tiny view (we all have a tiny view) of the world make you think that every hold a decent professional job in a decent western and industrial nation.
And no matter how much you long for it the past is not coming back and any UI that requires clicking "start" to shutdown is in wastebin and rightfully so.
>any UI that requires clicking "start" to shutdown is in wastebin and rightfully so.
Citation needed. Windows 10 still requires this, they just don't label the button "start" any more (nor does any other UI I've seen in the last 10 years). Mechanically, it's basically the same.
The win 10 is an all around mess and the typical user gets ads in their start menu.
It still has a few dialog from the nt4 era, some that use the snap-in nonsense, regedit is still a thing. Then for about half of what a typical user does they can use the nice pretty UI, as long as there are no unexpected errors.
> somehow we're supposed to accept this as normal for DEs?
No, but it is the normal result of an open source development process. Being able to fork at any time is exactly why the fragmentation exists. Don't like it? Use a commercial product.
I think it is even more extreme. It is the normal result of having multiple teams making UIs. Each team is going to try things that are different.
Consider how different the iPhone and windows 10 are. Or windows 7 and windows 8. Because the community driven teams don't stop when they run out of money they stick around, but they also have really small marketshare.
The problem is too many people with too much time on their hands. For the commercial teams, they're in-place and receiving a paycheck, so their managers want to make themselves and their teams look like they justify their salaries, so they come up with New!! stuff to do. For the FOSS people, it's partly the same: the Gnome3 devs are largely employed by Red Hat. For others serving in a volunteer, it's a lack of discipline and too much time on their hands which could be used for more productive stuff (like documentation, fixing bugs, etc.).
Writing code in leisure time has given us some really great FOSS applications. But like many things, it's a double-edged sword: it's also given us Gnome3 and a lot of today's unnecessary churn. However, unlike little 1-man projects, Gnome3 isn't just a leisure-time project, it's largely led by paid developers at Red Hat. So honestly, I'd say that the volunteers really aren't the problem, and if you eliminated the salaries of the Gnome devs working for RH, maybe we wouldn't have some of these problems; the volunteers aren't going to sustain all that effort on their own. The commonality I see with UI churn across the industry (proprietary and FOSS) is that UI people are being paid to do UI work. That means they have to justify their pay and their existence by doing more such work, even when it isn't needed and it's downright bad.
I disagree so strongly. I hate os x and windows on the desktop. even just the window managers are awful. the lack of a package manager is super dumb. there's a fundamental lack of power.
I used Linux for over a decade exclusively while good friends told me how great OS x was now. I finally had to support building an iOS app so I bought a MacBook.
Great hardware. the os is awful compared to Linux. the window manager is worse.
they have some really nice apis available. increasingly to use those you must sell your software in their store and give them 33% of your revenue.
I think a world where one or two companies get 33% of all software revenues is roughly maximally dystopian.
I absolutely agree that Windows needs a better package manager. But it does have one: PowerShell comes with one, and the Windows Store is finally getting serious attention from major players now that MS has made it easier to port Win32 apps to it.
The huge advantage of the Mac and Windows app stores over Linux is that installing a Linux app allows an app to vomit files all over your PC, while MS followed the Mac model with packages that install into a single location. Much cleaner.
That's funny, I find the linux package management far cleaner than windows. There's nothing that prevents an MSI or EXE installer from dropping files anywhere on your PC. Sure it may ask whether to install to a directory but it can install anywhere. In particular rooting out a random bunch of numbers from your registry is no simple task when you decide to uninstall an app and the uninstall author chose to leave behind remnants.
On the other hand unless a linux package has a post-install script that moves things around, I can just dump where everything landed by a simple package query. To each his own, I suppose.
I totally agree that the install model that most apps use right now is mostly horrible, despite MS's efforts to clean it up for the past few decades. But registry cleanup is pretty rare, in my experience. I haven't had to do anything in the registry since Windows 98, probably, although I have chosen to.
However I was comparing the newer package manager and Windows App Store install models to Linux's package management, since that is the most comparable model. Both installation models are closer to the Mac model, where an app package gets installed to a single location. Universal apps are containerized, and don't use the Registry at all[0]. The PS package manager installs packages into a single location, where you can pretty easily remove them using the package manager, or manually, if you want.
Windows and OS X have both gotten progressively worse over the last 10 years. Everything past Windows 7 is terrible, and the iOS-ification of the mac OS X platform has killed it for me.
I much preferred the UI/UX of the old Classic Mac OS. Apple should have just plopped that on top of Darwin (I would probably still be a mac user).
Fedora is good enough, and definitely breaks less often than a Windows install (which is a big change from even a few years ago).
>Fedora is good enough, and definitely breaks less often than a Windows install (which is a big change from even a few years ago).
My experience is completely different. Not with Fedora, but with Linux in general. I've never actually managed to hose my Windows system. Closest I've gotten is when installing custom themes on Windows 7 and done fucked my explorer.exe, so my GUI was gone, but I was able to fix it with cmd. On other hand I've mauled multiple Linux installations to a point where screens are just black or login screen won't even show or system doesn't even recognise MBR on the disk.
OSX so far I haven't been able to break, but it might be just because I haven't actually tried to tweak it in many ways. Currently I'm kind a tied in the Apple ecosystem, so I just run OSX on my laptop/hobby machine and Windows on my "gaming rig", but I use Linux daily for work.
