John Carmack on Linux ports [Archive]

Odin

07-21-2009, 02:12 PM

Wolfenstein is pretty unlikely, since it was developed at Raven, and published by Activision. There are no firm plans for linux ports of the idTech 5 titles, but it certainly isn’t off the table. I don’t think it will be very difficult to get them running on the binary nvidia drivers, but bringing them up to functionality and acceptable performance on other OpenGL drivers would probably be a more significant undertaking than we could afford.

http://www.linuxgames.com/archives/14005

simonau

07-22-2009, 10:48 AM

This seems very short sighted to me

Kano

07-22-2009, 11:01 AM

It seems that he does not like fglrx ;)

Ex-Cyber

07-22-2009, 01:50 PM

I wonder if anyone from AMD/ATI has already contacted him about this. Apart from the actual issue of running the game, it would be a pretty big PR fail if Rage got released for Linux and only ran on Nvidia hardware...

geamandura

07-25-2009, 01:38 PM

What a bunch of crap. The real reason is laziness. Id have 3 internal dev teams now and cannot afford a guy to port it? Come on, the engine was made with super portability in mind.

What pisses me off is that the devs' attitude is that they're doing us a favour and they think in terms of, can they afford the money to do us this favour of porting the game to Linux. When in reality they gain some sales and good image that translates to the future sales.

You could even go as far as say that important studios such as Id, Blizzard etc. are the market shapers. If Starcraft 2 suddenly decided to come out just on Linux, does ActiBlizz really imagine they'll lose sales? Of course not, Linux is free and every possible customer will simply do the shift from Windows. Their sales will stay the same, while they will have shaped the market with not much effort.

I can't stand hearing the big devs expressing their worries about the market trends when it is exactly them the ones who shape the market.

Also, until now I regarded Carmack as a super guru and a really good guy but it seems he just lost interest, and just goes where the wave takes him.

mirv

07-25-2009, 02:04 PM

I think perhaps carmack might have meant instead that he's familiar with the nvidia drivers. He would be likely to use certain extensions that, whilst not part of the official opengl spec, might be common enough (for example, occlusion_query before it was more standardised was an nv extension) to use. If the extension wasn't available, fallback solutions would have to be used, and that might have some bad performance penalties.
On the other hand, he could just be lazy - but given his track record, his words might be given slightly out of context. If there's a more detailed interview, it would be worth reading it (not there on the page linked).

-- btw, using a method, and being dependent upon it, only available from one company is a very, very bad idea anyway.

L33F3R

07-25-2009, 02:34 PM

the mighty do fall. please keep that in mind.

Dragonlord

07-25-2009, 03:52 PM

Not a problem... big ones will fall... new ones will grow. Just support the right guys folks, then victory is ours ;)

Yomp!!

07-25-2009, 06:48 PM

Yep, theres alot of small game development companies/indie games that do actually support Linux in full and look to have a very promising future. Whether or not we get another Linux port of any game from ID in the future ever again doesn't really matter, I'd like to see one as much as anyone else but lets face it, it doesn't look very promising right now.

Also I'm no programmer or OpenGL guru but if the ID games can run on Windows and MacOSx in OpenGL on ATI and nVidia hardware, shouldn't it be no problem to support them on Linux in a very similar way? It just looks like a load of crap hes saying:

"I don’t think it will be very difficult to get them running on the binary nvidia drivers, but bringing them up to functionality and acceptable performance on other OpenGL drivers would probably be a more significant undertaking than we could afford."

As far as the OpenGL calls and extensions and stuff, I would just assume it would be very similar or the same as Windows (especially MacOSX which ID does support), and the extensions on both drivers for ATI and nVidia are practically identical on Windows and Linux. What I mean by that is, ATI on Windows and Linux have similar extensions as well as nVidia, however ATI and nVidia have some different extensions from eachother when you compare the two in general. I guess those would be called vendor specific extensions, which no game can be built around else it would break compatibility and screw the customer.

I don't know, all I know is its sad to see them sell out like this but as you guys have said, big ones fall, new ones grow. I would just figure though that Carmack would come up with something better than that to say, why not just say: "ZeniMax doesn't want us porting to Linux as well as our publisher EA.", the end.

