Slashdot Mirror


Quake2 Engine In Java

An anonymous reader writes "Ok, so the game is old and there was a really poor web version some years back, but some guys at Bytonic Software in Germany have done a full source port of the Quake2 engine from C to Java. It's cross platform, performs just about as fast as C and has room for further improvements according to the developer. Also, there was another game engine that ran Q3 maps that was shown recently at JavaOne. Are first generation Java games that far behind?"

31 of 123 comments (clear)

  1. Java != Slow by kannibal_klown · · Score: 5, Insightful
    But I used java once, and it took forever to print Hello World. Java is slow. Java is stupid.


    Maybe this will lay waste to claims that java is slow, bloated, and sucks.

    I've only recently started doing heaving Java programming, and I have to say that the language is a dream to code with (provided you use a decent IDE). There're so many classes in place already, there's nothing you can't do. I'll take it over C++ any day, and MS's MFC is horrible on comparison.

    My only problem with it is the deployment; screwing with class paths and what not.

    People need to realize that most of the overhead they experience with their "hello world" experience has to do with loading the classes in the beginning. Once that is done, Java performs nicely.

    Sure, straight C is faster, but Java isn't the turtle everyone makes it out to be.
    1. Re:Java != Slow by blackbuddha · · Score: 5, Informative

      http://classworlds.codehaus.org/ will make your class path issues go right away. No more messing with the environment settings, no more screwing around with file paths, and only 29K of code.

    2. Re:Java != Slow by Prior+Restraint · · Score: 2, Insightful

      Maybe this will lay waste to claims that java is slow, bloated, and sucks.

      I would be more inclined to agree if this story were on the main page.

    3. Re:Java != Slow by FedeTXF · · Score: 4, Interesting

      Moreover. Im running Java 5.0 RC and the client VM has this nice feature called Class Data Sharing and I can see how much faster the startup times are. I'd say 2 to 3 times faster.

    4. Re:Java != Slow by macrom · · Score: 4, Funny

      I've only recently started doing heaving Java programming

      I hear ya, man. I did Java programming once and it made me heave.

      :^)

    5. Re:Java != Slow by GeckoX · · Score: 2, Insightful

      I believe you missed my point entirely. Stating that Java is as fast as C because an 8 year old game has been ported to it, and it runs fast on hardware that's 8 years newer than the original game would have been played on is NOT a valid comparison in any way.

      I'm not suggesting that Java sucks unless it can run on a P 233, which is how I think you took my comment.

      --
      No Comment.
    6. Re:Java != Slow by angel'o'sphere · · Score: 4, Insightful

      MFC is wonderful compared to 6502 assembly on the Atari 800 I dare to challange that.
      MFC was created a long time ago, during the 16-bit era. Thats no excuse for creating bull shit.
      It made some compromises for speed (e.g. message map macros instead of gazillions of virtual functions). A virtual function is in most circumstances faster than a message map macro.
      It was created before C++ was standardized, and frankly, true!
      before the OO community had learned some lessons (some of which Java directly benefitted from). wrong!
      When the MFC was crafted there existed plenty of better alternatives.
      a) MacApp, Apples C++ Application FrameWork
      b) TCL, Think Class Library, later Symantec, a simplyfied clone of MacApp
      c) Borlands Application Framework
      e) ET++, a cross platform C++ GUI and Application FrameWork
      f) Taligent research Frameworks, off springing from ET++
      g) BedRock a Borland + Apple + others attempt to make a Mac/PC cross platform Application FrameWork

      Every SmallTalk environment was allready far better than MFC in the early 70ties. CLOS and other lisp derivates allready had pretty well working GUI and Application FrameWorks.

      Bottom line: M$ was unable to craft a good C++ compiler. So they made a shitty incompatible slow monster with M$ specific extensions. Crafted a library relying on that extensions and sold that as "development environment".

      But it provided a nice MVC framework, and handles all your menu and toolbar and other GUI stuff for you. I started with C and the Win32 API, so MFC was "a dream to code with" in comparison. I wish we would evolve from saying something sucks or is horrible, and just accept that they "are", and pick the best one for each task.

