Any recent thought as to Linux OpenGL port of DCS ?

[DmitriKozlowsky]

Any recent thought as to Linux OpenGL port of DCS ?

Linux has grown up somewhat. Still no MS DirectX, but there is OpenGL, Nvidia Vulkan, AMD Radeon API. In my day job I use Maya, Houdini, NUke, other OpenGL. Looking at their installations, there is just Linux version, without disto-specific versions m that used to be norm.

[okopanja]

Same interest, but perhaps, we wait for Vulkan on Windows first?

 

[cfrag]

Or a native (not wine-based) Linux port of the stand-alone server could also help. 

Edited September 3, 2022 by cfrag

[SharpeXB]

 

[upyr1]

If I could do DCS in VR using Linux i would

[okopanja]

1 hour ago, cfrag said:

Or a native (not wine-based) Linux port of the stand-alone server could also help. 

 

This would be wonderful, and would make operating of servers both simpler and chreaper...

[SharkWizard]

6 hours ago, DmitriKozlowsky said:

Nvidia Vulkan,

Vulkan is from AMD originally I believe and has since become cross-platform api

Edited September 3, 2022 by SharkWizard
added info

[okopanja]

Well, we have 2 distinct linux areas:

• Linux DCS client probably ED not too interested although there is clear interest among us
• Linux DCS server, would be absolute blast:
  • no windows associated licenses => lower cost
  • linux => more secure
  • ability to deploy into containerized environments => better hardware utilization => lower costs
  • if combined with selective map downloads, lots of CM tricks: e.g. rapid server updates (while the server is running old version, new container is build with latest version)

Just to name some benefits.

 

[SharpeXB]

24 minutes ago, okopanja said:

Linux DCS client probably ED not too interested although there is clear interest among us

ED already said they have no plans for Linux. See link above. 
Not sure how you figure there is any “clear interest” among players as only 1.27% of this market uses Linux. 
https://store.steampowered.com/hwsurvey

This question has already been asked and answered. 

Edited September 3, 2022 by SharpeXB

[SkateZilla]

There are no plans for Linux,

Also OP, nVidia does not own Vulkan,

if you want to get technical, Vulkan was born from AMD's Mantle API, so technically it's a child of an AMD API.

[okopanja]

10 hours ago, SharpeXB said:

ED already said they have no plans for Linux. See link above. 
Not sure how you figure there is any “clear interest” among players as only 1.27% of this market uses Linux. 
https://store.steampowered.com/hwsurvey

This question has already been asked and answered. 

 

Thank for providing a link, and yes I have read the topic you have quoted there (I see more than once of the commenters in that topic are here, moved on here), including your rather fast remark that less  than that 0.86% of population uses linux. Therefore you can see that the user base of Linux did grow within one year significantly.

Let me try to elaborate this a bit more. Linux is indeed my primary OS and the only reason I have Windows on my machine is to play DCS. At the same I am professional SW developer and have in the past developed cross platform applications that run on Win/Mac/Linux. Making DCS cross-platform SW does require effort, and I totally understand that ED does not want to do that, but overall the experience shows that such codebases contain 99% of commonality, while only 1% is the code specific to the platform. In case of video games major challenge is GPU interaction, and writing your own adapter toward different APIs is usually not a good idea, since it may produce different experiences for different users. However the advance of the Vulkan in DCS will mean that there will be a common API across all of those platforms, and hence the primary hurdle will be removed. Additional benefit of writing the cross platform software is that it forces developers to write a cleaner and better quality code (different compilers tolerate ugly code in different way). Adding a single new platform typically uncovers a myriad of hidden issues on the original platform. This way as a windows user you would also benefit from it.

Again, I will ask you to read the second part of my comment, that relates to server, which we would all benefit from, including ED itself (for the sake of conducting tests).

In addition while troubleshooting the reasons for performance degradation of a DCS server, I found out that the server is nothing more than lobotomized client, which still makes the calls toward the things that one server should not interact with (e.g. checking flightsticks!). You see there is a plenty of the things that ED can improve.

Now I do not advocate for a "hostile" thread against ED or anything like that, but from time to time they should be reminded about the need to write better quality software.

 

 

[MAXsenna]

3 hours ago, okopanja said:

Now I do not advocate for a "hostile" thread against ED or anything like that, but from time to time they should be reminded about the need to write better quality software.

They are aware, and have even admitted to having "spaghetti code". You'd known if you'd been here the spring of 2020. This was heavily discussed then. 