L33F3R

07-25-2009, 06:49 PM

Yomp!!

07-25-2009, 07:01 PM

our scene? thats hard. few people give a bad rap to the many. All linux users dont buy software says leadwerks. We are open source zealot fags says the steam community.

The real question is, how do you plant that seed?

------

Carmack is a big deal, he has contributed alot. Indies are essential but they are not movers and shakers. What we need is a real big project. It needs to be very very linux and it needs to be opengl 3.

Of course Carmack has done great things for us, I am thankful for that.. I just don't understand why hes saying the things he is, I understand him not being able to support Intel graphics on Linux for various reasons, its hard to support them on Windows too with a modern game, but saying nVidia is the only way for him to get it to work makes absolutely no sense.

And as far as Indies and the big project, well, the big game has to come from a company like a Indie developer or a small game development company, no one else wants to do it, or wants to take the chance. A company can single handily make a name for Linux in the gaming world if they bring the right product that shines enough and brings enough interest, however, I don't know if you'll ever find a company or group in this day and age that would make a Linux exclusive game for the purpose of trying to spark a revolution. It just doesn't sound very realistic to me. Multiplatform sounds realistic though.

L33F3R

07-25-2009, 07:11 PM

Of course Carmack has done great things for us, I am thankful for that.. I just don't understand why hes saying the things he is, I understand him not being able to support Intel graphics on Linux for various reasons, its hard to support them on Windows too with a modern game, but saying nVidia is the only way for him to get it to work makes absolutely no sense.

And as far as Indies and the big project, well, the big game has to come from a company like a Indie developer or a small game development company, no one else wants to do it, or wants to take the chance. A company can single handily make a name for Linux in the gaming world if they bring the right product that shines enough and brings enough interest, however, I don't know if you'll ever find a company or group in this day and age that would make a Linux exclusive game for the purpose of trying to spark a revolution. It just doesn't sound very realistic to me. Multiplatform sounds realistic though.

his brain is getting old. Hes making dumb decisions. ATI works great on ETQW for example.

If indies moved as much as you say then we would be golden. This is not the case. World of goo for example was one of the games you speak of, but linux was an afterthought. Multiplatform is the only linux option, you are right. You can find a group of programmers and artists who can do a smashing job, you just need to give them the right incentives.

Yomp!!

07-25-2009, 07:25 PM

his brain is getting old. Hes making dumb decisions. ATI works great on ETQW for example.

If indies moved as much as you say then we would be golden. This is not the case. World of goo for example was one of the games you speak of, but linux was an afterthought. Multiplatform is the only linux option, you are right. You can find a group of programmers and artists who can do a smashing job, you just need to give them the right incentives.

Yep I agree, the game can't be late, World of Goo came pretty late, to the point where no one really cared, although from what I could tell it sold ok on Linux, I don't think it sold as well as MacOSx though. It could had sold better though, most people hate buying the same game twice just to play it on their preferred OS, but it came out late so they just bought the Windows version, played through the game, and left it at that.

As far as future games sparking any excitement on Linux, if the development team is very pro Linux and wants to help it out, they can release the Linux version a day or two earlier than the Windows and Mac versions, that would maybe put some incentive into the customer to say, hey, maybe I should try Linux! Or if they already use Linux in a dual boot scenario and aren't accustomed to purchasing Linux games, that might put in some incentive to do so as well.

The game has to sell very well too on Linux though for any change to take place. We gotta show the dollar signs to all of those big gaming companies out there. Because honestly I think there are enough people using Linux already right now, there just needs to be a game that interests most of the userbase, something highly anticipated and highly spoken of around the community, word travels fast, theres enough media outlets such as Phoronix. We just need the damn game ;) hehe.

Dragonlord

07-25-2009, 07:39 PM

You have to change the way the game is played. If you work with the aged philosophy of porting games instead of working with a truly platform independent solution things will never change. This is simply because if you have to port a game there will always be a major platform it is released for first and minor platforms it's ported too. In that game Linux has no chance. It only has a chance if a game is made once and 0-day playable on all platforms. This is what a revolution looks like. And this is not coming from AAAs since they are still clutching to their engine licensing business forgetting that the world is changing around them.

