Linux and Dual Processors support

[x582gr51 0]

Linux and Dual Processors support could be good things since there is so much addons used and missions have loops that go over 756 units adding to this many vehicles then it take a lot of cpu power and memory.

These days dual processor is affordable and many gamers use them, i am one.

Why linux then we all know that linux perform way better that windows. No slow virtual memory under linux only a swap that is rarely used, praticaly never.

I think these could reveal new paths and take the game further!!!

[void_false 1]

Yeah. LinuX would be great so we dont need to use silly emulators.

I love ID since they made Linux version of D3

[benu 1]

Well, you could bother Transgaming to finish support for OFP in WineX. It's nearly there, there are only some texture brightness issues left to work out.

But a native linux client would be cool.

[void_false 1]

IIRC WineX now called Cadgeta or something similar.

[Mr_Tea 0]

Yep, Dual Processor support and support for 64 Bit Processors and Systems is realy needed.

[Odin -AEF-Kampa- 0]

specialy for CTI :P

Maybe like this:

CPU1 for AI

CPU2 for Clients/Scripts

[benu 1]

IIRC WineX now called Cadgeta or something similar.

Cedega. But who would have known THAT name?

[x582gr51 0]

last time i was patenting wineX it was impossible to get directX applications working .

so a real linux port would be better

or i dont know what so special about directx else than it's microsoft product and 90% of computer users are running games under Windows XXXX. Maybe graphic stuffs are lighter than other engines. But if graphics could use OpenGL or any *GL then would be easier to convert ofp client to linux.

Maybe a BIS pro can clarify things bout OFP DX and *GL possibilities and limitations.

For the Dual processor dispatching AI to only one processor will not improve anything since the good thing about having more than one processor is the load balancing and the bad thing is that windows kernel isnt a good dispatcher neither a good balancer, unless some serious tweaks that are not availables to comon users.

Once again BIS pros are called to teach us about this aspect.

[LauryThorn 0]

Yeah. LinuX would be great so we dont need to use silly emulators.

Has someone actually ran OFP on some emulator in Linux? I didn't even try, I just bought a 700MHz P3 as a secondary computer and installed W2K+OFP on it. My main computer runs Debian. I've tried to play some Windows games on WINE, but I couldn't get even Close Combat II to run smoothly.

[benu 1]

I tried with cedega 4.2.1 a few days ago and it seem to run ok. There were some texture issues with older versions of cedega, but those were minimal, ie only in the mainmenu cutscenens and the briefing. I didn't multiplayer performance or mods yet. I am not sure how well it would run with other apps like ofp watch, teamspeak overlay or dxdll. For teamspeak exists a native linux client, so that would not be the problem.

But i have to agree that a native linux versions would be much better than having to emulate. Especially as those are faster most of the times than the windows versions.

[der bastler 0]

I second this, a Linux port would be nice. Perhaps BIS could specify the problem and how we --the community-- can help. It's not that they should stop the overall OFP2 development and spend their precious time just to program a client for a minority, but with the right project layout they could help us by providing the right foundation for a native Linux port.

Various SDKs for games under Linux are available

http://www.libsdl.org/index.php

http://www.clanlib.org/

Look at Doom 3, UT2004, Enemy Territory. You buy a CD/DVD with the game's data and install a Linux client over your own distribution's package management system (e.g. in my case: "emerge ut2004" -> copies UT2004 binaries from DVD to HDD). The client doesn't have to be open source (protecting intellectual property if neccessary) and uses platform-independent game data (graphics, sounds, models etc.). Even some mods are available via the package management system.

[benu 1]

Besides that, most of the native linux games (ut, quake3, rtcw:et) run faster on linux than on windows.

[shinRaiden 0]

Some musings on this in light of evolutions since the initial announcement of OFP2 some time ago...

I think the time if and when dual-processors make it to the consumer desktop in even negligible quantities are still some ways off. However, the likely market penetration of dual-core chips is much more imminently relevant. In either case, you have to significantly restructure the code to be much more multi-threading designed and optimized. I think in the process that better code design in that model could create better stream paths for closer to real-time priority for function components, regardless of the number of cores or processors it's executed on.