      Your luck was that you did ot know better, I allways went out for one hour to hack lumber to get my agressions off, when I needed to work on a PC in 1993 till 1998 in "Windows" C++ and "Windows" libraries. Using Symnatec C++ (Walther Bright, anyone?) softened that pain a bit. The pain was further softend by using the zApp GUI libraries and not MFC. After searching 6 weeks for a good GUI Framework, the GUI of my app was made in 2 days, not with MFC, but with a true framework.

      I really don't blame M$ and gates for making money and trying to dominate the market.

      I blame them for throwing the whole computer age back by 30 years at least. They did so by killing so many good ideas and good companies. It took until 2000 to get again new ideas established in using and programming computers.

      Now hey are copying all good ideas from unix/linux/mac os x from KDE and GNOME. If they can dominate again, by killing Sun and bying SCO, or whatever, a lot of progress will be "removed from the market" again.

      Compare AppleScript versus Visual Basic. The first one is ingenious easy and the second one is ruling front end programming, but is utter crap.

      angel'o'sphere

      --
      Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.
  2. 0.9.2 version released by tod_miller · · Score: 4, Informative

    First released notice in May.

    This is a good demo of the power of Java, it handles the game, then passes this smoothly to the native opengl rendering. Jogl is great, I hope I can find time to work with it some more.

    Those crazy Germans do deserve some awesome credit for this! (having lived in Germany I can say I love Germans, and they are crazy! :-)

    Sourceforge page

    --
    #hostfile 0.0.0.0 primidi.com 0.0.0.0 www.primidi.com 0.0.0.0 radio.weblogs.com
    1. Re:0.9.2 version released by FFFish · · Score: 4, Funny

      German effort, eh?

      I guess that's why they didn't port Wolfenstein.

      [FFFish waves good bye to his karma...]

      --

      --
      Don't like it? Respond with words, not karma.
  3. Hooray by Shinglor · · Score: 4, Funny

    Now you can have the eye candy of Quake 2 with the speed of Doom 3!

  4. But do we care? by Turn-X+Alphonse · · Score: 5, Insightful

    I don't care if it's running in Java C D E or bloody Z, as long as the game runs well and looks playable (AKA not shooting green blocks to make red blocks appear with a black block and various brownish walls. ).

    No gamers is going to go "Doom 3 running on Java!? I'm so not buying that!" or "Doom 3 is running only on C++ I'm not playing that!". No one but the coders themselvs and the modders will truely care what it's written in as long as it runs okay and looks good.

    So gonna get modded Troll..

    --
    I like muppets.
    1. Re:But do we care? by iwadasn · · Score: 4, Insightful


      I'm a Mac user, I care. Too often Mac games come out years late and with dozens of bugs added in just for the fun of it. If the development houses would either do as blizzard does, and develop quality software, or at least write in java so I get the same bugs as everyone else, I'll be happy.

      This is after all the bottom line. I don't want to wait 4 years for a buggy Mac port to come out, and I don't want to use windows at home just so I can run the latest games. If more people went this route then it's likely that there would be more high quality games for everyone. Just write it once, release the java for OS X, Linux, BSD, Windows, whatever, and then use something like GCJ to turn your java code into a good executable for XBox or PS2. Of course I'll hear the C coders say "But C can be portable." and it can, but only if you know what you're doing and actually put in the effort. History has shown that very few people fall into this category, so we need java.

      Everybody wins, especially me. :-)

    2. Re:But do we care? by kaffiene · · Score: 4, Insightful

      No, users shouldn't care at all. But I can tell that there has been huge developer resistence to the idea that Java can do games at all.

      That's relevant because games need good quality game libraries and noone will write them for a target they think can't cope with the task.

  5. The performance is alright now... by jmhewitt · · Score: 3, Insightful

    but what about when Quake2 came out? The code runs at about 60% performance on today's machines. So when Quake2 was released, Quake2 Java would have broken my machine, as it barely ran the orignal Quake2 version. Java can offer portability over platforms but games often push hardware to its limits and so may raise the bar on minimum system specs.

  6. The language does not always matter by Alban · · Score: 5, Interesting

    The important thing to know is that the majority of your performance gains are obtained by scheduling the hardware intelligently, keeping the CPU and GPU well balanced, i.e. busy at all times. And of course, that your tight loops are really optimized, that you do not fragment memory, etc. That, for the most part, has nothing to do with the language your using, but simply with your programming skills.

    By the way, what hardware have they tested on to claim that performance is similar? If it's modern hardware, then of course it will run at 60+ fps no matter what.

