NVIDIA's Proprietary Linux Driver Adds Support For Wayland, Mir (phoronix.com)

An anonymous reader writes: After being desired by NVIDIA Linux users for years, the proprietary GeForce graphics driver natively supports Wayland and Mir as an alternative to an X.Org Server. It's been a long time coming for the proprietary GPU driver stacks to support Wayland/Mir, but with today's 364.12 beta driver there is now the necessary DRM KMS kernel support and EGL extensions for being able to handle these next-generation display solutions. The new NVIDIA Linux driver also provides integrated Vulkan support, PRIME rendering support, and other additions.

Comments

Can't wait to see the performance comparisons

[Anonymous Coward]

Might be about time to move the gaming rig over to SteamOS.

Re:Can't wait to see the performance comparisons

[aliquis]

Might be about time to move the gaming rig over to SteamOS.

So far the performance has been quite a bit worse on SteamOS than on Windows. I don't see why this update would change that. So far Vulkan has performed worse than DirectX too but I assume that will change with new games/engine versions by competent developers.

Re:Can't wait to see the performance comparisons

[Anonymous Coward]

While a popular point of rhetoric, there exists ZERO evidence SteamOS (Linux) is slower than Windows. For example: https://www.youtube.com/watch?v=rISRVeJxhnE Very improperly, most people and articles are attributing poor performance to SteamOS when the real issue is sub-par ports of applications. Also note in the video you can observe a couple of glitches (texture loading) in the Window's demo whereas it's silky smooth on Linux ( - which is believed to be superior Linux file caching).

To be clear, not that I'm ignoring your last sentence, but your first two sentences are misleading; though I don't believe that's intentional. If a game's performance is lackluster, you can completely blame the game developer as it has absolutely nothing to do with SteamOS. SteamOS has all of the foundations to provide FASTER performance than Windows. There exists no evidence to substantiate a claim that SteamOS is slower than Windows and all evidence points to platform performance deltas being that of lazy ports and developers who lack platform knowledge to optimize for Linux as they have done for Windows. Hell, once again we have games running under Wine which are starting to outperform native Windows. Which is pretty profound when you consider they are frequently doing DX to OpenGL translations to boot. Meaning they are faster even with an additional abstraction layer. Bluntly, there is every evidence that SteamOS/Linux (with NVIDIA anyways) is faster than Windows and where applications falter, you can squarely blame the developers.

Additionally, the Vulkan comment is not true. So far, developers who are learning to use a new API, without redesigning to properly leverage Vulkan, and running on beta drivers, have performed worse than DirectX; which is a highly optimized and well understood framework. Furthermore, there are specific guideline recommendations for when Vulkan should be used and when it's possible to provide ideal performance. The current use cases are not even known if these satisfy the recommendations. As such, the examples people commonly use may not even represent ideal candidates for the newer DX12/Vulkan APIs.

In summary, while you have accurately parroted the rhetoric and FUD, it is not supported by any available evidence. That said, I believe the spirit of your comment can be accurately rephrased to say: Many games running on SteamOS have lackluster performance because developers have failed to optimized their games for the new platform (OpenGL vs DX9, DX10, DX11). As such, many games will run slower on SteamOS than Windows, through no fault of SteamOS. This is a subtle yet profoundly important distinction. And it's a distinction which is directly contrary to the common FUD.

Re:Can't wait to see the performance comparisons

[Anonymous Coward]

Arma 3 Bechmark results:
https://i.imgur.com/UfpUI5p.png
https://i.imgur.com/ypWsvqz.png
https://i.imgur.com/haTiRLZ.png

Worth noting, the port was achieved using a DX wrapper which converts DX calls into OpenGL. With an extra abstraction layer, it's still performing on par or better than Windows.

Re:Can't wait to see the performance comparisons

[Hylandr]

Mod UP +1 and Thank you for these informative posts.

