(Originally this was supposed to be a comment on "The Tek 0164", but after a user's suggestion I made it a blog post. So here it is. Enjoy.)
TL;DR - Read the title, walk away feeling satisfied you know something despite learning nothing for not reading the post.
TL;DR of the TL;DR - Don't be lazy, read the post.
TL;DR of the TL;DR of the TL;DR - Sarcasm!
Depends on how much traction SteamOS gets. We've seen a lot of movement from Indie game devs towards releasing games on Linux. And we've also seen quite a few older games being ported to Linux. I've also seen a lot of newer (not AAA) titles being released on Linux (along with Windows, and occasionally consoles as well).
The issue really is market share. Nvidia puts all their eggs with whoever has the biggest market share, and right now that's Windows. But that means once that big market share becomes much more divided, AMD would have the software and hardware upper hand, because they'd have more experience with Linux, Linus Torvalds would be more than motivated to unlock even more power from AMD on the kernel level, and we could even see more improvements coming from the Open Source community once Linux has a significant market share (say, 30% or more of online game sales being made on Linux-enabled devices).
Game play time doesn't matter. Companies care about sales, not how much time the player spends on their game (unless it's an MMO game, since activity from players would definitely affect sales, as it has with WoW and GW2, for example).
Nvidia will only move towards making their hardware and software work properly on Linux once it would hurt them more financially to ignore Linux than it would benefit them from reinvesting that money into Windows development (for drivers, software, installation, usability, UI, UX, etc).
The thing is, Nvidia doesn't give two s**ts about the user or the player. Nvidia is about profit, pure and simple. They never cared about Linux, they care about money. They don't care about OpenCL or OpenCompute unless it helps them sell more GPUs (and that's why AMD has had the upper hand in those areas). The BitCoin craze was a big demonstration of that. Adobe using OpenCL for the hardware acceleration of Photoshop and Creative Cloud is also another example of this. Companies don't like being hostages to a closed standard by a single company, since that means said company could at any moment abandon development of said standard or cease to develop products for it. If Nvidia did that with CUDA (but they wouldn't), Adobe would have lost millions of dollars in software development of CUDA-specific code. Adobe chose OpenCL because an open standard gives them (and their users) hardware flexibility of choice, allowing their clients (individuals and companies) to have near-adequate performance on Nvidia GPUs, but also have excellent performance on AMD GPUs, instead of great performance on Nvidia GPUs and no performance/acceleration from AMD GPUs. That inspires more confidence from clients and a better, more enjoyable experience, which in turns increases sales (and decreases complaints and bad reviews from users, also driving up sales).
Nvidia is all about money, that's why they make more money than AMD. AMD is about creating a better future; they helped develop DockPort (you should check it out - think of USB 3.1 + 100W power, side by side with DisplayPort 1.3, over a single cable sending the signals in a way similar to PCIe). AMD wants to make a better future, one that's inclusive, and they're playing the market for the long game by making sure they can compete in all arenas - Windows, Linux, ARM, etc.
AMD knows that being able to have the option to change where and how they compete and against whom is a very important thing in a market as subject to abrupt change as the hardware and software industries are. Nvidia doesn't know that, and has put all their eggs into a few big baskets: Windows, CUDA and PhysX. To an extent the server space is also somewhere they're competing in, but not really. The monitor space with G-Sync is interesting, but we'll have to wait for it - same with their GRID streaming service.
If there's a big change in the game soon that drives people away from Windows, we're going to see AMD pounce on that opportunity like a Hollywood producer on the rights to a script of "FernGully, The Prequel: The Before Story you never cared about, now with more gimmicky CGI than you can shake a stick at". AMD is ready for abrupt change in such a volatile market. Nvidia isn't. AMD has other markets it can tap into if the market share changes. And they're willing to sell products with lower profit margins (even in business-to-business sales), which is why they got big contracts like XBox One, WiiU, and PS4.