This might be primarily the reason Linux is out of the question currently. It's been said it will take years and 50M to make a total rewrite.

I wouldn't mind a server version for Linux, but for the client, too much stuff I use for DCS depends on Windows.  

[okopanja]

51 minutes ago, MAXsenna said:

I wouldn't mind a server version for Linux, but for the client, too much stuff I use for DCS depends on Windows.  

LOL, I wonder which ones. Keep in mind it does not have to come all in one package...

But nevertheless it would be nice to have linux server...

[5ephir0th]

16 hours ago, SharkWizard said:

Vulkan is from AMD originally I believe and has since become cross-platform api

 

Well, exactly AMD developed with DiCE (the guys from Battlefield series) the Mantle API but Vulkan, while based at first on Vulkan, was (and are) developed by Khronos Group

[MAXsenna]

20 minutes ago, okopanja said:

Keep in mind it does not have to come all in one package...

'Course it doesn't. Problems are, they don't come in any package.

One example, close to seven years ago.

null

 

[SharpeXB]

7 hours ago, okopanja said:

Thank for providing a link, and yes I have read the topic you have quoted there (I see more than once of the commenters in that topic are here, moved on here), including your rather fast remark that less  than that 0.86% of population uses linux. Therefore you can see that the user base of Linux did grow within one year significantly.

It doesn’t matter what changes there might be to this number as long as it’s always around 1-2%. I don’t think this has ever changed. It’s not a matter of programming science that makes Linux infeasible for DCS it’s a simple question of the market share. And it’s worth remembering that this is a tiny niche game. Spending $50million to reach an extra 1% of its potential customers is clearly not worthwhile. 

Edited September 4, 2022 by SharpeXB

[okopanja]

16 hours ago, SharpeXB said:

And it’s worth remembering that this is a tiny niche game. Spending $50million to reach an extra 1% of its potential customers is clearly not worthwhile.

I see you like crunching numbers and have very rational view on spending vs earning.

I hope you will not mind if I dive a bit deeper into this topic and wish to apologize from digressing from original post.

Let's assume that a developer's hour cost $80-$100 (e.g. this is a price of a developer including not only his salary, taxes, office, electricity, equipment, in short every cost related to developer) in developed world (in Russia likely 30%-70% of that value).

$50.000.000 / $100 per hour, translates into 500.000 developer hours, which itself is 62.500 developer days, translated to years: 284 (I used a working year of 220 days). Now you just need to calculate how soon you want it, e.g. in 1 year: you can hire the army of 284 developers for a period of 1 calendar year.

Off course the problem here is that this is not 100% distributive, and these $50M do not bring profit directly and immediately, but represent spending from account point of view (no direct income). Failure to view this as investment into future and competitiveness, is a common pitfall many companies make. Fortunately for ED, the competition is focused on less demanding customers.

Now I do not know where the $50M came from, nor will contest it (please do not quote me the link), purpose is just to give you an idea on how much the $50M is.

[cfrag]

While I agree that a Linux Client is far off, and a doubtful business proposal, I really hope that ED can make some headway wrt the stand-alone server part. Today's DCS is a bit of a two-headed beast: what appears to be surprisingly good netcode wedded to the worst possible network platform: Win. If ED manage to parcel out server and client (as they should and probably will do in the coming years), we can finally run the DCS much better (run in separate threads on the same computer so single-player performance gets better due to better thread utilization), and the opportunity to port the server part to *nix/*nux machines for much better net integration and virtualization. One of the reasons that dedicated servers are so expensive is that you need a frigging Win machine. Since happily, DCS server hosting is currently taking off as a business (as may be witnessed by the success the great people at Fox3 have), being able to host DCS standalone on Linux or Unix would help to drive down server rental fees, which would help to further popularize DCS MP.

Edited September 5, 2022 by cfrag

[SharpeXB]

5 hours ago, okopanja said:

I see you like crunching numbers and have very rational view on spending vs earning.

I hope you will not mind if I dive a bit deeper into this topic and wish to apologize from digressing from original post.

Let's assume that a developer's hour cost $80-$100 (e.g. this is a price of a developer including not only his salary, taxes, office, electricity, equipment, in short every cost related to developer) in developed world (in Russia likely 30%-70% of that value).

$50.000.000 / $100 per hour, translates into 500.000 developer hours, which itself is 62.500 developer days, translated to years: 284 (I used a working year of 220 days). Now you just need to calculate how soon you want it, e.g. in 1 year: you can hire the army of 284 developers for a period of 1 calendar year.