I am going to load up Linux on my desktop tonight and see how it goes. Going use a separate drive just in case though. :)

Re:Can't wait to see the performance comparisons

[TheDarkMaster]

Hum... If these benchmarks are true, so now you have my attention. I never thought about using Linux seriously as a desktop because of the poor performance of GUIs and the ugly mess that are the desktop applications for the same, but if the AAA games are getting to have good performance on it then is a step in the right direction.

Re:Can't wait to see the performance comparisons

[TheDarkMaster]

To be fair, the engine Unity is getting better https://www.youtube.com/watch?...

Re:Can't wait to see the performance comparisons

[aliquis]

I don't really get the need for the attitude, if there existed zero evidence for it then I wouldn't mention it, however as you said in my case "don't believe that's intentional" maybe the same can be said about the harshness in your comment. But I assume it was, but you kinda covered it up by kinda stating being too harsh may not be valid because my "lies" wasn't intentional so whatever.

Over to the posts.
The stuff I've seen HAS BEEN lower performance on SteamOS. As for WHY that's is the case I don't really care. The main reason reason people would be against switching OS would likely be lack of applications, in this case the games they are or want to be playing, the second reason for gamers would likely be for lower performance if they switched. Lower performance may cut it for some ideologists but it won't cut it for the majority which focus about THE GAMES and not the operating system.

There's what is now old tests of early versions of SteamOS, maybe I really shouldn't use those.
Here's one on Arstechnica from 13th November 2015, it shows performance for ValveÂs own Source-based games:
http://arstechnica.com/gaming/...
SteamOS 2.0 vs Windows 10:
Portal: 107.1 vs 146.2 FPS
Team Fortress 2: 89.2 vs 114.3 FPS
Left for Dead 2: 49.1 vs 50.1 FPS
DOTA 2 (Source 2 version?): 60.0 vs 70.6 FPS.

They also cover Metro: Last Light Redux:
Min settings, SteamOS 2 vs Windows 10: 40.0 vs 50.5 FPS
Max settings, SteamOS 2 vs Windows 10: 4.2 vs 9.5 FPS
and Middle-earth: Shadow of Mordor:
Lowest, SteamOS 2 vs Windows 10: 61 vs 95.5 FPS
Low, SteamOS 2 vs Windows 10: 55.3 vs 87.0 FPS
Medium, SteamOS 2 vs Windows 10: 42.1 vs 63.3 FPS
High, SteamOS 2 vs Windows 10: 39.2 vs 50.7 FPS
Very high, SteamOS 2 vs Windows 10: 35.9 vs 46.9 FPS
Ultra, SteamOS 2 vs Windows 10: 14.6 vs 34.5 FPS

Phoronix, 6th August 2015, test performance of Ubuntu OpenGL vs Windows 10, http://www.phoronix.com/scan.p...:
OpenArena, slight lead for Ubuntu, at most by 12.4%.
Xonotic, massive lead for Windows, at most 344.9% faster.

Arma III about the same in the clip you pointed out I'd say, but maybe I would had preferred the Linux version anyway because it's the slowest parts of the game-play which matter the most.
Googled for Total War on Steam OS and Windows and found this of TW: Attila which shows better performance on SteamOS: https://www.youtube.com/watch?...

16th November 2015, SteamOS vs Windows 10:
https://www.youtube.com/watch?...
Geekbench: Winner Windows 10.
Unigine Heaven: Winner Windows 10 (46.5 vs 48.8 FPS.)
Borderlands 2: Winner Windows 10 (34.1 vs 38.6 FPS.)
Metro: LL: Winner Windows 10 (37.5 vs 42.3 FPS.)
Alien: Isolation: Winner Windows 10 (46.8 vs 54.2 FPS.)
Shadow of Mordor: Winner Windows 10 (~42 vs ~47 FPS.)