Looking at the most recent Steam Hardware & Software Survey, Linux share is at 0.72% and shrinking (a year ago it was 0.81% two years ago it was 0.89%).
I already pointed out multiple times, that Steam survey is useless for evaluating Linux gaming market. Why would Steam even need a survey? They have full stats available, but they aren't public.
And what's important is sales potential, which you can evaluate by sales percentages per OS for cross platform games. Steam doesn't publish such numbers. Humble Bundle does though.
I already pointed out multiple times, that Steam survey is useless for evaluating Linux gaming market. Why would Steam even need a survey? They have full stats available, but they aren't public.
I wouldn't be surprised if a survey actually overestimated the number of Linux users. Using Linux for gaming is a choice and usually a more difficult path than gaming on Windows. So, Linux users are likely move vocal about their use of Linux, for various reasons: (1) it's often a principled choice; (2) by showing that a significant chunk uses Linux, more gaming studios will target Linux.
tl;dr: it would be nice if Valve released actual stats, but I think the survey is a good ballpark figure. In other words, the market is pretty much non-existent.
> usually a more difficult path than gaming on Windows
Just to speak against this notion (which absolutely used to be true, no contest there): anecdotally, I don't find that to be true anymore, at least using Steam.
It works really smoothly for me, just as on Windows. In fact I'm just about ready to drop Windows as a gaming plaform. Maybe some titles aren't available for Linux, but I still have enough to fall back on.
Perhaps by "difficult" here OP means "access to games" and not just setup.
If you choose to game on Linux, you are intentionally restricting the number of games you can play. And for some people that's obviously ok, but for most (as stats show), it's not ok.
I wouldn't worry about total number of Steam users or their percentages. It's pretty irrelevant. What's more important, is potential of sales per OS. So for all intents and purposes, that survey is pretty useless. But numbers from HB are actually something you can evaluate.
> by showing that a significant chunk uses Linux, more gaming studios will target Linux.
You're being pretty confrontational here, and I don't think it's necessary..
No matter which way you slice it, the Linux gaming community is tiny when compared to Windows and consoles. Sure, it's growing, and there are more and more games coming to the OS, and more choices = better of course.
But we are a long way from Linux being a platform that matters in the overall computer gaming world.
I'm not confrontational. I just prefer when those Steam numbers aren't used like some absolute indicator of some situation, and they often are. They aren't demonstrating sales potential, so developers shouldn't use them for evaluating viability of Linux releases.
Making statements like "if you paid attention" or calling people "desperate" is confrontational language, or at best unnecessarily harsh.
Linux gaming is obviously a subject that you hold dear - you've jumped in multiple times on almost every single top thread in this discussion - and that's ok, but ease off a bit..
Let your points stand on their own merit, you don't need to call people out because they're not as deeply informed on the subject matter as you are, or because they don't agree with your perspective.
Statistics are often twisted around or taken out of context to make an argument, which is wrong. So don't get upset if that is dismissed to make conversation fact based, not speculation based. And this particular stats example is waved around often for that incorrect argumentation.
Because they need permission from the user to collect the data.
Even if the data is undercounting Linux (maybe Linux games are less likely to take the survey) the trend is consistently down for ~4 years now. So you might make a case that the size is bigger than the survey shows, but not that it is growing.
Trend down doesn't mean the number of Linux gamers is decreasing. Even if you assume the survey is correct and representative, it can simply mean the number of Windows gamers on Steam is increasing faster. Besides, Linux gamers are less likely to use Steam than Windows ones (DRM-free preference).
From DRM-free stores, GOG unfortunately don't publish any stats, and stats from HB show pretty good Linux percentage.
I agree, it only means the share is shrinking, not the absolute numbers. Given the large overall Steam growth rate, it is certain the absolute number has increased. Nonetheless, it isn't a great story for Linux gaming.
I don't see how it's any story really. Steam grew. So? What's more relevant, is how much Linux gaming has grown. And it has quite a bit in the last several years.
When you purchase from Humble Bundle, they don't know what platform you would use to play. All they know is what platform you're buying from.
People who use their Android phones to purchase and pay Humble Bundle, get classified under "Linux". You can see big Linux platform numbers even for bundles that don't have a single Linux game.
Steam's survey reads the platform - the user has almost no role in actually filling it out beyond saying 'yes, you can collect this data'. This makes it significantly better than requiring each and every user to act to specify their platform.
The survey doesn't show up when you are in big picture mode. Since Steam machines boot into big picture mode, they will never get the survey unless they switch over to desktop mode, which takes some effort. That would account for some of the Linux under counting.
It just shows that Valve themselves don't take that survey seriously, if they don't even count specialized hardware targeted for SteamOS. So I wouldn't use those numbers for any far reaching conclusions.
Nobody uses SteamOS or Linux for gaming. It baffles me that people like you go to such lengths to push Linux for Desktop...
I tried using Linux on several occassions over the last 10-15 years. It was never easy or fun. All the GUIs felt like they were from the 90s and nothing worked out of the box. I tried Ubunutu, Mint, Arch, and a few others. My last attempt at it was ~2014. Maybe I'll hate myself enough to try Linux again in a year or two...
It baffles me when people use the "I couldn't get it to work so it obviously doesn't work for anyone" argument, even in the face of statistics that prove otherwise. Ho hum.