Nvidia can't compete with that. As a result, fewer AAA games that are cross-platform are coming out with PhysX support - the lower number of titles since 2010 (only 4): Warframe(68/100 on Metacritic), Daylight (50/100 on Metacritic), Borderlands: The Pre-Sequel (75/100 on Metacritic, only with PhysX support because it's basically BL2 redone, and BL2 had PhysX), Lords of the Fallen (73/100 on Metacritic) - and none of those games were anything special, to be honest. That means that only mediocre games included PhysX in 2014, even after Nvidia announced they would allow PhysX to work on the Xbox One and PS4. Meaning that PhysX is looking to be in bad shape. [1]
Nvidia does have some cool games that might come out with PhysX support in the future. EverQuest Next, Batman: Arkham Knight, Star Citizen, Project CARS, and The Witcher 3 are just a few on the list (but those are the only big releases; the rest is just fodder for increasing the number size in 2015 for the marketing team's commercials). [2]
AMD has given their clients, both corporate and individuals, flexibility to choose which hardware they want to buy. That gave them confidence and flexibility if they chose to buy something else later. Nvidia didn't to that. AMD gave their clients the choice of OS, CPU, GPU, or everything else in between, something that neither Intel nor Nvidia has been able to do. AMD is ready for whatever change might occur. Nvidia is still depending greatly on Windows, and while that's their strength, it's also their crutch. It's up to AMD to make the most out of this by helping Linus Torvalds and the Open-Source community. For AMD, improving Linux and OpenSource is more than just part of their corporate philosophy of choice and flexibility - it also means undermining the very foundation of their biggest hardware competitors, Intel and Nvidia. And AMD can capitalize on this, but only if their pounce on that opportunity while it still exists. Giving Nvidia and Intel time to adapt and adjust is bad business for AMD, and I sincerely hope AMD makes the smart move by doing even more to make the user transition from Windows to Linux even easier and smoother, and hopefully offering some financial support for the development Open Source versions of essential (read: Adobe Creative Cloud, foobar2000, or 1st World Problems) programs that aren't yet available on Linux. Once people can make the switch and the market share changes, suddenly developers will have to make a choice. And if one area of the market share is continuously growing, while the other is shrinking, you can bet it'll make software and game developers consider Linux as a good option. As more software and games get published on Linux because of Market Share, the more we'll see even more users migrate over.
It's this vicious cycle which has kept users locked in the Windows ecosystem. But if the grass is greener on the other side, and people can show that to Windows users, and also give them for free the OS and programs they want, convenience and cost will make Windows irrelevant. By not having to use Internet Explorer, reset your computer when you get updates, or deal with any of Window's multiple issues, convenience, cost and simplicity can win over users.
Linus Torvalds was also talking about the distributions he like, and how he no longer uses Debian, Fedora or Ubuntu, because he has to take care of the PCs in his house. And when his [Linus Torvalds'] daughter called him saying she needed the administrator password to print a document, he no longer could recommend those distributions to anyone. I think he's on to something here: the user interface (UI) and user experience (UX) are the most essential things for the user, just like how gameplay is the most important thing for a gamer, since that's the thing both of them will be doing when interacting with their software. If that doesn't deliver a good experience, or if it gets in the way of the user (or player), they'll pick the best (or least worst?) option they can. If AMD can support app developers and also support Linux UI/UX developers to make the experience of installing and using Linux much simpler for the user, we could see a big shift. But AMD would also have to use marketing to let user's know that the grass is greener on the other side.
So there is hope for Linux and AMD. And AMD has the ability to undermine their competition by strengthening the main competitor (Linux) of the crutch (Windows) which their competitors rely on most. That could be AMD's next big play. I can't wait to see how this unfolds.
Cheers!
UPDATE: Because monitor scaler manufacturers aren't allow to make G-Sync, since it isn't an Open Standard (nor an Industry Standard), monitor scaler manufacturers (which are present in pretty much any desktop monitor you can buy that isn't G-Sync) have "pledged their allegiance" to Adaptive-Sync (DisplayPort's version of AMD's FreeSync). [3] AMD has made Nvidia's big hype into an industry standard, making Nvidia's spending on marketing suddenly much less valuable, as well as taking away much of the ROI (Return on Investment) that would have come from the money spend on the R&D (Research and Development) of G-Sync.