8 February 2015, CS:GO:
https://www.youtube.com/watch?...
Kinda a wash, either side leads at times.

As for Vulkan:
17 February 2016: http://www.anandtech.com/show/...
Fury X:
OpenGL: 51.8 FPS
Vulkan: 56.9 FPS (beta)
DX11: 97.8 FPS
980Ti:
OpenGL: 62.6 FPS
Vulkan: 65.8 FPS
DX11: 91 FPS

Not on topic but somewhat related I think the performance using DX12 in Gears of wars was worse too? Or was it just that AMD cards performed worse with that version? (It is a GameWorks title.)

As I said with completely new games and engines from people who know what they are doing that may change but as is anyone who owns The Talos Principle who were exited about Vulkan support and who hopped for even bet

More Wayland & Vulkan: GOOD

[CajunArson]

Thank you Nvidia. It's about time that there was a broader support base for Wayland, and the launch-day support of Vulkan on Linux has been quite positive too.

Re:More Wayland & Vulkan: GOOD

[CajunArson]

What exactly does Wayland bring to the table that X doesn't?

You know every single vitriolic attack that's ever been made against SystemD.... well you could apply pretty much all of them to the X server with the main difference being that with X server they are actually true most of the time, while with SystemD most of the more salacious ones are flat out wrong and are just copy-n-pasted as trolls.

Further, if we were to take a Venn diagram of the subset of people who foam at the mouth with hate about SystemD and the subset of people who foam at the mouth with hate at the concept of using a graphics stack that isn't rooted in mid-1980s graphics hardware, there's a disturbingly large overlap.

Re:More Wayland & Vulkan: GOOD

[rahvin112]

In 99% of cases X is using a remote framebuffer scheme as well, it just does so in a way that most people don't know it's happening.

Re:More Wayland & Vulkan: GOOD

[Anne Thwacks]

We're no longer scraping for every cpu cycle on a 16 mhz 386 or 68030.

I married a mutant werewolf 386, you insensitive clod!

Re:More Wayland & Vulkan: GOOD

[CajunArson]

Worth mentioning is that you'll find extremely arrogant and ignorant users advocate that remote transparency is a minority use case,

Never seen that once on Slashdot. Ever.
Seen PLENTY of extremely arrogant and ignorant users with a God complex because they forwarded an xterm once* and decided that all technological progress in graphics must be halted permanently in the name of "network transparency" using one type of rather crufty protocol that performs like absolute shit over any non-LAN connection.

What's even funnier is how the word "network transparency" is repeated by these same people like a religious chant when they don't even realize that literally any modern graphical application shoved over a network socket in X is *not* network transparent but is pushing frame buffers in a massively less efficient way than all the other remote display protocols that these same people viciously attack as not being "pure" for some rather undefined value of "pure".

* And didn't realize how stupid forwarding an xterm over the SSH tunnel that's required for use with modern X is.

Re:More Wayland & Vulkan: GOOD

[serviscope_minor]

any modern graphical application shoved over a network socket in X is *not* network transparent

The death of network transparency has been greatly exaggerated. I can ssh to another machine, start a modern (or old) program and it pops up on my local display. You might not like how the network transparency is implemented, but the claim that it doesn't exist is patently false.

Re:More Wayland & Vulkan: GOOD

[CajunArson]

" You might not like how the network transparency is implemented, "

Yeah, I don't and neither do most people who actually have to do this stuff for something other than toy applications. Speaking of which, why the hell should a separate SSH server & login session be a requirement for secure forwarding of graphical applications? You're own attacks on any progress in technology whatsoever start to ring hollow when you're "perfect" solution includes a built-in kludge that accounts for X's inadequacies.

The way you make it sound, Vulkan (or hell, any version of OpenGL after 1.0 or so) never should have been developed since 3D graphics already existed and people should just shutup and enjoy GLXGears forever.

Re:More Wayland & Vulkan: GOOD

[serviscope_minor]

Yeah, I don't and neither do most people who actually have to do this stuff for something other than toy applications.

So first you said it doesn't exist, now you say it does but you don't like it. Which is it? Please make up your mind.

Speaking of which, why the hell should a separate SSH server & login session be a requirement for secure forwarding of graphical applications?

