No announcement yet.

John Carmack Shares More Of His Linux Views

  • Filter
  • Time
  • Show
Clear All
new posts

  • John Carmack Shares More Of His Linux Views

    Phoronix: John Carmack Shares More Of His Linux Views

    John Carmack, the co-founder of id Software that's widely known through gaming circles due to his remarkable work on developing Doom and Quake and other titles, sparked some controversy earlier this week when he promoted Wine for Linux gaming over native Linux game ports. He's now provided some additional clarification and thoughts...

  • #2
    What's all the hassle? M$ market share is already down to ~70% in Switzerland and continues to fall...


    • #3
      Originally posted by BO$$ View Post
      Microsoft is so last decade. They are dead. Most devs should focus on building on linux first and then maybe porting to windows. Hell even Microsoft thinks of porting office to linux...
      Not because Linux would yet be the more lucrative market, but simply because designing for linux equals to following best coding practice guidelines.

      Also, porting is unnecessary. Cross platform compatibility comes almost for free if you design the program wisely from the get-go. This is why i tend to agree with carmack on this issue. Use WINE for current titles, design next-gen titles to be crossplatform - forget porting old shit that was designed in a darker age.
      Last edited by varikonniemi; 02-06-2013, 11:53 AM.


      • #4
        Originally posted by varikonniemi View Post
        Not because Linux would yet be the more lucrative market, but simply because designing for linux equals to following best coding practice guidelines.
        If and whwn Linux gets high uality modern APIs and tools,maybe. it's currently way easier and less frustrating to get a modern renderer up snd running using Microsoft's APIs and tools. As a developer with a budget and timeline, that matters. The only thing Linux has which Windows is missing is Valgrind. Meanwhile, Linux is completely lacking in high quality GPU debugging tool, and OpenGL/OpenAL are simple awful APIs. OpenGL has an excuse in that it is ancient; OpenAL has none, it was just designed by morons who copied OpenGL. Which is why so few real games use OpenAl directly and pay for fmod or wwyse, and why game devs prefer D3D9/11 over GL. Better APIs save time and money, as do better tools.

        With OpenGL, if your renderer doesn't work, good luck figuring out why. Way easier to get everything working on D3D/Windows and port after you know your game is corect and works.


        • #5
          Valve already demonstrated how hopelessly poor windows/d3d is as a 1st class gaming platform when source runs much faster on Linux/ogl than it does after years of optimizing for win/d3d. You cannot have the cake and eat it too. Choose either simplicity or quality.
          Last edited by varikonniemi; 02-06-2013, 12:18 PM.


          • #6
            - Specifically about Direct3D translating: "Translating from D3D to OpenGL would involve more inefficiencies, but figuring out exactly what the difficulties are and making some form of “D3D interop” extension for OpenGL to smooth it out is a lot easier than making dozens of completely refactored, high performance native ports."
            I'm fairly sure valve is already using some sort of d3d > opengl translation layer for their ports (I recall reading that somewhere)


            • #7
              This is clearly Carmack responding to the great strides made over at Valve in the Linux user space. Wine is great for running Old games that have no chance of a port. It is a really neat piece of software but its not a optimal solution. Even Blizzard is thinking of releasing a Linux native title. I have played three WoW expansions, StarCraft 2, and Diablo III. All play alright but you can really tell the difference if you play them native. Believe me, I got them working but it was hard to do at times.


              • #8
                I think most of the AAA players will be the LAST ones to adopt Linux, while the indie guys, and Valve, who still has the indie spirit, will be the first to adopt it. With OpenGL ES 3.0 games coming to both hundreds of millions of iOS and Android devices, it should be easy to port the same games to Linux, too, especially with more engines like Unity supporting it. I hope Unreal 4 supports it, too. Last I checked they were undecided. But I hope they will.

                It's still disappointing Carmack is not on the "indie" side anymore, but what can you do?


                • #9

                  A good shim layer should have far less impact on performance than the variability in driver quality.
                  This sentence doesn't make any sense. Doing a shim layer won't make disappear the "variability" in driver quality.
                  You just throw back the problem to the Wine developers...


                  • #10
                    Originally posted by spykes View Post
                    This sentence doesn't make any sense. Doing a shim layer won't make disappear the "variability" in driver quality.
                    You just throw back the problem to the Wine developers...
                    I think he means the shim is a less important factor on the loss of performance than some drivers that don't have high performance, but both will affect it(in other words there are issues regarding performance with higher priority than the ones coming from the shim).
                    Besides, at least from my understanding, what wine does is that when the wine dll is called, it calls a linux native function instead of a windows native function, at worst you get an extra function call in between. Whether the linux native functions or code within the wine dlls are optimized is of course another matter, not connected with inherent issues of the implementation.