Also, it seems many monitor scaler manufacturers will just need a simple firmware update on their scalers in order to support AdaptiveSync (DisplayPort 1.2a's implementation of variable refresh rates, or AMD's FreeSync). [4] This means many monitors out there right now, if it were possible to easily update the firmware of the monitor, could be freely converted to AdaptiveSync monitors (as long as they have a DisplayPort connector on them). You can't get cheaper than a free firmware update, as opposed to paying an extra 100$ (or so) dollars for G-Sync. Money you could spend on a better AMD GPU, like going from an R9 280X (around 230$) to an R9 290 with aftermarket heatsink (around 330$).
UPDATE (2) : While I've said AMD should support app developers directly, there are some apps which I don't think AMD would be able to support. An Open-Source competitor to Adobe CreativeCloud would be against their best interests, since Adobe's adoption of OpenCL/OpenCompute is huge - it enables them to get far more hardware sales and profits, as well as helping bring Adobe away from the whole proprietary nonsense of CUDA.
Another point is that the list of apps which AMD should support (financially) are:
LibreOffice - Fund better compatibility, so switching over is easier. Also add new features, like better image background placement, scaling, positioning, filters, etc. Closer similarity to PDF format and what's shown on-screen. Maybe add some extra features specific to book writers, like different text alignment options, different font size options (like small capitals, where the size of the small capitals is variable rather than fixed, so many different types of fonts can look good too). Adding better hardware acceleration for LibreDraw could also help.
Firefox - Fund better hardware (GPU) acceleration on Linux using OpenCL/OpenCompute, so there's better performance, and also fund better CPU threaded optimization (removing single-threaded bottlenecks wherever possible).
Filesystems for Linux - Adding better ECC/Parity/Checksum of files (without introducing much read/write latency) could be very important. Having corrupted files because of hard drive problems or unexpected shutdown can be a huge frustration on Linux, and if that makes the user's experience with Linux better, more power to them. (This is one advantage ZFS has over ext4.) Also, adding functionality that allows only certain files (like system-critical and kernel files) within the filesystem to have error-checking (rather than an all-or-none approach) could save space and read/write time. We also need Integrated LVM (Logical Volume Management) so partitions don't have to be used on non-UEFI devices, but it could also allow for better caching to happen on-the-fly. Having a file change log that doesn't destroy previous data would also be amazing - that way a file could be rolled back if/when needed, and file corruptions wouldn't be the end of the world. (Also, having a file system that could have multiple copies of system-critical files and the kernel could also save people's bacon in case of data corruption on one area of a hard drive, or due to NAND cell damage on an SSD. It would increase reliability. Not only multiple copies on a single partition, but also on multiple storage devices.)
Automatic Caching for Linux - Having RAM and SSD Caching happen automatically could be a big performance increase for Linux. It could also save on energy consumption on mobile devices, such as laptops, Intel tablets and smartphones, because if a device has more RAM than it's using, being able to read from RAM or an SSD rather than the HDD could decrease energy consumption. Having Linux lead the mobile industry with smart energy saving through software caching could be huge, and being able to take Linux to the mobile computer/tablet arena with Intel's new Tablet/PC hybrids (as shown in CES '15) could be a great way for people to be introduced to a better online experience. This would require better prediction of what the user might use, based on what programs the user uses, how often, when they're used, and the benchmark data of the random reads and sequential reads of the storage devices they use (for example, where is it best to store file X? SSD or HDD? K-brand SSD, or I-brand SSD?).
Better music players - foobar2000 is a great music player for Windows, but AMD could bring that to Linux as well. AMD's TrueAudio wouldn't be needed for a standard music player, but using VR and TrueAudio, we could see AMD make listening to classic music in a concert hall something that's enabled using an Oculus Rift, for example. That could make the music listening experience much more interesting, where someone isn't just listening to music in the background, but enjoying it in a virtual world with physics-enabled acoustics for greater realism. AMD has the technology, they just have to use it and support developers.