I suppose you should use it today, instead of making your impression on the times long ago, and claiming Linux isn't usable because of some long gone issues.
In total number yes, but in percentage of the total market, there's hardly any change, Windows and Mac gamers are also growing segments too.
And the GP comment was right anyway, 90% of the Market is for sure not running Linux for gaming on PC. 1~2% is nice but it's hardly enough to convince large publishers to support Linux.
They don't need Mesa for that, their "worsktation" customers are using their closed driver. But if you missed it, unlike Nvidia, AMD have a team which contributes to Mesa directly.
if games ran on linux, i wouldn't use windows for anything. i love everything else about linux for the most part. it's a chicken and egg problem. if (AAA) games were actually released for linux, linux market share among gamers would grow hugely. i don't use linux for games because no games i want to play run on linux.
It's a chicken and egg problem. I want to play some AAA games as they are released and not have issues with Wine, which is why I'm still on partly Windows at home. It's mainly Blizzard and Bethesda for me but EA still neglects Linux as well.
You can't deny that if the big AAA players supported Linux it would bring a (large?) influx of gamers. If they'd all just target the latest Ubuntu LTS or whatever SteamOS run on, the community would make it run on their own distros.
Legacy publishers aren't innovators. They are too lazy to think out of the box. For me it's forget Bethesda and Blizzard. There are more than enough developers who are worth supporting on Linux, and who are making good games.
But if you are really so desperate to spend money on those who ignore us, Wine is improving all the time, and now already approaches decent DX11 support, so even games like Overwatch are now playable in it.
So have I, but it's actually gotten there for me, my main/gaming PC has been linux for 6 months now and I've been pretty happy. Most of my gaming is strategy games from paradox, and they've all supported linux for a while now. Apart from that there are more linux games than I have time to play, the only change is that I no longer buy a bunch of games during a sale that I will never play.
Define very. More game studios release games for Linux today, than several years ago. So not very few for them anymore. There is no "this year". There is clear growth. It's a process, and it's going in the right direction.
Obviously not. But neither would most want to do HPC on a computer they are also running a graphical desktop on either. In my experience, most serious academic computation happens on headless servers that are certainly not running Weyland or anything similar.
The pipelines of many animation and entertainment studios use Linux. If Nvidia just wants to sell to gamers, that's fine, but it seems like the work to have Linux drivers would be worth it just to keep selling hardware to studios.
The point was more about that they are directly profiting from Linux in various forms, yet refuse to be part of the community and provide proper support for their hardware, unlike say Intel.
The reality is all the money they make off Linux is with workstations cards or HPC. Neither of those market niches are likely demanding Wayland support since stability is a hard requirement and Wayland really isn't sufficiently mature yet (nor does it really have sufficient market adoption).
They're a business, so when their customers need it, they'll support it. Until then, only hobbyists or an insufficient part of their customer base is interested, so understandably, they're not investing in it.
Your forgetting the embedded device market that the Tegra K1 and newer are used in. Is Nvidia going to orphan this whole market segment they've built? I know they replaced a ton of WinXP boxes in labs across the world, and it seems as though they have captured most of the automakers on that same platform, from Toyota to BMW all running Xorg in their cars.
Will Toyota go for running legacy, unmaintained Xorg in their cars? How about BMW? I don't see Nvidia ready to dump this market they've spent a decade building, so short of them being complete idiots, they will probably walk back on this.
They're talking about not supporting XWayland, not Wayland proper.
Given this -- if Toyota doesn't want to run "legacy, unmaintained" X11 in their cars, then.... they don't have to, they can just move to Wayland, which will work, and NVidia supports Wayland just fine -- and they don't have to worry about X11 at all anymore?
So, everything would be fine? I'm not sure what point you're making here...
Those markets are also not demanding support because many of them have no display at all. This is definitely true for HPC GPUs and will be true for deep learning servers.
Wayland is probably going to be the default in all the next stable releases (RHEL, debian (is it already the case?), Ubuntu). Not supporting it sounds risky.
Risky, but for whom? If my GPU stops working with the Linux distro I use, I'm definitely more likely to change my distro. GPUs are expensive and there isn't much choice between vendors. Linux distributions are free and there's plenty of choice between vendors AND between versions.
Selling a GPU takes time and loses you money - and that's in PC case, because on a laptop you don't even have that option.
Moreover, if you have an NVIDIA GPU, you most likely need a decent GPU for something - high-end video games, 3D modelling work, GPU-accelerated computations, etc. Linux distros are, frankly, not much different for one another - they're pretty interchangeable for almost all tasks I can think of as a user/developer. GPUs, on the other hand, are a bit less interchangeable between vendors. You really have only two players to choose from (Intel GPUs don't count for anything serious) - so if both suddenly become incompatibile with a given Linux distro, then it's obviously bye bye to that Linux distro.