                    • #11
                      If Gaming companies made games that worked flawlessly with wine with an easy deployment (that installs their wine hack separately if necessary) and if they guaranteed patch and corrections and a good framerate and immediate releases.. then I wouldn't mind if the game is native or "Wined" ... and would pay for a good gaming experience.
                      But I wouldn't pay if I had to hack my wine install to play ...

                      But as already mentioned before : Opengl is already used in games ported to Mac, IOS.. So it shouldn't be so difficult to port the games to Linux (not to mention that lot's of games use "3d engines" that are for some already available on Linux) ...
                      Id software did create a version of rage on Ipad and ported rage on the Mac... So they thought it was good for them to port games on other platform than win pc... for Linux they perhaps don't think it is worth the costs .. that is their view after their previous trials and "errors".
                      For games and apps It has always been a problem of market size and distribution ... if you don't provide a sufficient market, they won't make apps or games .. if you don't provide apps and games, not enough users will come to make a potential market ... it will be a niche market.
                      Even if Carmack made commercial games available on Linux years ago before the other guys, it was not "as easy as steam or Ubuntu store", for the gamers to find or install them .. And it was a different time with low bandwidth, less 'inline' payments and less Linux users ...
                      Today with Steam, games are easy to find, buy and install (but there are just a very small number of games compared to windows and Mac markets)..
                      Ubuntu software center should have helped game companies to port games long ago (with a "kickstarter like page" with "pre-purchase" in their store to port games or "Wine enabled them" for example - I understand that companies wouldn't want to do it for free)...
                      According to steam Stats : "Ubuntu" has now 1.12 % usage of the steam platform during January 2013 (+ other distros ('other' = unidentified platforms in steam) have a part of 0.76% of steam usage) ..
                      That is between +1/4 and -1/2 (if you count all the "other") of the mac usage (4.41%) on steam.. That means if not everyone, a large portion of Linux users wants to play on Linux... and Gamers are ready to pay for it. And regarding the "market share" attributed to Linux VS Mac it is in proportion a LOT of people ...

                      If you care and give a good product for the users and easy access to the product, they will probably come ... Steam decided not only to port games but to "make/build the market" ... Will they succeed ? Is it a problem if the market is not big enough ? At least they have tried and they have given the linux gamers "a tool" to make their voice and money count ... If after that companies are able to make money they will surelly come like they came to embrace the MAC and IOS market...

                      So "wined" or "not wined" is probably not relevant - easy access, easy deployment and good user experience are the only things that matters for games and for apps (in fact the goal of any OS should be that everything is totally transparent and works as designed) ...

                      So let's think about "the future of gaming on Linux platform" ... If a steam box comes out with Linux system, what would the best option to sell it ? If I were Valve I would sell it "bundled with games" for a limited period ex: "A steam box with the whole 'half life' package (Half Life 3 included) [Half Life 3 won't be a real exclusivity but the box would be the first platform to "deliver the game" .. That would make gamers consider the box or Linux as a viable gaming option.. but that is just a "wish"


                      • #12
                        Just read this:


                        It seems Valve will be making the "business case" for Linux, and for Carmack. If they sell millions of these Linux-based Steam boxes, then Carmack will come back to porting games on Linux before you know it.


                        • #13
                          That was expected. Id is dead and he is getting old. But the funny part is:

                          Back in days, he wanted to write native version of Mario on PC, but Nintendo decided Mario belongs only on its consoles.
                          Today, he is asked to write native versions of his titles on Linux, but he decides his windows-only code is better run emulated/wrapped.


                          • #14
                            Grandia 2 the PC port (Win 98) is going to be my dead horse here. Windows XP couldn't even run it! Betcha anything if they released the code someone would port it in days. I could give a flying fuck about most new games I just wish games that no longer sold had their source released so the world would benefit. Still, yeah Wine for old titles... like we have a choice, what logic. Someone posted it sooner write cross platform from the get go.


                            • #15
                              Originally posted by elanthis View Post
                              If and whwn Linux gets high uality modern APIs
                              don't tell me you call windows apis anything near of high quality and modern!

                              .net in its latest version is the first api wrapper from microsoft that SOME WHAT managed to provide something half way usefull.

                              i've coded for over 20 years, most of my paid stuff for windows and it is ALWAYS a BIG relief when i don't have to code anything involving micorsoft api's. biggest bullshit ever designed!

                              but sure, when people grow up and get educated in colleges and universities only with such stupid things they got used to and like most people think the things they already know are the best.

                              if i weren't payed so well for coding for the windows plattform i wouldn't evet do it again.