Because it works well enough that no one's bothered to implenent a different mechanism? I'm guessing people tend to run X over ssh for the same reason that most people run rsync over ssh rather than rsync:// and why some people run RDP and VNC over ssh.

Basically SSH is good and you can rely on one group to implement authentication and encryption really really well. Then you can use dumb, plaintext sockets naievly but get the same security you'd have as if you had done as good a job as the ssh developers.

The way you make it sound,

I took you to task for making up blatant lies (you literally claimed remote windowing didn't work). In your mind that's apparently the same as opposing technological progress.

Jeez, with friends like you, Wayland advocates don't need enemies!

Re:More Wayland & Vulkan: GOOD

[jedidiah]

> Network transparency is also a pretty narrow edge case. There hundreds of millions of remote screen devices and deployments today

It makes for a better argument if you don't immediately contradict yourself.

Also, conflating RDP and VNC is just moronic. Microsoft has made RDP a worthy offering. VNC is just a nightmare and the prime example of why you don't want to treat network transparency as an afterthought.

As bad as X is supposed to be, it's still better than VNC. It's WAY better than VNC. X with a few tweaks is almost on par with RDP (even going across the Internet).

Network transparency is by no means a narrow edge case as the example of RDP demonstrates. It's now a common feature that the vast majority of corporate users take for granted.

It's not 1994 anymore. While you X haters were stuck in your bubble the world moved on.

Re:More Wayland & Vulkan: GOOD

[jedidiah]

> You know every single vitriolic attack that's ever been made against SystemD...

Except you didn't really answer the question.

If anything, X is much like init. For many of us it "just works" and gets stuff done. It sits quietly in the background and does it's thing. It does not make itself a problem.

"It's icky and crufty" isn't really an answer either for X or init.

It seems like the best thing the Wayland fanboys can say for themselves is that you won't notice the difference. That's if we are exceedingly lucky and Wayland development is unnaturally devoid of problems, bugs, or other unexpected issues.

Re:More Wayland & Vulkan: GOOD

[jedidiah]

> X is *not* network transparent but is pushing frame buffers in a massively less efficient way than all the other remote display protocols

You're on crack, or just a clueless idiot that's never actually used this stuff. There seems to be a lot of this going around. People like to copy Apple products without actually having used them.

Case in point: VNC versus X. This is one great example that contradicts your statement. VNC is god awful. It is slow. It's so slow that it doesn't even work well on a Gigabit LAN. If it is "more efficient" than X then it certainly doesn't show. It's a total PIG.

Re:More Wayland & Vulkan: GOOD

[jedidiah]

By all means try. It might be amusing.

In the meantime, I will be utilizing a remote GUI to get some work done.

Re:More Wayland & Vulkan: GOOD

[CajunArson]

Easiest way to find somebody who has a God complex because he forwarded an xterm once and thinks that everyone else is an idiot: Correctly point out that the modern X server running any real application made this century is no longer network transparent.

Bonus points for the idiots not understanding the fact that the ability to forward a window (poorly) over an X server does *NOT* make the X server "network transparent" because the word "transparent" has a specific technical meaning that the little bot with the God complex doesn't understand because "details" are beneath him.

Here's a guy who has spent almost 20 years developing X talking about why Wayland is better, he also says that modern X is not network transparent.. because it isn't: https://www.youtube.com/watch?...

Watch the whole video and before you spew, remember that this guy has done more real programming for X in one day than you have done in your entire life.

Re:More Wayland & Vulkan: GOOD

[duke_cheetah2003]

Probably for the same reason when you make a tar file, you use a separate program like bzip or bzip2 to compress. Because that's the type of environment *NIX systems were designed around. A tool does one thing, and does it well, and you combine tools in clever ways to achieve the end result you're looking for. I think dragging X through SSH tunnels is awesome myself. Just goes to show how versatile X is.

