Related
I am just getting started with coding and app development using LibGDX and i´ve got some general and some more specific questions about it.
Before I start: I know that some of my questions might have already been asked somewhere on the internet but I found it really hard to get a straight answer to them.
So, the most basic question first:
Is it possible to create an average application (like angry birds or doodle jump) in LibGDX in a reasonable amount of time?
Browsing the internet I came across many people stating that engines like Unity are "just so much easier to use" and that you can create a game in a fraction of the time it would take you to code a game in LibGDX.
Is that really the case?
My second question connects to the first:
Is LibGDX really worth the time? Picking up any engine or Framework seems like a lot of work, so I wondered if LibGDX is a growing platform worth learning or just an irrelevant framework of many.
The third question is a bit more specific (and the most important one for me):
How can I make LibGDX game pretty?
I know, that sounds weird at first but looking at hundreds of apps in all kind of appstores I´ve noticed that most unity apps look really neat whereas most apps found in the Badlogic gallery look just awful to be honest.
So I wondered is there a simple way to add animations, particle effects, shadows, etc. to a LibGDX application. If I was coding a 2048 clone (just as an example) how can I make my tiles slide together and blend nicely?
Creating good looking animations is something I struggle with the most...
(If you want you can have a look at what I created)
I would really appreciate any answer to any of my questions!
Thank you for making it through this huge post.
MrMorph
Question 1:
If you're good at LibGDX and know quite a bit about the API, it would be easy to make a game like Angry Birds or Doodle Jump. Unity is easier to use, because GDX doesn't have a GUI editor - meaning you have to position things using coordinates.
Question 2:
I think LibGDX is worth doing. It's free and royalty-free, which is good unlike Unity, and it's on Java so it's relatively fast.
Question 3:
To make particles, there's a 2D and 3D particle editor which you can download of the LibGDX website or run from your IDE if you import the 'tools' module. For animations, you can make sprite-sheet animations easily with the Animation class and you can move things around smoothly (for your 2048 clone) using functions like MathUtils.lerp() to smoothly interpolate a value to another, over time, slowing down as it gets nearer.
A late answer, but hopefully someone will find this useful.
Checkout my answer to a similar question here: What I need to get android game with graphics like this?
I've used both libgdx and Unity and talk about my experience in the answer. At this point I'm in favor for Unity for most small/medium sized games. The editor really does make development much easier (especially for a one-man team). Libgdx + Overlap2D + Ashley (entity component system) is basically a very light-weight version of unity imo. Going the open-source route, you will definitely learn more since you'll be more involved in gluing these tools/concepts together. To answer your individual questions directly:
Question 1:
There are many good games made with libgdx. The reason you see a lot of crappy looking games in Gallery is because the submissions are sorted by time. The gallery does not prioritize good games to be shown first, this was a community decision a while back. Engines like Unity of course show off the best games first. To see some good looking libgdx games, checkout games by Robotality. Also, take a look at the "Showcase" section of libgdx forums; sorting by reply count will show some interesting games made in past few years.
As for the the claims that Unity is easier to use, I agree. I've made the switch myself just because how easy it is to come back to a project after months, and just starting playing around with it. Not as easy when your game is mostly code.
Question 2:
Libgdx is the better way to go if you want to learn. Because it's a framework, you'll be dealing more closely with the components that make up a game engine. You'll be putting your own game engine together with the APIs provided. Unlike Unity, you'll have to choose whether you want to use an Entity Component System like Ashly. You'll decide when physics engines get updated, when you render, how you load and manage assets, etc. Once again, unlike Unity, there is no enforced structure in place.
You'll be spending time reading libgdx's very well written source code as well as the excellent technical github documentation. I'd say the overall effort to make a game is greater when using libgdx, because of the learning. It's great if you have the time, but can be frustrating if you're developing only once in a while as a hobby.
Question 3:
The question is a little unclear. As I've stated above, there are plenty of crappy looking games made with unity, they just don't show them on their gallery/showcase game. Plenty of games with bad art are made with all popular game engines.
Libgdx has the things you want: checkout Interpolation functions for smoothly controlling object movements, particle editor to create & load particles, not sure about shadows (for 3D?).
Hope this will help someone decide what tool to use.
Yes it is very possible to create a simple game in libgdx. I used to use this framework all the time. If you are semi new to it YouTube has some awesome tutorials.
Now I use unity primarily, and here is why:
It is easier from a productivity standpoint yes.