But for getting such a project off the ground it lacks on one thing: artists. They are nearly none around which are worth and dedicated to make a great game and without the content the tech doesn't help much.

Kano

07-25-2009, 07:48 PM

Well maybe Nvidia's It is meant to be played program would help him in case of problems and would fix the driver if needed. Nvidia does much more for games developers than ATI, they even help em to use Physx. ATI should really care more about popular game developers.

xav1r

07-25-2009, 08:16 PM

Dragonlord

07-25-2009, 08:53 PM

Man, you should work in the game industry, or better, you should be the guy calling the shots at a big game developer. Why not start your own?? :D
I'm not saying more here as those who know me know what I do :D

L33F3R

07-25-2009, 09:59 PM

But for getting such a project off the ground it lacks on one thing: artists. They are nearly none around which are worth and dedicated to make a great game and without the content the tech doesn't help much.

If thats truly the biggest issue its not that difficult to fix. Start small but start big. Look at a genre thats way overlooked, remember super smash bros. 2.5D punch'n'!@#$. Now take art elements that any idiot could accomplish. Make it feature chars like Tux, Penny, snowball, blue screen of death, ect. Target a middle-low polygon platform to maximize compatibility and run on say, pandora. Now instead of modeling your maps which should be done, because your stuff is low poly use csg. A dog can make a map using csg. Set the players to run on x axis and make them map w.e they damn well want. For textures, search web development forums and say they get credits. Those people are looking to get der names on everything, im serious. models are relative to the polygon count, you can make a model in any program and it will export to any other app. Now you have a low price game that will evolve with its user base. It would be easy enough to make levels (i remade the zelda map from SSB in 2 minutes) so having a storm of crap like we see on fps banana would not be very hard. Obviously I am being very general and random but you can abstract the idea. You dont need to make a AAA game for it to be a seller. If people want to ditch halfway through the project tell them to f-off and use what they made anyways. If der shit sucks then tell to do just that.

When you have a track record, then you can attract the best. Such an idea like the above one I just came up with does not require a large pool of art talent. Im not saying to make crap, everything takes more than 2 minutes obviously. Please expand :D.

xav1r

07-26-2009, 12:01 AM

If thats truly the biggest issue its not that difficult to fix. Start small but start big. Look at a genre thats way overlooked, remember super smash bros. 2.5D punch'n'!@#$. Now take art elements that any idiot could accomplish. Make it feature chars like Tux, Penny, snowball, blue screen of death, ect. Target a middle-low polygon platform to maximize compatibility and run on say, pandora. Now instead of modeling your maps which should be done, because your stuff is low poly use csg. A dog can make a map using csg. Set the players to run on x axis and make them map w.e they damn well want. For textures, search web development forums and say they get credits. Those people are looking to get der names on everything, im serious. models are relative to the polygon count, you can make a model in any program and it will export to any other app. Now you have a low price game that will evolve with its user base. It would be easy enough to make levels (i remade the zelda map from SSB in 2 minutes) so having a storm of crap like we see on fps banana would not be very hard. Obviously I am being very general and random but you can abstract the idea. You dont need to make a AAA game for it to be a seller. If people want to ditch halfway through the project tell them to f-off and use what they made anyways. If der shit sucks then tell to do just that.

When you have a track record, then you can attract the best. Such an idea like the above one I just came up with does not require a large pool of art talent. Im not saying to make crap, everything takes more than 2 minutes obviously. Please expand :D.

You make it all sound so easy. Is it really that easy?

L33F3R

07-26-2009, 12:19 AM

no

--------

geamandura

07-26-2009, 04:32 AM

Now take art elements that any idiot could accomplish. Make it feature chars like Tux, Penny, snowball, blue screen of death, ect. Target a middle-low polygon platform to maximize compatibility and run on say, pandora. Now instead of modeling your maps which should be done, because your stuff is low poly use csg. A dog can make a map using csg. Set the players to run on x axis and make them map w.e they damn well want. For textures, search web development forums and say they get credits. Those people are looking to get der names on everything, im serious.