Better image and video editing software - AMD could create "bounties" for certain improvements they'd like done on GIMP, like more effects, different filters, extra plug-ins. While it wouldn't compete toe-to-toe with Adobe Creative Cloud, it could definitely help get more people moving to Linux. As for video editing software, support LightWorks is all AMD needs to do. Find what's missing from Sony Vegas and other professional video editing software, add "bounties" (read: cash rewards to the first person who publishes complete open-source code to add the desired functionality in, much like how security vulnerability research is done by Google).
UI / UX - Linux could use an overhaul, including their desktops. Having something that looks and feels familiar to the Apple and Windows desktops is important, and while KDE and GNOME have improved a lot, they still aren't quite up to the same level of polish or familiarity that Windows and Apple have on their desktops. So why not change that? Make it simple, intuitive, and fast! LibreOffice has some great guidelines [5], and I think Linux should learn from them (even Linus Torvalds mentioned these, and how some distros seemed to blame the user for not using the software right, which he really didn't like).
While command-line mode may be great for veterans who know what they're doing, the simple truth is that most user's don't. And they need a UI that's simple enough for them to grasp, use and like. The UX (User Experience) needs to not be an unpleasant one, so fixing problems like requiring the administrator password to print can't keep on happening. The usage of Linux needs to be simple, for simple-minded users, so the UX and UI should reflect that. But that doesn't mean Linux can't have an "Uber-mode" or other distros that aren't meant for advanced users. Linux just needs an "entry-level distro" with a catchy name that your Average Joe can remember (like iOS, Android, Windows, etc), a simple interface, and something that's powerful enough to take care of work, simple but so much that it gets in the user's way, and slim (read: minimalistic and light on RAM, CPU, GPU) so that it can use the performance on things the user actually cares about (because let's face it, when's the last time you heard your average user complain about the kernel's performance or kernal features rather than a slow-loading app?).
(Having a bright, simple UI that focuses on readability that's also polished, like Apple's could help make the OS seem more friendly to users. Having the UI also avoid being too flashy could help make it more attractive to business owners, and also make it perform better on legacy hardware, so long as it doesn't look outdated.)
Better online backup and service integration - Seagate and other companies have talked a lot about "the cloud" and having a "personal cloud". Why not add some type of service hosted by AMD that acts like a dynamic OpenDNS server to allow users to connect to their home NAS, of their service of choice (Dropbox, Mega, or whatever). It has to be simple, and something that just works, allowing them to set up a username, connect their devices, and then let them sync/backup as needed (with incremental backups, rather than full copies; the whole conservation of bandwidth thing).
Better included games - While Linux has several games, why not make the included ones better? It may seem small, but having something for the kids or the teenagers who visit could be an incentive for parents and grandparents to adopt the OS.
Better parental filters - Linux could greatly benefit from this. Having parental filters could make Linux go from the perception of being "super genius difficult" or "built in someone's basement" to a very friendly (and family-friendly!) ecosystem (like the perception people have of Apple or Nintendo). This could help increase OS adoption.
Better OS and driver installation experience, and updating experience - Linux needs a simpler installation experience, as Linus Torvalds did say once (infamously). He also said that he hadn't tried installing Debian or other distros in a long while, though. As a Windows user, I'm often frustrated when I can't find the correct driver to install (because it's buried in some website designed by a chinese person who wants to see the world burn, because they don't understand that plain text and links need to be in different colors!). I'd love to see an OS that identifies the controller and manufacturer, and then installs the proper drivers automatically (and updates them automatically too, downloading only the incremental changes from version to version rather than a whole copy of the installation file). I'd also like to have a better updating experience, since it majorly s*cks on Windows.
Better default apps - I think Linux could benefit from having better default apps within a distro. LibreOffice is a better office suite, and I think having it included by default would make distros better.
Faster Desktop Environments - KDE and GNOME are crazy fast. But AMD has some great hardware acceleration experience, and they know a thing or two about writing software to use it. So why not let the KDE, Gnome and Cinnamon team benefit from their expertise by lending them some of their programmers to help them improve the performance, and maybe allow them to make the graphics more beautiful and polished with the same performance as before, or better performance while looking the same.? AMD could only benefit from a stronger Linux, and Nvidia and Intel can only lose from a stronger Linux. It's a win-win, plus great publicity for AMD.