Hold on. They didn't say they won't support Wayland - they will. They said they won't support Xwayland which is needed to run X programs under Wayland. I don't think that's entirely unreasonable. The community has said Wayland is the future, and they are finally on board with that. But now the community is dragging their feet in actually getting there. We still don't have upstream Firefox running natively on Wayland. Gimp, Inkscape, others, they have it on the radar but have been putting it off for years now. The only thing nVidia is doing here is not allowing the free software world to lag behind in their own goals ;-)
That sounds like some bullshit to me, the amount of effort for Nvidia to ease pains over the transition period -- which will surely last at least ten years before even 80% of applications are patched for native Wayland support and will and involve dumb shit akin to the systemd debacle like "We hate Wayland so we're forking Debian and calling it DebianFuckWayland" -- is marginal compared to the benefits. Xwayland will be here and necessary for a very long time.
Didn't mean to do that. The whole thing has been a point of contention for a long time, so I wanted to clarify that they are actually getting on board with Wayland. They are by no means fully backing with Xwayland. I want a full Wayland experience and nVidia is doing the bare minimum to enable it on their hardware. For me getting all the way there really does involve desktop applications embracing it and adding support too. X is deprecated, and so should be Xwayland.
Not supporting deprecated software for a deprecation period is just rude, especially when said software is integral to your product on a certain platform and present on 99.69% of Linux workstations. Nvidia is not some kiddy startup and shouldn't play the "move fast and break things" meme.
What % of nVidia's sales go to Linux systems running Wayland on in a scenario where X is desirable or even necessary? Far and away the vast majority of their GPUs in Linux systems are either in supercomputers or embedded applications like cars.
Yes maybe... and also the percentage Linux systems with nVidia GPUs running not in a SoC Application (e.g. Car Infotainment) or in a Supercomputer is probably less than a percentage point.
Right. And if you have a server running Linux and want to run a basic Wayland environment on it you can now do that with nVidia hardware. For car infotainment I thought the genivi guys were going Wayland already.
>By not treating Linux end user devices as first class citizens, I don't see how Nvidia intends to have a viable business long term. Who wants a tablet, phone or desktop GPU without drivers?
The more relevant question is "who wants a desktop Linux system". And the answer is very few.
That's why they doesn't care about that market. The huge majority of desktop users will go either Windows or OS X.
That was the case when Mac Pro would still be a viable option for a workstation. Now it isn't, and Windows, despite improvements, isn't there yet. So many researchers and HPC users, including myself, are stuck with Linux desktops. I am not happy about that, as my Mac Pro used to be much better. But there's nothing I can do about it now.
(And, of course, Nvidia is the only choice for now if you want GPGPU computing).
Apple has released some rumblings that it's going to bring back a modular Mac Pro and keep it on x86 even if it goes to ARM for the Macbook, but I wouldn't hold my breath.
And as earlier stated, most low-end to midrange GPUs are going to be Intel only, Apple has completely abandoned NVIDIA. On high end, there's still AMD to compete with. Their ARM chips seem to have failed, nearly every chip is Samsung, Qualcomm or Apple. Is NVIDIA really in a position where they can just blatantly ignore communities like Linux?
>> By not treating Linux end user devices as first class citizens, I don't see how Nvidia intends to have a viable business long term. Who wants a tablet, phone or desktop GPU without drivers?
Desktop Linux with Wayland has maybe 0.1% market share rising to maybe 1% over the next years in the best case. In the worst case, Wayland will just be another Linux Desktop fail. Other devices (Phone/Tablets) do not use Wayland.
The amount of resources any given company can allocate to Linux platform bullshit is limited. The "serious" NVIDIA Linux customers will probably stick with X anyway, in the foreseeable future.
To a very good degree of approximation, nobody uses Sailfish OS.
I'm not ruling out that there exist other pointless mobile operating systems with 0.0% market share that use Wayland or even XWayland in some capacity.
My point however is that the "market" argument is nonsense when the market is practically non-existent.
this isn't "no plans to support wayland" - these drivers are built to support wayland. It's the Xwayland translation layer they're talking about here, which (I think - my understanding is a bit thin here) is for running applications that are compiled to use X11 graphics in a wayland environment.
So: it could well be that this is an issue of resource allocation more than anything else. Over time, most applications should move to using wayland (or whatever window / surface library emerges as the clear winner, but that looks like wayland is the future here) for their graphics. A lot of effort into supporting Xwayland would mean work spent on a library that should be of limited utility - only useful during the transition between platforms (like the old carbon layer for OSX in the early 10.0, 10.1 days). Right? Or maybe I'm misreading it.
From what I read on reddit, Xwayland is X running within wayland, and (for some reason) needs some special code in the drivers. I'm not fully clear as to why or what needs to change.
The motivation for Xwayland is that 1. a lot of applications use no GUI frameworks (or outdated versions thereof, e.g. all these apps that are still on Qt 4), so if the frameworks are ported to Wayland, it doesn't help them. And 2. a lot of applications talk to X directly even if using a framework (e.g. an application that takes a screenshot needs to use xlib or xcb directly because that's typically not abstracted by a GUI framework).
Xwayland enables these types of applications to continue working in a Wayland session, so that desktops and distributions can move to Wayland without having to either wait on the legacy applications or remove them from the distribution.
AFAICT XWayland is a server that speaks the X protocol and not an Xlib replacement compatibility layer that emulates X11 client+server in-process and speaks Wayland protocol (though, I imagine various mismatches prevents that sort of thing from working well at all).