Re:More Wayland & Vulkan: GOOD

[saigon_from_europe]

I was also very skeptical about Wayland advantages, esp. as I considered X to be one of most cool features of Linux/Unix systems. But when you take a look to some details, then you see that it is not so cool. When X was designed, it was designed to draw primitives - lines, fonts. And it was not designed to draw bitmaps. In its current usage, it mostly draws bitmaps, i.e. true rendering is done in applications (using libraries like GTK, KDE, Qt...) while X is just slapping it together. And it does it in a very, very complicate way. Even the network transparency is not done properly. Wayland is way more simple, more suitable to how graphics is used today. It does not have a network layer when not needed; at the same time it brings a layer where you can cut and implement network transparency if you want. X is a dead end.

With the current state of affairs, X is not too useful. For example, there are very few X implementations. I needed a X server* for Windows. There are only two of them that I was able to find - one proprietary Hummingbird's and all others are based on Cygwin. Another example - rendering of fonts depends on the font library on the client's* machine - that should not be if X was used as originally planned.

*) Not to mention the counter-intuitive names for "client" and "server"

Re:More Wayland & Vulkan: GOOD

[serviscope_minor]

A server dishes out resources. In this case the resource is screen. It's counterintuitive if you think server= compute server, but there are other examples like print server too. It does make sense.

Re:More Wayland & Vulkan: GOOD

[dbIII]

Yet some of those older Kindles mentioned in another article use X. My phone uses X. Even a very low end WinCE terminal thing I have at home uses X. Somehow it works even on very crappy hardware incapable of running Wayland while you are pretending it does not.

Re:More Wayland & Vulkan: GOOD

[dbIII]

Never seen that once on Slashdot. Ever.

Lucky for you. While you've been asleep we've had to put up with that "X Sux" shit every fucking time Wayland is mentioned.

Re:More Wayland & Vulkan: GOOD

[dbIII]

Better performance.

I asked you for benchmarks last time and you had nothing to back up that statement. It may be the aim but Wayland is not performing better yet - especially since most of the drivers are cut and pasted from X.

Effective remote over high latency.

WTF? Remote access is explicitly not supported at all!

Re:More Wayland & Vulkan: GOOD

[KiloByte]

Almost two decades ago, my friends and me ran four instances of Quake, running it via X forwarding from a Linux box to our IRIX workstations. It was playable.
Sadly in a window as those IRIXes had insane (for those times) display resolutions, but it worked well.

Try getting a fluent animation over VNC or similar crap today, on so many orders of magnitude better hardware.

Re:More Wayland & Vulkan: GOOD

[saigon_from_europe]

Yes, it does make sense, but it is counter-intuitive when you hear that for the first time. Exactly the case you mentioned = compute server runs clients and they attach to the server which is actually my work station. They could have chosen some other wording, no matter that technically X server is indeed a server.

Re:More Wayland & Vulkan: GOOD

[thsths]

It is not just counter-intuitive, it is also wrong. The limited resource is not your screen, but your desktop. That may be a small distinction, but it leads to all kinds of problem in X. For example, running multiple screens is still a bit of a faff. And you can run only one window manager.

In a clean architecture, the window manager would be the server, and the screens would be clients.

Re:More Wayland & Vulkan: GOOD

[Blaskowicz]

I'm not sure many are interested in multiple window managers, but if so then you run multiple X servers or a nested X server.

Re:More Wayland & Vulkan: GOOD

[dbIII]

Wayland is using an architecture that has obviously better performance

As the benchmarks would show if there were any and you were correct. Reality is that it is a work in progress so still inferior performance, which is expected and not a problem apart from clueless fanboys lying about it when there is no need to do so.

WTF is your problem?

Re:More Wayland & Vulkan: GOOD

[dbIII]

Please inform the readers how expected future developments are "delivering" those things that are not yet available.

Just in time for Ubuntu?

[jordanjay29]

It seems well timed to coincide with the release of 16.10 later this year, which, if all goes well, should use Mir by default for Unity 8. This gives NVIDIA 6 months or so for early adopters to work out the major kinks for them. Smart plan.

Re:Just in time for Ubuntu?

