John Carmack Pushes Wine For Linux Gaming

[entropy]

Carmack is an overrated developer.

This.

Carmack farts the web is filled with articles reporting it? He has become irrelevant in gaming for quite some time... Try to remember the games he produced in the last 5 years and tell me how good they were, technology or gameplay-wise...

Hysteresis effect of what he did in the past. Might never change.

Similar to AMD - they will probably never get kudos for their binary drivers.
No matter how hard they try. It's just the other way around.

[newwen]

Wine is only good enough for older software that would never get ported to Linux. But for new software, designing and writing code with platform independence in mind from start is a better software engineering practice.

This said, if Valve wants to be successful in their Linux push, they should integrate Wine into Steam to run older games that won't get ported. I have Windows games in my Steam library that I have to run using the windows version of Steam on wine.
They should test popular windows only games against a specific version of Wine included with the Linux version of Steam and enable running windows games that are proven to work good enough (gold) with a wineprefix that works for that specific game. That's a ton of work though and only an interim solution.

[TemplarGR]

Carmack is an overrated developer. I remember having arguments with other CS students about this. While most worshipped him because they played Quake and Doom, i never did. Since i am now a professional developer myself, i don't consider anyone that good and deserving of such high praise.

His engines weren't even the best of their times. They were good, sure, but others were better. And the games he created were pretty boring actually. I viewed them more like tech-demos than worth-playing games.

So i don't get what the fuss is about. Why every time Carmack farts the web is filled with articles reporting it? He has become irrelevant in gaming for quite some time... Try to remember the games he produced in the last 5 years and tell me how good they were, technology or gameplay-wise...

[Mike Frett]

Goodness

This guy has clearly lost his Marbles. Wasn't he raging before about how Linux was a failure and how he hated to write his games natively for Linux?. I respect Mr. Carmack for the things he has done in the past, but my opinion of him has been quickly changing as of late.

[moilami]

ID what?

Anyway, they don't want my money for their new games, but other companies will gladly accept my money and deliver games to me.

Times are good. There are much more games for GNU/Linux than I have time to play.

Besides I haven't even finished Quake I with both mission packs on Dark Places engine, a game I have the fancy collector's edition box. Was nice to support ID, but now it is Full Steam ahead

[madbiologist]

Carmack actually said "emulator"? I thought WINE stood for Wine Is Not an Emulator?

[losinggeneration]

No, I did not know that. Now I understand why the Doom 3 engine source is a complete mess.

Well, I feel I should clarify a bit. He's not a CS graduate, but he has been a professional game developer since the late 80's if I'm remembering correctly. That longevity goes a long ways.

With regards to the id Tech 4, the engine was a bit messy because it was their first mostly C++ project. My naive wc on doom3 & ioquake3 gives 578515¹ & 355606² respectively. I believe John Carmack has actually commented on this and expressed that the id Tech 5 engine is actually in better shape. That said, I don't think the engine is a mess, from an architecture standpoint. There are things that could have been done better for sure, but that could be said about pretty much every piece of software.

1. find . -iname "*.cpp" -or -iname "*.h" | grep -Ev "neo/(curl|openal|sound/OggVorbis|renderer/jpeg-6|tools)" | xargs wc -l
2. find . -iname "*.c" -or -iname "*.h" | grep -Ev "code/(AL|SDL|zlib|jpeg|libcurl)"| xargs wc -l

[kokoko3k]

D3D is a shader compiler, openGL the same, CG the same. It compiles programs(shaders) to a form that a computer(GPU) can understand. Its useful because there are many different computers and you cant write code just for one. The target and optimized libraries that are required to target a specific hardware are inside a GPU driver (the same is true for OpenGL compilers). The bad thing is that wile with a new CPU they give as those libraries(BDver for example) for GCC, for a GPU they don't give the analogous MESA parts. That is happening because they don't want MESA to grow, because then other companies will come and break their monopoly (software is half the GPU). Those two evil companies (ATI_AMD and NVIDIA) instead of using only OpenGL, they co develop with Microsoft DirectX, the closed all games under Windows and they are privileged by this deal (to be the only ones for long time). Also they did attack open pc closing games under consoles, that's another crime and not for money (the profits are better if you sell more cards and more expensive cards for pc). Now as for Wine the situation is simple: Except from the compilers there is also the rasterizer inside a GPU driver. Even if we can run D3D libraries on Wine we don't have the D3D rasterizer with the Linux GPU drivers, so we need to translate and loss FPS. Some times with an Nvidia card only we can set this translation off (GLSL=disabled), that uses the old and not efficient compiler and gains some FPS. In order to solve those problems Wine started an llvm_hlsl_shader_compiler, that uses the efficient llvm to compile HLSL very close to GLSL efficiently so the translation becomes very fast. When they succeed we will no longer need D3D libraries ether.

Oh, thank you very much for the detailed explaination.
D'you have any source where i can follow the wine llvm_hlsl_shader_compiler development?

[ElderSnake]

I don't know what's up with Carmack lately. IMO what he's suggesting is actually kind of dangerous (if, like me, you want to push for multi-platform tech and lessen this reliance on Windows only tech)

WINE is great, for what it is. But it should never be the answer for new titles. For old existing games or games which just happen to be Windows only/exclusive, WINE can be a God-send for the Linux gamer if it happens to work okay.

I'm really loving what Valve is doing with Linux and I hope it does so well it makes JC (unfortunate initials) very very jealous.

[kaczu]

And I do not disagree with you, for the most part. I just think some of the language and allegations being thrown about in this thread are unfair.

I think they are actually kind of warranted. From a business perspective it makes no sense to target wine when 2/3 of the consoles don't and 97% of the mobile devices out there don't either. Right now there's a war going on and it's not about operating systems, but about API's. MS is losing this war in a big way because it's mobile marketshare is horrendous and no sane developer wants to limit the platforms they can sell their products on. Meanwhile gaming on mobile units is exploding. MS knows this and so does Carmack. OpenGL and it's offshoots are eating MS alive. So why target Direct X just so that when you go to port your game to a mobile unit you have to spend that much more to make it possible?

I also don't buy the source argument. Yes, source is very important. But it's only as important as the technology it relies on. Most of the gaming engines being produced / developed now support output to OpenGL/ES. This isn't by accident.

What SteamBox, Ouya, Android, iOS are doing is providing a platform for developers to sell their content without the MS lock in. Windows 8's store by itself isn't what is pissing off developers. It's the API lock in with Metro.

What Carmack is saying that it's harder to target an executable for Linux (which he has admitted to it only taking a few months) than to chase wine versions + their platforms over time. I'm not seeing where it makes a whole lot of sense.