    What's more, I would guess the bulk of the work on the CPU for quake2 consisted in traversing the BSP tree, building the scene (with transforms still being performed on the CPU at that time) and collision detection. The rest is taken care by the graphics hardware so that's totally independant of the language you've used.

    There is one thing that bothers me with Java though. You never know when the garbage collection will be performed. Sure, recent virtual machines make it possible to perform garbage collection in smaller but more frequent iterations so you don't halt the system for a few seconds like early virtual machines would do. But still, if you're in a tight loop with your data and instruction cache perfectly populated, and all of a sudden the garbage collection kicks in, then your cache is toast and data will have to be refetched to it when execution resumes. That would result in a horrible performance loss provided you are already really close to the machine's limit. Also, what I'm saying is only pertinent on a console with no (or almost no) OS, because on any PC operating system, your process can be interrupted at any time by the various system tasks that are running, so the garbage collection interrupting your tight loop would only be one of many possible interruptions.

    I don't believe java can be as fast as native code, although probably extremely close. And sure, a good java compiler will generate faster code then a crappy C++ compiler.

    Another thing I don't like about java is that you have no control over memory (not that i know of, maybe some recent VM extensions allow you to have some control over that?). I really like to be able to give different sets of alloc/dealloc routines to the different subsystems in a game. A subsystem that is known to perform very small allocations/deallocations very often could be passed alloc/free routines that are customized to its use so that memory fragmentation is kept to a minimum. If such a component was allowed to get memory from the same pool then your other subsystems, it would wreck havoc on your memory.

    Anyway, it's not such a good idea to compare java and C++ (or whatever other languages) on a system where resources are abundant (PCs).

    By the way, Jak & Daxter for the ps2 is written in a lisp derived language (GOAL), yet that game outperforms and looks better then almost everything else on the ps2. Yet lisp is not perceived as a high performance language. But the people at Naughty Dog have developed their very own compiler that is extremely specific to their needs (see their gamasutra post-mortem, very cool read) So, it goes to show that the notion of performance shouldn't be tied with a language, but rather with that language's runtime & compilers.

    Ok enough ranting!

    1. Re:The language does not always matter by Anonymous Coward · · Score: 3, Insightful

      I don't believe java can be as fast as native code, although probably extremely close. And sure, a good java compiler will generate faster code then a crappy C++ compiler.

      The usual 'support' for this argument goes: "Java programs have to be interpreted from bytecode and then run, but native code only has to be run. Bytecode interpretation slows you down.".

      That seems to ignore HP Labs' Dynamo project, which showed that an HP-PA 8000 emulator running on an HP-PA 8000 could run code faster than native HP-PA 8000 code.

      If interpreted HP-PA 8000 code is faster than native HP-PA 8000 code, I imagine a bytecode would have to be pretty poorly designed to perform worse. In fact, since JVM was built to be interpreted, I hope it's a lot better.

      Java probably isn't the best example of this, but I see no reason why a high-level, byte-compiled language couldn't run as fast or faster than native code, assuming a standard operating system with libraries and so forth.

    2. Re:The language does not always matter by the+eric+conspiracy · · Score: 4, Informative

      if you're in a tight loop with your data and instruction cache perfectly populated, and all of a sudden the garbage collection kicks in, then your cache is toast and data will have to be refetched to it when execution resumes.

      Actually, in the more recent JVMs garbage collection improves cache coherency through memory locality - newly allocated memory is likely to be close to already allocated memory, and thus in the cache.

      The same techniques can be exploited to cause stack allocation instead of heap allocation, another speed win.

      As far as compilation goes, Java compilers don't extensively optimize. Java optimizations typically occur at run time, which gives you a far greater range of possible strategies than the static compile time optimizations that are possible with C. Some people now think that run time optimization will eventually lead Java to be a higher performance language tham C.

      I don't believe java can be as fast as native code, although probably extremely close.

      It will be interesting to see how this works out, I'm betting that for a significant class of problems, large general purpose software, Java's runtime optimization will make it a clear winner.

      For small programs that can fit in a CPU cache though, I think Java will be slower to overtake C, if ever.