This also adds to the MP scalability capability, by extending server processing capability to stupid amounts of CPU and memory. Me personally, I want clustering support with dynamic cross-over, but I also want streamed 1m res terrain dynamically generated from GIS data too. On the other hand, reworking the code for better multithreading would also be a great time and opportunity structure that code so as to do as concurrant as possible simultanious builds for WindowsMP and LinuxMP.

Anyway, on to Linux issues. There are several issues involved, and not just video API's. OFP uses DX API's for Video, Audio, Input, and Networking, plus voice-chat. The net-code sucks, no need to refresh that and the voice-chat complaints. That's why we have sockets. I'd rather see a VoIP frontend, but what really needs to happen is for the gameDev industry to make a XML VoIP schema so that the games can remote-control an external VoIP chat app, and let an external provider handle that.

SDL libs would help with the Audio and abstracted input methods somewhat. Creative is rather aggressive with their EAX pushiness, and unfortunately that saddles you with a noticable CPU hit and nasty IP messes. Dolby has their own share of IP issues historically, but at least you can off-load the CPU work to a standardized performance system. Your input methods are pretty simple cross-platform, however joysticks can get complicated where you have specialty wrapper drivers that custom config the joystick controls. These often involve fake-keyboard cheats and such, and often are exclusively for Windows.

The big item of course is the graphics system being done currently in DirectX. This isn't an HL2 vs Doom3 or ATI vs Nvidia issue like AA/AS double depth or cG programming or some other little obscure detail like that, it's the simple fact that Microsoft has every reason to not port DirectX to Linux, and the Linux community has every obligation to not accept the kludgey insecure DirectX layered hooks into Linux. OpenGL is stable, reliable, and documented on both platforms, DirectX is not. That's the basis of the arguement. Does M$ hold you hand and make simpler support and make D3D more dev-friendly? Yes, but at a price.

On top of this though is the cold hard reality of market demographics. Is Joe Walmart going to have even a dual-core machine in the next five years? Is he even going to care or notice? Bob Bestbuy doesn't even change the passwords on his brand-name PC, let alone know what Linux is. What's the honest potential market, and ROI for shipping to the Linux market, other than tweakers, hackers, and crazy server admins?

The biggest positive thing that could come from this though is that the code could be cleaned up enough in the process to make it really begin to sing and dance, regardless of whether it ever actually made it to Linux. Like my previous rants about what I believe to be the realistic potential of OFP:Xbox, I do think though that the exercise/effort is valuable in fixing up the PC version.

There's one other item I neglected to mention. There are two additional licensed libraries, the Intel JPEG library and the Immersion Foundation Classes (IFC). I'm not positive atm whether the IFC libs are physics or DirectInput or both, but it's still a significant component that would need to be compensated for.

This brings up one other item that is admittedly a tweaker component for the near future, but may have significant future growth potential, the dedicated physics processor card. One solution was recently announced by Ageia at the recent Game Developer's Conference with announced support by Ubisoft and the next Unreal platform. Initial solutions are targeted for Christmas 2005. It's going to take some time before a product of this kind gains wide market penetration, but it still has a lot of potential.

[ohard1 0]

this would havea to be THE best idea i have ever heard.

i am a lover of linux, open source and the whole anti-microsoft idea, and OFP for linux would be the best!

i am looking at some of the source codeing for an emulator now, and i can see straight away (even with the basicness of by linux programming ability) why this is such a huge problem. The way the video..., im not going to bore you, but a linux version would be the true answer.

Then i can throw my crappy windows partition out, and have a 100% linux system.

for those of you who have a huge problem with emulators, try upgrading you current version of KDE or GNOME, and if that dont help, try switching to the other.

E mail me (ohard1@eq.edu.au) with any question or comments please: i probably won't be here for a while (going on holiday) so i wont be able to see your replies easily (my handheld has a problem with this site)

ohard1