Come on, man, I wouldn't even bother downloading something done as you desctribed it. Hell, it would even make the Linux gaming status quo worse to have a crap like that thrown together and advertised. People still play games using Wine, there's still stuff like Nexuiz, FreeCiv, Wesnoth... I mean there's already a standard in effect here. It's not Blizzard but it is something already.

If you're talking about minimizing effort but keeping it look good, then the solution is a minimalistic art design, something done by an art guy who knows his stuff real good, but has very little elements instead of an UT 3-like heavy art design. E.g. that Dyson game or the games from Puppy Games (www.puppygames.net). No compromises.

L33F3R

07-26-2009, 08:45 AM

i think you greatly perverted what im saying. If you want to see crap look at all the FOSS games in synaptic. That alone proves you have no idea what your talking about. Start calling some good non FPS titles, you will be fairly short on a decent number of them that are 3D. That idea took 30 seconds to come up with, its hardly anything special but it demonstrates how the artwork hurdle might be accomplished, not will be.

BTW, IMO Puppy Games are boring as hell after 5 minutes. Great ideas and good looks but perhaps something that doesn't make me want to click the X on the window.

geamandura

07-26-2009, 11:57 AM

To be clearer, I wasn't talking about all the Linux games having good standards. I was trying to refer to raising the bar on Linux games. If there's a recipe for art that can effortlessly be put together by somebody that doesn't have true art talent... better not bother.

L33F3R

07-26-2009, 12:42 PM

oh hell I dont want to see an effortless product and neither does anyone else. I push when i say how easy it would be, its not. It would take a long time for everyone and the core team would need to be very handy. My point was to show how you can grab a team. Dragonlord can attest to how many hammer users are on moddb. Same goes for generic modelers. Some of them, might actually even be good :D. Photoshitters are everywhere and sound peoplez are less difficult then one might expect. In my example i showed how you would be able to adapt the title to the artists instead of the other way around. One such ways is using a very generic XML to be the geometry allowing the developers and as an afterthought, the people who play it, to use whatever program they want. Artists dont want to learn another program, we are a stubborn bunch.

yogi_berra

07-26-2009, 02:30 PM

but saying nVidia is the only way for him to get it to work makes absolutely no sense.

It makes perfect sense or have you forgotten how the Doom 3 release went?

The game ran fine on nVidia hardware but crashed to desktop on ATi in Windows and Linux. ATi, now AMD, has never been good on the bleeding edge rendering technology that Carmack likes to program.

Yomp!!

07-26-2009, 03:13 PM

geamandura

07-26-2009, 04:16 PM

One such ways is using a very generic XML to be the geometry allowing the developers and as an afterthought, the people who play it, to use whatever program they want.

Collada? :)

Dragonlord

07-26-2009, 04:47 PM

@L33F3R:
Wrong direction to mount the horse. Down-grading a project to the people just puts emphasis on Linux sucking for real games. There's only one way: do it right. Now this doesn't mean to be the biggest graphic miracle but without a certain degree of quality the cause is lost. And as much as I do know hammerers on ModDB as much do I know lazy bitches. Most of the people don't have the will to stand such a project through. That's the sad truth.

@yogi_berra:
Which is bull. I'm deving on ATI stuff on bleeding edge tech. That's not what is the problem. A crashing app is simply one which doesn't do things properly. Granted ATI had some crash-worthy parts in their drivers but so does nVidia. If it crashes it is first a problem in the game not doing their pointer arithmetic correct not the graphic card doing something wrong.

@L33F3R:
XML is good for only one thing: definitions. Never use it for large data-objects like models, terrains, images or whatever. It's best use is for using definition work and for higher level data ( for example maps but only defining entities no geometry/brushes or other binary stuff ).

L33F3R

07-26-2009, 05:12 PM

Down-grading a project to the people just puts emphasis on Linux sucking for real games. There's only one way: do it right. Now this doesn't mean to be the biggest graphic miracle but without a certain degree of quality the cause is lost. And as much as I do know hammerers on ModDB as much do I know lazy bitches. Most of the people don't have the will to stand such a project through. That's the sad truth.