Off course the problem here is that this is not 100% distributive, and these $50M do not bring profit directly and immediately, but represent spending from account point of view (no direct income). Failure to view this as investment into future and competitiveness, is a common pitfall many companies make. Fortunately for ED, the competition is focused on less demanding customers.

Now I do not know where the $50M came from, nor will contest it (please do not quote me the link), purpose is just to give you an idea on how much the $50M is.

It’s funny that you write about man hours as if ED or anyone else with a job doesn’t know all this. I was referencing another post here for the $50M. No idea if that’s correct or not. 

2 hours ago, cfrag said:

While I agree that a Linux Client is far off

Let’s be clear, it’s not “far off”, it’s never…

Edited September 5, 2022 by SharpeXB

[okopanja]

Just now, SharpeXB said:

It’s funny that you write about man hours as if ED or anyone else with a job doesn’t know all this. I was referencing another post here for the $50M

The point is that major rework would cost certain amount for sure but probably not that much. They already have to do that either way. Already now, the current architecture struggles with simulating everything on time to be rendered. If they try to solve this brute force, by pushing us to newer, faster CPUs/GPUs, likely I can follow, but large number of user base would in this case suffer, or simply leave. Just consider the costs it take to build a PC, Windows license is not small part of it, and those $100-$200 for windows license could be spent elsewhere.

[SharpeXB]

55 minutes ago, okopanja said:

The point is that major rework would cost certain amount for sure but probably not that much. They already have to do that either way. Already now, the current architecture struggles with simulating everything on time to be rendered. If they try to solve this brute force, by pushing us to newer, faster CPUs/GPUs, likely I can follow, but large number of user base would in this case suffer, or simply leave. Just consider the costs it take to build a PC, Windows license is not small part of it, and those $100-$200 for windows license could be spent elsewhere.

But to reach an extra +/- 1% of their customers it will likely never be worthwhile. They know this already, it’s their business after all and they’ve answered the question.
It’s not a “large number of user base” I don’t know what would lead you to believe it’s some large untapped market. The vast majority of gamers or consumers just don’t use Linux. It’s not a worthwhile savings to avoid Windows because your machine wouldn’t be able to run most any software let alone games. Linux just doesn’t make sense for the average PC owner. 

Edited September 5, 2022 by SharpeXB

[cfrag]

18 minutes ago, SharpeXB said:

But to reach an extra +/- 1% of their customers it will likely never be worthwhile

I think that the 1% isn't the point here, that would indeed be pointless. It's refactoring DCS to a multiplatform code base, so it can be easily/instantaneously distributed to multiple other platforms (that's how big game engines do it, and ED is approaching big game manufacturer status), and Linux may be be a lucky winner here. Usually, game companies target consoles to grow their market because of their huge established base; see some other big-name flight sims that shall not be named here. When you gain the ability to multi-target different platforms at once, you often end up with some form of *ux or *ix as a target. If you can get that for nearly free, it may become an interesting low-hanging fruit. So Linux may have an accidental victory in this regard.

1 hour ago, SharpeXB said:

Let’s be clear, it’s not “far off”, it’s never…

"Never say never." - You don't speak for ED, we don't know the future, and it's not the first time ED managed to make Hell freeze over.  

[SharpeXB]

17 minutes ago, cfrag said:

I think that the 1% isn't the point here, that would indeed be pointless. It's refactoring DCS to a multiplatform code base, so it can be easily/instantaneously distributed to multiple other platforms (that's how big game engines do it, and ED is approaching big game manufacturer status), and Linux may be be a lucky winner here. Usually, game companies target consoles to grow their market because of their huge established base; see some other big-name flight sims that shall not be named here. When you gain the ability to multi-target different platforms at once, you often end up with some form of *ux or *ix as a target. If you can get that for nearly free, it may become an interesting low-hanging fruit. So Linux may have an accidental victory in this regard.

"Never say never." - You don't speak for ED, we don't know the future, and it's not the first time ED managed to make Hell freeze over.  

Hell would indeed freeze over before Linux becomes the dominant OS

[cfrag]

8 minutes ago, SharpeXB said:

Hell would indeed freeze over before Linux becomes the dominant OS

Technically, it already is: Android. 

Man, it's getting cold in here!

[SharpeXB]

26 minutes ago, cfrag said:

Technically, it already is: Android. 

Man, it's getting cold in here!

We’re talking about PCs like DCS runs on.

Funny how you guys gotta keep this stuff going after the question has been clearly answered and the reasoning for that answer is completely obvious.

Edited September 5, 2022 by SharpeXB