DirectX is coming to the Windows Subsystem for Linux • Hacker News Board

Re: DirectX is coming to the Windows Subsystem for Linux

#531

Post by golergka » Thu, May 21, 2020, 6:09 PM UTC

cycloptic wrote: post #494

golergka wrote: post #480

Earlier quoted context omitted.

These prominent open source companies that you're taking about make money from consulting or cloud services. There's no well understood and well rested way to do that as a game company. It doesn't mean that there's no such way, but it does mean that attempting to find it is very risky. Making games is already a very high-risk, high-reward industry, do adding this amount of risk to the equation is advice that's insane…

A multiplayer game is just like any other cloud service in terms of economics. And a game that allows a certain degree of customization is analogous to a consultancy. Find the pain points your business customers/partners are having and then charge for solutions. If you don't have any business customers/partners then get some, not having them is a much bigger risk than choosing any specific development model. One of t…

> A multiplayer game is just like any other cloud service in terms of economics.

No, it's not - this is an absurd statement. Cloud services get most of their income from big and medium companies, billing tens and thousands of dollars in privately negotiated agreements. Digital Ocean and AWS don't live on single developers paying them $5. All their real money is on B2B.

Multiplayer games either get a flat subscription from all their playerbase, or rely on micropayments. Either way, they spend much less on account management where they interact with customers as individuals, and much more on analytics, where they measure and work with their customers in aggregate. These are completely different business models, that completely shape companies that operate on them.

> Find the pain points your business customers/partners are having and then charge for solutions.

Games are not solving any problems - they are entiertainment product. I've seen people who have tried to reason about games in the terms you're using, and it always yelded hillarious (but also sad) results.

Re: DirectX is coming to the Windows Subsystem for Linux

#532

Post by prirun » Thu, May 21, 2020, 6:27 PM UTC

scbrg wrote: post #381

m12k wrote: post #346

Earlier quoted context omitted.

Could you explain to me what you think an Extinguish step would look like in that sequence? Because I don't think there is a realistic one, and I don't think Microsoft thinks so either.

What you're saying is essentially: If we can't come up with a way in which Microsoft could proceed to the Extinguish phase, they can't. Because clearly, if we , after spending five minutes thinking about it, can't come up with something, then it is also impossible for this company that has spent three decades perfecting the concept, and literally has billions of dollars to throw at the problem. Presumably because we…

I agree with you. Microsoft plays a long game. Through giving out promotional / free copies of Excel, with 1-2-3 compatiblity, they killed Lotus. The same with Word and Wordperfect. The same with Netscape. And those companies completely dominated their markets. As market leaders maybe they were lazy, taking profits, and not enhancing their product. That happens with most market leaders. But the fact is, it is impossible to charge $150 for WordPerfect when Microsoft is giving away a compatible word processor. As their only / primary source of revenue, WordPerfect and Lotus could not survive a free competitor.

They don't have to kill Linux, and in fact, would not want that IMO, because then someone might bring another pesky antitrust suit against them. But if they can dent it in any way, they will.

For those who say Microsoft has changed, is a different company, is embracing open-source, etc.: if that truly is the case, wouldn't they release DirectX for all Linux distributions, not just theirs?

Re: DirectX is coming to the Windows Subsystem for Linux

#533

Post by prirun » Thu, May 21, 2020, 7:19 PM UTC

Here's the "new" Microsoft:

https://news.ycombinator.com/item?id=23255591

Re: DirectX is coming to the Windows Subsystem for Linux

#534

Post by monocasa » Thu, May 21, 2020, 7:37 PM UTC

wtallis wrote: post #530

monocasa wrote: post #487

Earlier quoted context omitted.

SR-IOV doesn't make sense on most GPUs, because they have their own MMUs already with different semantics. SR-IOV is terrible for dealing with a large bank of RAM on the target device, because the MMU is in the wrong part of the architecture. It's all the way on the root complex, so all GPU VRAM reads and writes would need to take the slow path through PCI-E and back. This is basically a nonstarter for GPUs that have…

How does AMD deal with this on their server discrete GPUs that support SR-IOV?

It's not really SR-IOV in the way you'd generally consider it as it only virtualizes the GPU's access to main memory not VRAM. That still requires that you setup GPU page tables and manage them on the host side with a bridge driver in the guest like Microsoft's work here is doing.

The same with Nvidia's A100.

Re: DirectX is coming to the Windows Subsystem for Linux

#535

Post by mcv » Thu, May 21, 2020, 11:20 PM UTC

simiones wrote: post #522

mcv wrote: post #517

Earlier quoted context omitted.