Yes and no. Yes to the lazy asses on moddb :p

I do not believe it would downgrade the quality of assets or the project as a whole. it would encourage shit work. Whether or not shit work is acceptable is up to standard. You dont hire someone simply because he/she is available you do it because they are qualified. Just because more people apply for a job doesn't mean that the rate of crap goes up. You pick your apples. As for a smaller weight it makes alot of sense. OGL 1.4 is good for mobile devices and netbooks. Make it run unexpectedly well on older hardware and more users will get on. This is 1 reason why WoW is such a success. That, and its like crack...

XML is good for only one thing: definitions. Never use it for large data-objects like models, terrains, images or whatever. It's best use is for using definition work and for higher level data ( for example maps but only defining entities no geometry/brushes or other binary stuff ).
Yes but it can be used for the others and if my irrlicht fps counter doesnt lie it works very well. Now i wouldn't make a large terrain level with it. That school project, what was it, urban warfare online. They used it aswell and besides the shitty collision it worked wonderful under wine. One might opt for a binary format but you would be chopping your options. That said, artists dont particularly like people poking around the assets so it has ups and downs.

Dragonlord

07-26-2009, 05:50 PM

I do not believe it would downgrade the quality of assets or the project as a whole. it would encourage shit work. Whether or not shit work is acceptable is up to standard. You dont hire someone simply because he/she is available you do it because they are qualified. Just because more people apply for a job doesn't mean that the rate of crap goes up. You pick your apples. As for a smaller weight it makes alot of sense. OGL 1.4 is good for mobile devices and netbooks. Make it run unexpectedly well on older hardware and more users will get on. This is 1 reason why WoW is such a success. That, and its like crack...[quote]
WoW is only a success because it's based on a well known franchise ranging back to the original WC. If it would have been new stuff without basing of their old franchise it would most probably not have had such a success. That's what I call "IP Whoring" or "Franchise Whoring". Base your project on an existing IP/Franchise and you'll have lemmings... erm... fans... in no time. Do something original and creative and you'll get only a few people.

[quote]Yes but it can be used for the others and if my irrlicht fps counter doesnt lie it works very well. Now i wouldn't make a large terrain level with it. That school project, what was it, urban warfare online. They used it aswell and besides the shitty collision it worked wonderful under wine. One might opt for a binary format but you would be chopping your options. That said, artists dont particularly like people poking around the assets so it has ups and downs.
XML does work as a format, that's not the problem. The problems are
1) slow to load large files
2) very large files compared to equal binaries
I use those too for testing formats and engine features but in the end for a final product you need binaries. Not a problem with my engine though since in the IGDE you will be able to mass-convert files when doing a deploy build ( since every resource module in the engine is required to know loading and saving ).

deanjo

07-26-2009, 11:07 PM

All and all, not a surprise at all. Carmack basically just confirmed what I said months ago. http://www.phoronix.com/forums/showthread.php?t=16332

Yomp!!

07-26-2009, 11:18 PM

All and all, not a surprise at all. Carmack basically just confirmed what I said months ago. http://www.phoronix.com/forums/showthread.php?t=16332

Yep, and the recent ZeniMax deal I think put this one to bed.

Svartalf

07-27-2009, 12:29 PM

It makes perfect sense or have you forgotten how the Doom 3 release went?

The game ran fine on nVidia hardware but crashed to desktop on ATi in Windows and Linux. ATi, now AMD, has never been good on the bleeding edge rendering technology that Carmack likes to program.

That's less due to the silicon (which is as good or better than NVidia's...) and more due to the drivers. Their codebase has been historically unstable and finicky. AMD's GLSL/HLSL compiler has always been quite a bit more picky on what was fed to it, for example. When it was done "right" (which is typically a slavish insistence on explicit compliance with the OpenGL and DirectX standards documents...) it works WELL and no issues. Otherwise...heh...

As for crashing, etc. I can't say that I ever noticed that with my dual screen setup I had during that era when I worked for Nexa Technologies (stock trading software company...)- my setup before it cooked itself (poor heatsink on the Dell OEM ATI adapter in it...) ran WELL with Doom3 under Linux, albeit strung across BOTH my screens. I ended up getting an NVidia part because it was actually cheaper than the comparable ATI part because of a sale that week.