[rahvin112]

Mir is a joke NIH solution to a problem that's already solved with Wayland. And just like all their other NIH solutions to problems they will abandon in a year when it's clear how much it's going to cost them to support it. It's the same story at Canonical over and over again.

Re:Just in time for Ubuntu?

[jordanjay29]

That may be absolutely correct, but Ubuntu vanilla still accounts for a large number of linux "users" for which NVIDIA would be wise to target in one go like this.

Re:Just in time for Ubuntu?

[jordanjay29]

Eh, I have a hard time seeing Android replacing too much on the desktop for several years. It might be a great drop-in substitute for ChromeOS, but I don't see it encroaching on distros like Ubuntu or openSUSE or Arch anytime soon. Plus, I'll bet there are plenty of Linux users who'd prefer not to have such a tightly coupled existence to Google.

Re:Just in time for Ubuntu?

[Junta]

Of course, 'if all goes well, should use Mir by default' has been a missed goal a few times in the past.

Re:Just in time for Ubuntu?

[TheRaven64]

Mir exists, in part, because there actually are some real problems with X11, such as the complete lack of anything resembling security on the input path. These problems were not things that the Wayland developers decided to fix. I'm not a huge fan of Mir, but if you're going to replace X11 then replacing it with Wayland is at best a sideways step.

Re:Just in time for Ubuntu?

[serviscope_minor]

Mir exists, in part, because there actually are some real problems with X11, such as the complete lack of anything resembling security on the input path. These problems were not things that the Wayland developers decided to fix.

Compositing window managers intercept nearly everything so that they can re-map events to arbitrarily shaped windows etc. Which security flaws aren't easily fixed by compositing windowmanagers on X?

Re:Just in time for Ubuntu?

[andydread]

Fedora and Arch and just about every other Linux distro are NIH solutions including Android. If you are not using Slackware then you are using NIH solutions. What's your point? SystemD is a NIH solution, Wayland is a NIH solution, Pulse Audio is a NIH solution, Gnome and KDE are NIH solutions. I could go on and on. Your argument that people should not innovate to suit their own needs and use cases is moronic at best. The NIH excuse is tired and beaten to death. trying to use NIH to justify your hate for Canonical is quite adolescent at best. Grow the fuck up. - smh

Re:Just in time for Ubuntu?

[neuro88]

No, Wayland fixes all of this. In fact, this is in part why Mir uses the exact same input library as wayland: libinput.

Re:Just in time for Ubuntu?

[neuro88]

No, Wayland fixes all of this. In fact, this is in part why Mir uses the exact same input library as wayland: libinput.

Responding to my own post because I realized that citation is need:
https://lwn.net/Articles/51737...
https://blog.martin-graesslin....

Re:Just in time for Ubuntu?

[bregmata]

It is not coincidence. It's planning on the part of Canonical and nVidia.

Vulkan is the important bit

[zhen.sydow]

Nowadays the headline should be "NVIDIA's Proprietary Linux Driver Adds Vulkan 1.0" and in the article text you can also mention Wayland and Mir.

Re:Vulkan is the important bit

[CajunArson]

Except that Vulkan support isn't news.
Nvidia had Vulkan support for Linux on launch day. The news is the official Wayland support.

Re:Vice versa

[mrvan]

That is not how you use vice versa. You just ruined English and Latin in one go.

That's not how you contribute to the Internet. You've just ruined a post and my time in one go! :)

Re:How can I make tech work for me

[jedidiah]

??? You are switching to the libre driver because of VDPAU? That makes no sense.

This lie again?

[dbIII]

This lie again? The same guy also said only five people understood X input. I was once on a mailing list with about three hundred of those five.
The lie is based on how the current version of GTK is designed to rely on fast local graphics hardware and does not support being displayed remotely so X gets the job done with a fallback. The fuckup is in GTK not X.

VR wayland desktop!!!

[jmooney538]

Now we can have nicely accelerated VR linux desktops. https://www.youtube.com/watch?... https://github.com/evil0sheep/...