I've been following the X11 vs Wayland story since very close to the beginning and have built and used wayland and weston multiple times over the years. Cutting through all the FUD on both sides, my conclusion is that Wayland was created in the context of some very specific use cases which essentially preclude it from being a general solution that can replace X by itself (it could succeed if there was a single project to reimplement X in a sane way, but leaving it to GKT and QT is NOT a solution by any stretch of the imagination). Nvidia's engagement with linux simply does not cover the use cases that Wayland was created to address. In my view desktop/workstation use cases are not sufficiently covered by Wayland, and that is where nvidia's focus has been.
tl;dr Nvidia has no need to play in this space, anyone using X will just use X without the translation layer.
While your post is interesting, it is also vague. Can you specify what Wayland is missing that can't be adequately handled by QT or GTK, or do you just have the feeling there is such a thing?
Without speaking too much FUD from the X11 side the main issues facing workstation use cases are input handling and permissions. The current wayland party line is "just use dbus to communicate between windows." This is complete madness because it basically means that if you want programs to be able to work together and not steal each others keybinds they all have to depend on a message passing framework. Full disclosure I have done everything in my power to avoid installing dbus and having dbus dependencies, but it is getting harder and harder. That said, it doesn't matter what you think about dbus, it is completely unrealistic to expect that every piece of gui software written for linux to be rewritten to include dbus. This is basically going back to the days of cooperative multitasking and I think we all know how that ended, the overhead is just too high.
What exactly do you mean by "stealing each other keybindings"?
From what you have described so far, it would seem to me that you are describing problem specific to X11 where global keybindings are often implemented in spyware manner, i.e., listening to all keyboard events (regardless of current window focus) and reacting accordingly.
They may conflict with each other, they may interfere with each other, they may spy on user, in other words they are not working together at all.
There is no standard way to register global keybindings under Wayland yet, but this is bound to happen eventually.
Not as a builtin part of Wayland, but it doesn't matter.
It will of course use D-Bus interface, as KDE does it now for example - but if an application doesn't support this interface, it can't steal keybindings, it just doesn't get to install global keybindings itself.
Not sure what you have exactly against D-Bus, but clearly modern Linux desktop embraced it.
I have something against dbus. Bout the protocol and the implementation are very much over-engineered. The protocol at least has an excuse.
In the desktop things, that wayland aims to do, dbus is useless. (actually i don't know what it's useful for at all, other then quickly hacking some object oriented (gui)client-server desktop program)
I do recommend reading the dbus protocol (the protocol, not the API).
Sorry to sound negative but "modern linux" now means "desktop crap", and it is all crap. Freedesktop.org (the linux "desktop" "standards" authority) has gone from "ehh, passable" to "complete crap" in the last... idk 5-10 years (the "desktop" quoting is because it has gone pass the actual desktop problems).
I only follow Linux graphics casually, but AIUI one of the principles of the Wayland initiative was to replace the monolithic model of X with a combination of more tightly scoped components. For example, libinput was created to take responsibility for actually handling input devices, so that Wayland compositors would never need to carry that code, and the input device handling in X could be decommissioned.
From that point of view, "Wayland" is kind of a short-hand for a number of different stacks that have some components in common, and speak the Wayland protocol. If you choose not to use a complete stack on a system, then the result can't be fully functional.
Personally, I don't have any sense of how much dark matter is in the area of Linux GUI applications. We can easily know about Open Source, where most modern applications sit on toolkits and desktops that are moving to Wayland, but presumably places like CERN have some really old internal applications that will only ever run on X until they are replaced. I'm genuinely unsure how large that problem is.
I have to say that I'm always skeptical about arguments that something can't be done for performance reasons. People have been using that argument against new technology since at least the days when C was decried as new-fangled, decadent luxury.
Can you provide an actual example of what is impossible without dbus? Getting mouse/keyboard input for your window is part of libwayland, and not dependent on dbus at all.
> (it could succeed if there was a single project to reimplement X in a sane way, but leaving it to GKT and QT is NOT a solution by any stretch of the imagination).
Qt and GTK are not responsible for implementing the compositor at all. Compositors aren't actually the problem at all, the problem is applications.
XWayland is one of the main ways we are going to bring legacy applications to Wayland desktops. It works about as well as X does, and lets the other applications take advantage of the sane architecture of Wayland.
> In my view desktop/workstation use cases are not sufficiently covered by Wayland, and that is where nvidia's focus has been.
What use cases? The only thing that's missing for me is input method integration. It has full display colour management, handles multiple refresh rates in the same session well, handles scaling, supports every input device I own (including graphics tablets). What's not really firm right now is native support in every application, that's coming along at some rate, and XWayland is bridging that gap quite well.
Qt and GTK and responsible for implementing how all of their windows talk to each other. My point is that it is not reasonable to expect or require programs to depend on a widget toolkit just so they can get proper keyboard input. I realize this is a bit of an exaggeration, but X11 is more than something that draws windows and from my understanding wayland doesn't address many of the other parts. For example, from my understanding wayland does not provide a standard way to support window manager level keybinds that override the binds set by the developer of the window that currently has focus. For me this means that the primary way I interact with my computer is now broken on wayland unless every single program I use implements the ability to pass those custom keybinds to the program that actually handles them. Sure, everyone can implement their own way of dealing with window manager level keybinds but that completely defeats the purpose of having a standard in the first place.