    3. Re:The language does not always matter by FyRE666 · · Score: 2, Interesting

      There is one thing that bothers me with Java though. You never know when the garbage collection will be performed...
      Actually on mobile phones, where you're working with extremely limited hardware and very little RAM, it's common practice to force the garbage collector to run regularly. I usually call the gc in the central game loop to prevent "hitches" during play as a large(ish) block is freed up randomly.

      Since just about every game I've written is under 30k in size (including graphics and sound), and most run in phones with as little as 100k of usable RAM, no hardware sprites etc, I'd say that Java can be pretty efficient!

    4. Re:The language does not always matter by angel'o'sphere · · Score: 2, Insightful

      With delete/free maybe?
      maybe? ... maybe not?
      To which OS exactly do you refer?

      Neither: Mac OS X, Mac OS 4 to Mac OS 8, DOS, Windows (3.1, NT, 95, 2000, XP) , Unix, Linux, or Mainframe OS does that.

      You simply have a misconception what freeing of memory (using free() etc.) means.

      angel'o'sphere

      --
      Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.
    5. Re:The language does not always matter by voodoo1man · · Score: 2, Insightful
      There is one thing that bothers me with Java though. You never know when the garbage collection will be performed.
      Here's a hint: it happens after you've allocated too much memory. Even if you are so deprived that you can't even turn the garbage collector on and off, if you don't allocate in your inner loop, the garbage collector won't kick in! Who would have thought? In Java allocation is very easy to control (a big reason for this has to do with it's separation of built-in types and classes), and as long as you take care not allocate yourself and be aware which library functions do, you are fine. As well, you can always use object pools to safely do manual allocation/deallocation. Any half-decent VM will allow you to tweak the garbage collector at run-time, providing a functional interface for forcing full or partial collections, changing the amount of memory allocated before triggering collection, or even turning the collector on and off.

      The thing to remember from all this is that the garbage collector doesn't work in a vacuum, and cannot be blamed for poor performance per se. You application and the garbage collector together form a system about which you can draw conclusions on performance. A well-written application can give hard real-time performance with a naive garbage collector, but any number of stupid things you can do while writing a program can bring even the best garbage collector to it's knees. (A fun thing to try in systems that support it - try re-defining a class to be it's own superclass. Makes a great practical joke!).

      --

      In the great CONS chain of life, you can either be the CAR or be in the CDR.

  7. garbage collection? by dayeight · · Score: 2, Interesting

    Now I haven't used java since 2000, but have they fixed the seemingly random garbage collection? I remember seeing a raytraced wolfy3d java demo years back, and garbage collection would make it go to a stand still at random intervals.

  8. Turtle language by John+Harrison · · Score: 5, Funny
    Sure, straight C is faster, but Java isn't the turtle everyone makes it out to be.

    No, that is Logo.

  9. High performance Q3/D3 graphics engine by GameNutz · · Score: 4, Informative
    For those of you who need to see what is happening with a high performance Java game engine running more modern games, check out the Aurgia 3D game engine http://www.auriga3d.org/pictures/picture.cgi?pict= q3dm1-1

    A brief snippit from the developer site:

    Auriga3D is an advanced real time game engine built on an extensible plugin framework. Presently, the engine is geared towards BSP style rendering popular in FPS games such as Quake3 & Doom3. In the future there may be additional plugins created that allows other rendering techniques such as height maps to be accessible. A goal of Auriga3D is to allow independent developers the ability to develop new games with existing content creation technology.

    Feature Glimpse:
    # OpenGL Binding Agnostic; supports JOGL & LWJGL
    # Quake3 Map Support
    # Texture Mipmapping
    # Multitexturing
    # Trilinear Filtering
    # Multiple Vertex Array Rendering or VBO Rendering
    # Lightmap Rendering
    # Potential Visibility Set
    # Frustum Culling System
    # JPEG, TGA, PNG texture support
  10. Re:Quake II .NET by bay43270 · · Score: 2, Insightful

    By "ported to Managed C++" you really mean "compiled to IL to run on the CLR". There's a big difference. This code is not managed, its just unsafe code running in a VM (there was more information on this on Channel 9 last week). The radar program they created was written in managed code.

  11. Wow, what a crock by Anonymous Coward · · Score: 4, Informative

    The summary says "Runs almost as fast as the original code!"

    The article says "Runs somewhere between 60 and 85% of the speed of the original", and this is on modern hardware.