Having said this, the drivers really aren't where they need to be. They haven't been there for a while (When I can't switch users without being black-screened with an R600 part, but before the drop of R300-R500 support, the R420 I had COULD do it right... Something that is still with the current fglrx drivers... It's not "there"- what else did they miss/break?) and they need to do SOMETHING to get them back on track. I'm not wholly sure what they need to do, but...

Perhaps Larrabee will be as good as the reports are. It's alleged that Intel will FOSS the stuff they're making for the Linux side on it- if so, it'll be a second answer and maybe present a more appealing situation for iD. As it stands now, I can't fault them for saying they're not going out of their way for us, but it's not a "no" on Rage at this point.

Svartalf

07-27-2009, 12:35 PM

All and all, not a surprise at all. Carmack basically just confirmed what I said months ago. http://www.phoronix.com/forums/showthread.php?t=16332

Heh... Before this, there wasn't any real reason to think that you'd be right, deanjo- they'd not made any statements along the lines of what John just said there. YOU didn't have anything other than a gut level feel to go on- something you've not been 100% good on any more than I.

Moreover, I think this is a combination of our 3D story not being there right at the moment, coupled with their recent merger. If there's no money to be had (hey, you can get the rubes to buy the Windows version and run it in WINE, so why do the extra 10% for that platform, especially when the adapters aren't there...) in the parent's estimation (whether it's wrong or not matters little- they're the ones in control...) then why go to that extra effort?

yogi_berra

07-27-2009, 12:44 PM

@yogi_berra:
Which is bull. I'm deving on ATI stuff on bleeding edge tech. That's not what is the problem. A crashing app is simply one which doesn't do things properly. Granted ATI had some crash-worthy parts in their drivers but so does nVidia. If it crashes it is first a problem in the game not doing their pointer arithmetic correct not the graphic card doing something wrong.

I highly doubt you are pushing rendering technology the way Carmack does, but whatever helps you sleep at night.

btw - if you want a definitive answer as to whether a game will be ported, ask TTimo, not Carmack. Quake 4 was a raven game and we got the native client in the same week.

deanjo

07-27-2009, 12:55 PM

Heh... Before this, there wasn't any real reason to think that you'd be right, deanjo- they'd not made any statements along the lines of what John just said there. YOU didn't have anything other than a gut level feel to go on- something you've not been 100% good on any more than I.

I disagree, all the signs have been there for people to view for quite some time. Over the last year Carmack has expressed his disappointment with how openGL 3 was handled, deflecting queries about linux versions of future games, his views on where the future of gaming lies with consoles and portables, and his replies to interviews saying iD and him are not as interested in linux as they once were. The writing has been on the wall for a while.

yogi_berra

07-27-2009, 01:02 PM

That's less due to the silicon (which is as good or better than NVidia's...) and more due to the drivers. Their codebase has been historically unstable and finicky. AMD's GLSL/HLSL compiler has always been quite a bit more picky on what was fed to it, for example. When it was done "right" (which is typically a slavish insistence on explicit compliance with the OpenGL and DirectX standards documents...) it works WELL and no issues. Otherwise...heh...

There shouldn't be any divorce between the hardware and drivers in this situation, because without a good driver the hardware may as well be a piece of corrugated cardboard glued in the pci slot, which is to say a useless waste of space.

L33F3R

07-27-2009, 01:12 PM

There shouldn't be any divorce between the hardware and drivers in this situation, because without a good driver the hardware may as well be a piece of corrugated cardboard glued in the pci slot, which is to say a useless waste of space.

well now not quite. Perhaps if you have a clear sided case you could view beautiful artwork on that cardboard. :D

Dragonlord

07-27-2009, 01:38 PM

I highly doubt you are pushing rendering technology the way Carmack does, but whatever helps you sleep at night.
Carmack pushes in a different way than I do. You can't compare the two directly. But no matter which way you push a crash in a game is to 99% of the case a problem in the code base not the driver. OpenGL is rather strict there and I only got things to crash so far if I did not comply with the specs.