Probably the most salient issue with wayland is that it prevents programs from communicating with each other in the name of 'security' and 'sandboxing' when preventing those things at the display server level adds literally zero security and just makes everything more painful for application developers (imo to the point where they simply won't touch the stuff). If there were a 'windowing protocol' that went along with wayland that covered many of the things wayland does not then I do not think there would be as much of an issue, but the thought that we can force a new display server on people without the other bits as well seems at best naive. Again, I think that the absence of such a 'windowing protocol' from wayland is because one of the main use cases it was developed for often had literally one program running (car backup camera displays).
I do not dispute that xwayland bridges the gap. But if people are using X right now then nvidia has no reason to support xwayland right now because there are no wayland only applications that their customers are demanding support for. Down the line that may change, but from a corporate decision making point of view there is zero reason to do this.
> Qt and GTK and responsible for implementing how all of their windows talk to each other. My point is that it is not reasonable to expect or require programs to depend on a widget toolkit just so they can get proper keyboard input.
You don't need to. Just like X11 has Xlib and libxcb, Wayland has libwayland. It is actually easier to get input from libwayland than it is to get it from Xlib, and about on par with libxcb.
> I realize this is a bit of an exaggeration, but X11 is more than something that draws windows and from my understanding wayland doesn't address many of the other parts. For example, from my understanding wayland does not provide a standard way to support window manager level keybinds that override the binds set by the developer of the window that currently has focus.
You're right about that, global keybindings will have to be allowed through the compositor, and I don't know if there's a protocol for that yet. Global keylogging is one of the things Wayland was designed to avoid, so the way we implement that feature will have to be a bit different. It's for good reasons though.
> For me this means that the primary way I interact with my computer is now broken on wayland unless every single program I use implements the ability to pass those custom keybinds to the program that actually handles them. Sure, everyone can implement their own way of dealing with window manager level keybinds but that completely defeats the purpose of having a standard in the first place.
I don't know that it defeats the purpose of a standard! I agree that these use cases need to be addressed, but I don't think the way to do that is to give every Wayland client the ability to globally change the way input reaches other clients without permission from the compositor.
> Probably the most salient issue with wayland is that it prevents programs from communicating with each other in the name of 'security' and 'sandboxing' when preventing those things at the display server level adds literally zero security and just makes everything more painful for application developers (imo to the point where they simply won't touch the stuff).
Yeah, I think that the XDG compositor extensions should loosen this stuff up a bit, or at least make it easy for an application to ask the user for permission to spoof everything. I'm unlikely to be displaying a hostile application on my machine. That said, doesn't exactly add "zero security" though. It is helpful for partially-isolated applications, like those in containers. I would like to be able to trust the isolation of applications some day so that I can run spotify on my workstation without breaking into a sweat.
I think it would be perfectly adequate to have the application ask for permission to do the heinous boundary violations that make X11 fun and dynamic.
> I do not dispute that xwayland bridges the gap. But if people are using X right now then nvidia has no reason to support xwayland right now because there are no wayland only applications that their customers are demanding support for. Down the line that may change, but from a corporate decision making point of view there is zero reason to do this.
For what it's worth, NVIDIA is looking to support Wayland on desktop. They already support it commercially in their embedded products (especially targeted toward automotive). So it's not exactly accurate to say that they have no customers asking for Wayland-only application support. The question is, when will it be convenient on desktop Linux? and that's hard to say. It is somewhat surprising that they aren't currently planning to support acceleration in XWayland, given that they have at least put some effort into supporting Wayland in general.
Ouch. Is Xwayland still the only way to do remoting? https://wayland.freedesktop.org/faq.html#heading_toc_j_8 just punts on it and speculates about what could theoretically happen someday, but I don't even know how or why I'd start using a backend that can only run on one computer.
Why is support from nvidia even needed? I remember that there was always this weird thing where graphics drivers were written for X11 instead of being part of the kernel (or a module) like all other drivers, is this still the case? Why can't they write a kernel (framebuffer?) driver so no specific support is needed for a desktop environment?
Indeed they can, For basic framebuffer, but for doing 3d acceleration, to get many nice desktop effects people are used to you basically have to speak their proprietary protocol to the GPU. You can do a little better than framebuffer, like the "nv" Xorg driver, but it's still not really sufficient for composting, video playback and other things commonly expected from a desktop OS.
There's an effort to reverse engineer this called nouveau, they've done a surprisingly good job in a surprisingly short time, but it's still far from perfect.
> There's an effort to reverse engineer this called nouveau, they've done a surprisingly good job in a surprisingly short time, but it's still far from perfect.
Nvidia has been putting a lot of hurdles on their efforts since the Pascal GPUs. If you look and pre-Pascal GPU the performance of Nouveau is really impressive, but Pascal GPUs are crawling to go beyond a few frames per second. It's really sad.
Basically, if you're not a gamer, the open-source Nouveau driver should provide all you need, assuming it works for your card. But parts of your system may stop working, or cause other bugs, from not using the official drivers.
edit
I might have misinterpreted your question, so here's an alternate answer. If you mean "Why can't we just do 3D rendering with a framebuffer?", a framebuffer is just a copy of the rendered screen in memory. You render it in your local CPU, and then copy it over into video card memory. So it doesn't use hardware rendering, and can work with pretty much any card.