Sure, companies may love it, but developers hate it. I use a Windows laptop at the company I currently work for, because everything is locked down and I wouldn't be able to get my own laptop connected to the network. (Or so I thought; I co-worker managed to use the Windows laptop as a bridge to his own Macbook.) Now you're right that as long as I stay in the IDE, it's not so bad. But every once in a while I need to d…

You generally have some good points, and some points that have more to do with preferences, or luck with hardware. I would note that the lack of anti-virus software is one of the reasons a lot of companies don't want to run Linux on employee systems. Locked-in Windows systems at least have the advantage that even if you Run as Administrator some .exe that you got emailed, there is a good chance that your AV will not…

I admit I've never really looked into how detailed Linux is in access rights, but on Unix systems, it's very normal to have install rights for specific directories. If Linux doesn't allow that, that would be disappointing, but I strongly suspect Linux allows this just as much as other unixen. So that would mean you can install stuff without sudo rights as long as you get group rights to the right directory. And that's a much safer approach to security than all-or-nothing.

Re: DirectX is coming to the Windows Subsystem for Linux

#536

Post by halbritt » Fri, May 22, 2020, 2:17 AM UTC

NeverFade wrote: post #528

5partan wrote: post #525

Earlier quoted context omitted.

Except that Linux usage on Azure has surpassed that of Windows a long time ago...

And I was talking specifically about Microsoft products like Office365, Exchange, and Sharepoint.

Microsoft's interest in on-premises business with these products is waning, just as the market share for them is also waning. They most assuredly aren't trying to win the war against Linux on the server. That war has been won (by Linux) and MS is off chasing other revenue streams that are growing.

Re: DirectX is coming to the Windows Subsystem for Linux

#537

Post by iforgotpassword » Fri, May 22, 2020, 7:38 AM UTC

skohan wrote: post #243

iforgotpassword wrote: post #240

Earlier quoted context omitted.

While this was my initial thought as well, on a second thought that doesn't make too much sense. Well depending on whether you're wearing your tinfoil hat properly, you could say this is for testing the waters, but who the fsck would target their game at this? Write against naive Linux but use DirectX for the graphics? Just why? Sounds like the most stupid thing ever. The CUDA support thing from other comments below…

EEE is incremental by nature. This represents a lever they could use to create an ecosystem of Linux software which only works on WSL. > but who the fsck would target their game at this? It doesn't have to be games. Maybe MS releases an ML visualization library for Linux which requires D3D. It's not so far fetched. Given the cumulative pain developers have had to deal with supporting IE6 over the years, I think it's…

Eh, still not convinced. Porting that to OpenGL is half a day of work. Like I said, you might say they're testing the waters, but I'm just not paranoid enough to scream EEE yet.

Re: DirectX is coming to the Windows Subsystem for Linux

#538

Post by jillesvangurp » Fri, May 22, 2020, 8:27 AM UTC

majewsky wrote: post #495

jillesvangurp wrote: post #252

Earlier quoted context omitted.

This feels more like a reverse takeover to me. MS actually has a legacy problem in the form of the windows ecosystem. Under Ballmer this was a problem that required defensive moves and rhetoric to not lose people to Linux and OS X and it didn't work. Under Nadella, a bait and switch was initiated where the problem (windows legacy) is slowly switched out for Linux. This is not the embrace and extend policy of last cen…

> Under Nadella, a bait and switch was initiated where the problem (windows legacy) is slowly switched out for Linux. If that is what's happening, then what's the endgame? Switching WSL around so that "Windows host + Linux guest" becomes "Linux host + Windows guest", i.e. Windows becomes a Linux distribution running native Windows apps in a VM with seamless integration? I'm somewhat intrigued by the possibility, but…

A large part of their business is already running on or compatible with Linux. The endgame is that they get your money regardless of what OS you are on.

I'd expect the importance of the windows kernel for revenue growth to be increasingly less important over time. Of course they won't drop it outright; at least not right away and it's likely to stay relevant for e.g. pcs/laptops and gaming. As for games and vr content, a lot of game studios already use cross platform sdks and Linux support for games with and without emulation is pretty decent these days. Also, Android and IOS are big targets for games.

Most hardware vendors don't target windows exclusively. Some do of course but a lot of hardware works fine on other platforms even if vendors don't actively support that. Anything intended for data centers runs linux primarily. Most laptop vendors have a few linux friendly laptops at least to not lose out on the pro developer market that tends to actually buy their more expensive laptops. Likewise, most graphics card vendors want to support e.g. machine learning and that requires linux driver support.