    Let's see how it performs on hardware that was actually used to play Quake 2 before we start lying about how fast java is compared to the native code. When the hardware in question is capable of pulling 300 frames per second, it's pretty damn likely it's not even being used to its full potential. Even the K6-2 machine was getting near 60fps. The only people who got 60fps in Quake 2 when it came out were the people with monster machines.

  12. Yes, sorta by daVinci1980 · · Score: 2, Informative
    Are first generation Java games that far behind?


    Sorta. There are numerous games out already that are based on Java. Pretty much all of Popcap uses Java. Java is also used as a scripting language for several games (sorry, no links), as an alternative to Python, Lisp, C++, Lua or any other interpretive language (or home-brewed language).

    However, it will be a very long time (if ever) before developers switch over to Java as their language of choice. Why? Because it is only the last generation or two of games that have started to use C++.

    Game developers are constantly trying to push the edges of what can be done within current hardware limitations. The problem with languages that do a lot for us is that... They do a lot for us. Which is nice a lot of times. But always seems to happen at the most inconvenient time (like when we're doing matrix operations, or animating units).
    --
    I currently have no clever signature witicism to add here.
  13. Vampire: the Masquerade by clamatius · · Score: 3, Interesting

    I happen to know of at least one AAA game which used Java - it used it as a scripting engine.

    Nihilistic's Vampire: the Masquerade - Redemption, back in 2000. As I recall, in the Gamasutra postmortem, they commented on how well it worked out for them.

    Sadly, I don't know what JVM they were using - but they did say in the postmortem that they didn't write it themselves.

  14. Re:Native Libraries? by OmniVector · · Score: 2, Informative

    those are the native implementations of jogl and joal (java opengl and java openal). you have to have native c interfaces with the system libraries to achieve these sorts of things.

    --
    - tristan
  15. Nothing new by mirabilos · · Score: 2, Interesting

    Tr3B (Robert Beckebans IIRC) has written a Q2 clone
    in Python some time ago, and currently hacks on
    qrazor-fx aka http://xreal.sf.net/ which is a Q2
    with graphics quality in the range of Q3 or even
    Doom 3.

    --
    My Karma isn't excellent, damn it! (And /. still does not get UTF-8 right in 2012. Wow.)
  16. Re:If Java is so portable... by Decaff · · Score: 2, Insightful

    And I'm a grown up, so I can *decide* if I want to use pointers, gotos, or multiple inhieritance. If you don't like these features in C++ THEN DONT USE THEM.

    We may be all grown-ups, but its a sad fact of IT development that we all make mistakes, and pointers aren't an optional extra in C++, they are fundamental to the way data is stored. The use of C/C++ has been a disaster in many areas of IT, leading to buffer overruns, buggy software and virus susceptibility. (I remember spending days trying to trace a buffer overrun in a C++ program on DOS).

    Unless you are writing OS kernels, device drivers or hardware controllers its hard to see why anyone would feel the need to use raw memory pointers in an application.

  17. First Generation Java Games by jvsquare · · Score: 4, Informative
    > Are first generation Java games that far behind?

    The first commercial Java game was Tom Clancy's Politika, published in 1997. This was followed by Tom Clancy's Ruthless.com in 1998, and Shadow Watch in 1999. I'm not exactly sure what the comment above means, but personally I consider those games first generation. They were burnt onto CD and sold in stores like many a C or C++ title.

    Whereas some C/C++ games have used Java as a scripting language, the approach of these games was to use Java for the core game loop and game logic, and to write new native libraries for the performance heavy stuff like graphics and sound. For Politika, we wrote our own movie and sound code (for both PC and Mac) because Java didn't have very good support at that point. We almost got some faster and leaner 2D graphics support in there too, but ran out of time. This approach was written up for a paper (Writing Java Games: How We Did It), which was presented at CGDC 1998.

    In Ruthless.com, the native libraries were improved and the 2D graphics support was added. Basically a wrapper was created around DirectX using JNI. In addition, the whole game was compiled to native code instead of using the Java VM. This whole system hit its peak with Shadow Watch, which is an awesome little turn-based strategy game; think Rainbow Six Tactics.

    So this has been possible for many years -- it's just that nobody's carried it on. The big accomplishment here is that they may be using only Sun libraries, which only means that Sun has finally gotten their act together and put out well-optimized code.