Professional applications for small and medium businesses - Payroll software, POS (point of sale) cash register software, contact information, groupware software (for collaborative work), group chat (VoIP, text, VR, telepresence), telecommute, etc. Linux could take over a large portion of the industry by offering small to medium business owners a simple to use set of tools. While it wouldn't be possible to have custom code for each and every business out there, having WYSIWYG editors with a drag-n-drop type interface could help business owners tailor a base software to their needs, and creating online tutorials (like on YouTube) that are professionally-made (and AMD should know something about video creation, with their FirePro cards and whatnot) could help improve adoption of Linux, since many business are dependent on those types of software. Also, having conversion software to convert old legacy data into other formats could also help.
AMD needs its own distro - There are many distros out there. Too many, actually. SteamOS is ready to show the world that Linux is ready for gaming. But AMD needs to show the world that Linux is ready for the desktop, at home and at work. Few companies will work on Linux until it has greater adoption, and for Linux to get there it needs help. AMD can create a distro to show the world how it's supposed to be done, by doing it right the first time. Only then can Linux take over the business workstation and the home desktop, currently dominated by Intel and Nvidia. If AMD takes off, neither Intel nor Nvidia can compete on Linux, so AMD would regain the performance crown once more, and not only that but have more experience on their software team, giving them a real leg up on their competition. And the increased sales of their hardware due to better performance AND being at a lower price than their competition could increase their profits, which could be further invested into funding Linux app development to further undermine their competition, but which could give them free publicity.
UPDATE (3):
Make Linux Context-Aware, with Syncing possible - Users don't like to have to change how they do things just because they switch users. Having better synchronization could allow multiple devices to be synced in the same ecosystem. Apple has this with their iCloud, but why not allow Linux users to configure their own storage medium or service of choice (NAS as home, NAS at work, Dropbox, Mega) and let them sync their files. Having Linux be context-aware could allow personal files to be encrypted on certain devices by default (like on your smartphone or laptop), but not on others (like your workstation). It could also show or hide icons or programs from your desktop depending on what time of day it is, what day of the week it is, where you're connecting from, or which device you're connecting from.
If you're at home, connecting from your own laptop then you shouldn't have much of an issue having that poker program for Linux running. But if you're at work, or using the company's laptop, then not so much. Having Linux sync files depending on context is very nice. It paves the way for more Linux-based tablets, smartphones and other devices, but also for an Internet-of-Things ecosystem for Linux. The big concern is the setup and UI - Linux is perceived as complicated or difficult, so the setup and usage needs to be simple, and there needs to be a clear separation between basic and advanced settings and features.
Linux needs simple, professional, friendly tutorials on YouTube and social media - For Linux to become more accessible, it needs to be taught. While often people in the Linux community may frown on "holding the user's hand" through everything, that's something the user may end up liking. People don't want to tinker and experiment, at least not most people. Most people want something that works so they can go on with their lives, which is why most people don't program, mod their games or build their own PCs. We're a minority in the world, and while we're [the Tek and TekSyndicate users] awesome in our own way, we're not the typical user. Having a simple UI and UX is great, but there needs to be something to spread the word. Social media works great for that, and having video tutorials to help grandma or my aunt install Linux on their machines could help a lot.
There's no reason why Linux couldn't benefit from simple, professional, well-made, friendly tutorial videos. They could even be included in the distro, if using compressed audio, video (maybe video with transparency to conserve on data, and also to not occupy too much of the user's screen), and perhaps presentation software for text, images and button elements. Having an interactive presentation that works like a game tutorial with a friendly voice and simple, clear instructions/text could really help newcomers to pick things up faster. If tutorials can work well for some games, there's no reason it couldn't work for Linux too.
[1] - https://en.wikipedia.org/wiki/List_of_games_with_hardware-accelerated_PhysX_support (sort by Release Date)
[2] - http://physxinfo.com/index.php?p=gam&f=indev
[3] - http://techreport.com/news/27071/monitor-scaler-makers-commit-to-freesync-hardware
[4] - http://www.tomshardware.com/news/amd-project-freesync-vesa-adaptive-sync,27160.html
[5] - https://wiki.documentfoundation.org/Design/Principles