But I get what your saying. My observation is that MS is very friendly lately with Ubuntu. I don't think Mark Shuttleworth is interesting in selling that outright but an MS distribution might be a logical next step given their increased dependency on Linux on the Desktop and in the cloud.

Re: DirectX is coming to the Windows Subsystem for Linux

#539

Post by smaudet » Fri, May 22, 2020, 4:26 PM UTC

cs702 wrote: post #439

lalaland1125 wrote: post #213

It appears that Microsoft has now initiated the "Extend" phase of their classic Embrace, Extend, Extinguish playbook. The key is that the proprietary Microsoft specific API added by this patch is only usable in a WSL environment as it relies on many pieces of proprietary closed source software that Microsoft is unwilling to open source. This patch does nothing but fragment the Linux ecosystem and encourage people to…

Keep in mind, a ton of people and companies are asking for this, because they genuinely want to use hardware-accelerated PyTorch and TensorFlow "through a Linux-thingy" on Windows. BUT I MUST AGREE WITH YOU that, over time, this is likely to fragment the Linux ecosystem and encourage people to develop software that works only in Windows "with the Linux-thingy installed" environments. That's because sooner or later, p…

I do think that's a real concern...however in some ways this is an admission of failure - users like what's in the WSL enough that they want more linux, versus just some subsystem which is consumed by other apps.

Also, I think linux users (desktop/servers) who actually write code and provide updates for programs, aren't likely to pause the world for WSL. Its not exactly 'linux' nor the destination for their code, WSL stands a risk of becoming obsolete, if not properly maintained etc.

Its also good in some sense, it exposes users to linux capabilities and the like, who might not have had the guts to run a live cd or deal with modern hardware foibles.

Re: DirectX is coming to the Windows Subsystem for Linux

#540

Post by smaudet » Fri, May 22, 2020, 4:43 PM UTC

Chouhada wrote: post #349

franga2000 wrote: post #306

Earlier quoted context omitted.

> As a game developer, for most mid-tier inside titles releasing Linux version costs more than it makes, so, no. What are the main costs of releasing a Linux version of a game when using Unity or Unreal? My limited experience with Unity has been "check the Linux box, click the build button". It even cross-compiles no problem. People I know tell me Unreal is similar. It's not like they have to do a full run of play-te…

> What are the main costs of releasing a Linux version of a game when using Unity or Unreal? My limited experience with Unity has been "check the Linux box, click the build button". It even cross-compiles no problem. People I know tell me Unreal is similar. For a simple game that uses the entire UE4 stack, you might be able to get away with that if none of your code is Windows-specific and works exactly the same on t…

Probably the most relevant issue here is the DirectX stack or using some windows specific API (why? DirectX I maybe got at one point, the others speak to some absurd Microsoft-Is-Best-Cause-I-Paid-Money-ism).

> Many won't have a Linux compatible version, and those that do may only work against specific distributions and hardware. Even then: Have they changed the window manager? What have else they customised?

Outside of truly exotic window managers, the biggest issue I've ever seen with ports has to do with dependencies. Flatpack/Snaps solve most if not all of those problems.

> There have been many issues with anti-cheat solutions over the years on Linux.

That's assuming anti-cheat systems are ethical, even. Most games I play I want to be multiplayer, and moddable. Most games I play can do both of those things and work in a cross platform modding language. Not hard. I don't want any anti-cheat systems for a complex game with tons of modding.

The only place anti-cheat systems make sense is with e-sports style games. Of which, the best anti-cheat systems work server side. Anything else is invasive and tantamount to illegal spying. Shame on you, go burn in peeping Tom hell you perverse game dev.

> For a simple game, you may get away with this. Anything larger will require significantly more testing.

I find this incredibly hard to believe. If you are working within the engine the majority of the time, the majority of the game will play wherever the engine works. Which brings us back to one of your first comments:

> For a simple game that uses the entire UE4 stack, you might be able to get away with that

For which games are you not just using the UE4 stack? I.e. what specific problems are game devs normally experiencing causing their code to be non-portable? A bunch of low level assembly code and non standard C++?

> but as soon as you start digging down into serious development

I challenge you to define 'serious' development. That sounds suspiciously like you're wasting your time fighting infrastructure, or just doing things in a non-portable manner. Solving problems is cross platform, so you can't be doing much serious game coding, its probably problem solving for your particular platform (e.g., as I already mentioned, doing a bunch of work in DirectX shaders or something).