Libgdx can be a lot of work when trying to work with graphics.
Plus it was an easy transition from Java to c#.
And yes in most cases unity can move a bit faster, again from a productivity standpoint, in libgdx there is more of a process to setting up graphics and viewports and etc.
libgdx is 100% worth it. A big part of what libgdx helped me with was grasping how the backed of these applications work. This has helped me a great deal in debugging problems in unity and just advancing my knowledge overall.
If you have good graphics and understanding of the framework then you can make a libgdx game that does look beautiful. The difference is that unity simplifies the process for you. And all the demos you see are from a huge pool of projects. With libgdx being such a good start to game development, there are going to be plenty of projects, with no proof of how much work and or time was put in.
I plan to build a game for mobile. My target platforms are iOS and Android. I hope to be able to carry out testing on windows and/or linux. I'm a bit over my head in this as aside from general game development experience, all the technologies are fairly new to me. I've done a decent amount of research and have concluded that the game should be written in/ported to C++ to ensure that most of my engine can be easily ported to multiple devices.
Another thing I'd prefer is to write the game initially in java and port to C++ since that's where I'm most comfortable and will get it done the fastest.
Now I have the issue of choosing whether to use a game engine such as cocos2d-x, or to write the engine from scratch and use OpenGL ES 2.0 for rendering. Initially I thought it would be better to use cocos2d-x since I don't have a lot of experience with OpenGL. My problem with using cocos2d-x, however, is that since I want to write the game initially in java, that I'm going to have a hard time porting the engine to conform to cocos2d-x (or I'd have to learn all about the cocos2d-x engine to begin with and then write my engine to mimic the cocos2d-x engine.. seems redundant).
Upon further consideration, I thought that writing my own engine using OpenGL would actually be the better option. I'm able to use the PowerVR SDK along with JOGL to emulate a GLES environment. Also it seemed nice since I would be able to allow GL to do most of the work for me in terms of collision detection and transformations. My only issue with this is that since the game is going to have multiplayer support as well, the GL collision detection and such is basically moot since I'm going to have to do collision checking server side anyway to prevent the game from being easily hacked. Of course for the single player game play this method is viable.
Obviously this decision is subjective and depends on my personal preference, though I hope to have given enough background for some more experienced persons to lend their opinions.
That being said, my question is: given these parameters - would it be better for me to use cocos2d-x and suffer the head trauma of building the game from the ground up in C++, or would it be better to write my own engine initially in java and struggle through the OpenGL aspect of it?
Building your own engine from scratch would be too much of a work and would attract lot's of bugs to creep into. The tried and tested engines are better choice in my view. The porting would be far more easier to do, since you would be having a working game.
I am using Marmalade engine currently and I am quite happy with it's performance. It gives you direct access to the OpenGL libraries, although I've never tried it. You should give it a try too, although it's not free, but the 90 days evaluation license is free to check it out.
I've decided to write it from the ground up in c++ using cocos2d-x. It hasn't been as bad as I expected.
I've made my share of 2D games on various platforms but I have never developed a 3D game.
I want to make a small "mmorpg". I already made my server in python and it works just fine with my flash 2D game but I decided I want to step it up and try out 3D. I want to make a 3D game for the web browser and I think Java might be a good choice for this.
So basically I'm just looking for a straight forward and well documents 'framework' to make LOW-END 3D games. Keep in mind that I will be targeting peoples with very low-end PC's (plus my 3d modeling skills aren't great so I wouldn't mind hiding it somewhat, haha)
If you care to develop your own software 3D engine, which is pretty cool, Developing Games in Java is a complete walkthrough, step-by-step, of developing a 3D engine in pure Java, capable of rendering textured and lit polygons. You learn a lot about the math involved and you realize that it's really not a terribly hard thing to do; in addition, the engine is all yours, so you know it inside and out and you don't have to learn an API. On the flipside, it might be outdated. It's been sitting on my shelf for a number of years now, but it is made with Java 1.4 so it's not all too old.
Otherwise, I would definitely recommend JOGL or its competitor LWJGL; however, both require OpenGL knowledge, so if you want to just deal with loading 3D models and moving them around, jMonkeyEngine could be a better option for you. There is also the lesser-known Xith3D engine, somewhat a competitor to jME, though it hasn't been updated in over a year.