If you do want to do hardware rendering (or most any feature of your card at all) you have to write a kernel driver (the DRM) to use the chip, and then some userland software (the DRI) to manipulate that driver, and send data to it somehow from your client app using a standard API for 3d graphics (OpenGL's GLX extension for the X11 protocol). The devil is in the details though, so to work around some limitations, some people wrote new graphics software to allow the client app to render without a "middle man" in the way. That's where you get Wayland.
FWIW, I use an nVidia GTX960 with the nouveau driver on openSUSE Tumbleweed, with two displays. Basically, it Just Worked(tm) out of the box.
Performance is good for regular desktop use, including video playback (then again, my CPU is a Ryzen 1700), 3d gaming performance is not overwhelming, though. Plus, Kerbal Space Program gives me weird display errors and sometimes manages to crash Xorg.
I am not much of a gamer, though, so I can live with this arrangement.
That's just too bad, especially for those who use Wine (Wine on Wayland now looks completely stalled, and XWayland would be the only way to use it in the foreseeable future). Firefox is also far from ready.
But Mesa really caught up lately, so switching to AMD was never as good. Just forget Nvidia and enjoy properly integrated graphics stack.
Note that this is not equivalent to saying "we plan to not/we will not support Xwayland." It just means it's not currently in the pipeline (and may or may not be in the future).
I mean, that depends, doesn't it? nVidia also refused to support KMS for a long time but they finally added it as it became a harder and harder requirement. KMS was a big overhaul and pre-requisite for modern stacks like Wayland. If that can get added, I'm reasonably confident that Wayland support will show up somewhere along the way (assuming that Wayland continues on its track to widespread adoption).
No, it doesn't depend. The two statements are equivalent. NVidia could certainly change their plans, but the particular way they phrase "no support" is irrelevant.
It's not as if "we have no plans for support" could result in accidentally having support.
Wayland will be supported by nvidia. XWayland will not be. XWayland is a transition layer than runs X11 applications under Wayland. Generally speaking, most users will not need to do this.
"Speaking of the Linuxes, Ubuntu is tops among them with 12.3% of the entire OS market for developers. Fedora, Mint, and Debian accounted for 1.4%, 1.7%, and 1.9% of all responses, respectively."
> Their current X11 drivers tear like crazy, and I have hope that Wayland will fix that.
No they don't. I have had 3 different nVIdia configs, and presently 2 for gaming, and through my experience hopping from one distro to the next I find that the problem is actually unrelated to nVIdia drivers - but the compositor. For example, in Mint if you switch to compton, it makes all tearing disappear. This is well documented.
I tried to do that, unfortunately for the high end AMD is simply not competitive. If you want to game in 4K NVidia is the way to go (or at least it was 6 months ago when I bought my card).
Yeah well I doubt they keep that stance for very long. Nvidia is investing like crazy in keeping the deep learning community happy about their products and they are disproportionately on Linux.
If/when Wayland becomes the default for Ubuntu, Nvidia will follow. Can't let them ponder the amount of time it would take to code the missing layers for AMD hardware
As a Fedora user I can understand the reluctance. The recent upgrade to Fedora 26 has given me a ton of gnome crashes which I think are Wayland related. Commentary on Fedora channels tends to be along the lines of "Wayland is great and works for everyone (OK, maybe not those with NVidia cards)".
If Ubuntu has millions of users it may have millions of NVidia users. Putting them onto unstable technology to pressure Nvidia into supporting Wayland wouldn't be a great move.
Ubuntu is the default distro for gaming, too, and I doubt they will jump until it's well ready, and certainly NOT before nVIdia provides support. It would make no sense.
Wayland is the default for Gnome now, and Ubuntu is switching to Gnome for its desktop next year. Have they said anything about whether they plan to change the default from Wayland to X for their installations?
Not exactly on the deep learning side, but I am a scientist who does some computatinoal work on the GPU.
You're right that we don't use our computer card for our graphics and, as many people pointed out, the computation will wind up running on a headless server. On the other hand, during development, our code will be running on our desktops and will be using the GPUs on the graphics card. I've witnessed a developer's frustration with X11 driver issues cause a project to transition from CUDA to OpenCL.
Granted, he didn't just throw a hissy fit an port the code. Rather, it inspired him to write up a tradeoff analysis of OpenCL versus CUDA that showed that OpenCL offered this specific project significant benefits and that the main selling points of CUDA weren't being used in this instance. He also wound up rage coding a prototype port to OpenCL to go along with the proposal.
The project would have moved eventually, since it was the better engineering decision in this case, but it was about a half decade away on the timeline until the lead dev's compositor stopped work.
I can only speak from my own experience (2-3 GPU rigs running tensorflow) but I have never had any trouble. Full Ubuntu desktop. Training models on the GPU does not cause any problems with display lockups etc.
No not really. The finger was about not giving enough details for opensource driver and pushing closed source blob instead. This issue is about closed source blob not supporting next gen linux desktop until the desktop has gotten enough market share to force them to.
What's the best supported graphics card line on Linux now and in the near future, anyway? I recently tried to search and it was difficult to get good information.
Any of AMD's recent discrete graphics cards are very well supported. Both the proprietary blob driver and the Mesa-based free driver share an upstream kernel module. The Mesa-based driver tends to have better performance for most applications, but currently lacks some of the bleeding-edge display technologies from the proprietary driver. That is likely to be completely rectified in the next couple of kernel release cycles.
They continue to expand their team, and there is a certain amount of broader community involvement in the Mesa-based driver. Valve software has at least one full time developer working on the driver in addition to the full time developers working directly for AMD on it.
The peak performance of AMD's top (pre-vega) cards right now is less than the peak performance of NVIDIA's top cards, but for similarly priced cards, the performance is basically on par. Finishing out 2017, and starting 2018, I think AMD may reach or exceed NVIDIA in flagship performance.
> Any of AMD's recent discrete graphics cards are very well supported.
This is true and I'm a happy linux gamer these days but I wanted to point out that these are new drivers and the mesa support is really only in new kernels. Ubuntu 16.04 - don't bother, Ubuntu 16.10 - getting there, Ubuntu 17.04 - worked great for me since day 1. Arch will also work fine.
Gamers know to either use rolling distros, or to make sure to install recent kernels and Mesa from backports and etc. instead of being stuck on LTS kernel / drivers.
Even without going out of your way to get newer versions, the kernel and video drivers get updates on Ubuntu LTS releases as a part of the hardware-enablement stack. https://wiki.ubuntu.com/Kernel/LTSEnablementStack
Oh, you're right. I always forget that I'm running Arch and building upstream drivers (also fiddling with development). Easy to forget to tell people that bit. :P
You don't really need to compile upstream drivers for a good AMD experience. I have a Fiji card and have been running (also on Arch) stock drivers and kernels since around Linux 4.5 (when they merged powerplay support for Fiji into the amdgpu driver). Haven't had a single problem; just installed Mesa and xf86-video-amdgpu and everything just works.
I have a Polaris 10 card, and I'm interested in build options which improve performance somewhat. The stable release works fine, but I have peculiar interests. ;- )
What do you need it for? If it's not for gaming or intense 3D requirements, I have been using Intel for a long time, and quite happy (lower cost & reliable).
I do play some games, but it's on a 3 year old desktop which is also running 4 fairly busy VMs (i7-4770 CPU @ 3.40GHz with 32 GB RAM and the built-in Intel graphics card).
As of recently, I have a mixed 1920x1200 and 4K setup on my home desktop, and 1920x1200 + 2560x1440 setup on my work laptop. I was very interested in what I'd heard about Wayland supporting per-monitor scaling and hoping to move over whenever KDE's Wayland support stabilised.
Well, I guess I'll still be doing that on my work laptop with its Intel graphics... here's hoping AMD does better for my next desktop GPU upgrade.
You'd be surprised by how many apps support Wayland natively. Gtk+ and QT support it transparently for app developers, and I assume other frameworks do as well.
And Wayland is a huge improvement over X. Whenever I go back to X, it really suffers from tearing on portrait sub-1080p DIV-D monitor.
I've consistently had issues with Vulkan on the 960M my laptop has. Nvidia sure makes it easy for me to go back to AMD.
The graphics market is dangerously locked up by two players and its time some new players took the capital risk and stepped in. Nvidia would probably give a shit if someone bothered to challenge their share of the market in both graphics and computational space.
I currently have no plans to buy an NVIDIA card ever again for display; and if AMD does well enough, probably not for compute either. They've been consistently useless in supporting the platforms I want to use for work and play, and while they've gotten a shade better over the last decade, not as better as their competition in the discrete GPU market.
Good luck using Wine with Wayland on Nvidia driver.
I'm really glad I ditched Nvidia and switched to AMD. With Mesa progressing at fast pace, Nvidia will become simply inferior in many aspects, including the above one.
Yup. Can't believe that such great drivers were written from nothing but Reverse Engineering. Since mine is entry level laptop GPU the performance has no limits. But Nvidia should follow AMD is releasing Opensource drivers. (But they won't)
You should be happy to hear that while AMD still has a proprietary driver for some outlier workstation cases, the officially recommended "default" driver is the fully open-source stack.
In my experience, dual graphics on Linux is more of a pain than either solely Nvidia or solely integrated. Both my current and past laptop have had dual integrated/Nvidia setups (the first because I didn't know any better, and the latter because I mistakenly thought it was discrete only), and both times my solution has been to just not set up the discrete card at all because I've been unable to get the dual setup working in a reasonable amount of time and I don't do anything graphics-intensive enough for it to matter.
Intel driver support is pretty bad. If you aren't running with i915 at close to head there are easy to hit graphics bugs. Running most Steam games on Linux yields much worse performance and stability than running the same games on the same hardware with Windows.
It's far from being a single variable here. WHen you talk about games, it's not just about the drivers, it's about what graphics API used as well, and how well the engine is optimized to run on OpenGL in the first place. Most engines are made with DirectX in mind so any quick port to OpenGL will lead to worse performance regardless of driver quality.
To be honest, I can put up with the less than graphics. What really got me was the regular hangs that would happen. I spent a fairly long time trying out different things to debug it, and eventually reported the hangs to the DRM list. While they were helpful, their answer was to update my kernel. It worked, but I am not comfortable having to run at head just to play old video games.
Intel is actively trying to kill off lowend and mid-range GPUs in the desktop, and has locked them out in the mobile space, leaving Nvidia with just the high end GPU market and their ventures into the ARM SOC space.
By not treating Linux end user devices as first class citizens, I don't see how Nvidia intends to have a viable business long term. Who wants a tablet, phone or desktop GPU without drivers?