P.S. Ever seen RuneScape? It used JOGL, though now I think they switched to their own port of only the OpenGL functions that their code uses, kind of like a stripped-down version of JOGL.
JOGL would be a good possibility. You could look at the older "Java3D" framework as well.
You might want to check out the jMonkeyEngine.
I would advise against Java3D. We're using it for a project and frequently run into gotchas. If we had the resources, I'd migrate to something else in a second.
Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
Questions asking us to recommend or find a tool, library or favorite off-site resource are off-topic for Stack Overflow as they tend to attract opinionated answers and spam. Instead, describe the problem and what has been done so far to solve it.
Closed 9 years ago.
Improve this question
I've been a web programmer for a while and I can also program in Java. I have an idea for a small, multiplayer RPG game that I want to work on. It will be played through a java applet in the user's web browser.
I have written the design document and specifications of the gameplay. What I'd like to know now is how I can develop the game? I've worked only with windows-like business apps in the past, with built-in widgets for textboxes, dropdowns, etc. With game programming it seems that I would have to build my own widget/controls for the UI of the game.
These are the specific questions I have in mind:
1) How to display a 'loading...' message with a progress bar while the game's images, sound, etc are being downloaded. (Using the java applet)
2) How to create the UI of the game with its own menu, controls, etc. Such as by clicking the map icon it would show up a map to them. Clicking the friends icon would let them chat to their friends, etc.
3) And other, general game development issues that i should know about, like whether I should use 2D or 3D graphics, physics in games, etc.
If there's a good recommendation for a book that will help me, do share.
If this is your first game, a multiplayer project might be too ambitious.
Loading screens are not terribly hard - I implement them by counting up how many files I have to 'load' and then advancing the progress bar (and doing a screen refresh) every time another file has been loaded successfully. You can do this with as much granularity as you please - it might complicate your loading code to add the UI component. I wouldn't worry about it for now; maybe just throw up a basic "Loading..." frame and then implement a full progress bar later when the game is more solid. I've also seen some good implementations with multithreading.
The other two will come with experience; I think what you need more of is a general tutorial for game development than the specific answers. You should definitely start smaller. Once you understand the structure and problems of a smaller game, it will be easier to apply those to larger games.
Most reasonable game programming books will go over basic game structure; I like Game Coding Complete but it's quite complicated for a beginner (it covers more complex ways to approach large projects). Game Architecture and Design is similar, but might be better suited to what you're looking for since it also covers some minor project management "best practices."
There's a lot of different ways to do UI, from using the Java primitive UI types (depending on what other libraries you're using) to self-writing your own "HUD" implementation with just what you need.
One thing that is probably worth being familiar with is multi-threading. For example, with the "load screen", you might have one thread displaying the progress bar, while other threads simultaneously load data.
Similarly, much of the interactivity will happen by dispatching multiple threads to do many things at the same time (move different characters on the screen) or to handle interrupts by user input.
I would look at O'Reilly Books
Killer Game Programming in Java
Coding For Fun
The easiest route is to package everything in a Jar file. The default screen does show a progress bar with some small ability to customise. You can write custom code to keep track, manage and download files but I would personally advise against this route. If you search for applet loaders you will find more information.
I'm not 100% sure what you mean, but you can use Swing components in the same way you can use them in applications. Use a JButton with an image is quite trivial, then hook the event code in the actionPerformed method.
The biggest problem you will probably come across is animation and the EDT. I asked about this earlier.
This page has a whole bunch of useful links for game development. Pulp core is an open source framework worth checking out - even if you don't use the framework you can investigate the code.
Whether you should use Java applets or not seems out of scope of this question, but a lot of the above answers give objective (or no reasons at all) about whether to use Java applets. If it's a game for a personal exercise to learn Java then it's a great approach. If you wish to make it public you need to consider whether the current adoption levels are high enough for your needs.
Things have changed in the applet world recently. Since 1.6 update 10 it is much more competitive with Flash - the download size is smaller (at typically under 4Mb), the startup time is reduced and a new scaling look and feel was introduced.
http://nehe.gamedev.net/
Has tons of tutorials from basic to advanced using OpenGL from various languages and systems, from C to C# and Python to Java.
I found this very very useful and is a great resource to bookmark.
It should get you started with the basics of game/3D programming on your platform/language of choice.
good luck!
As others have said a java applet is probably not the medium in which you want to present your game. The second most people see an applet start to load they run in fear.
If you have no game design experience at all, you are setting the bar very high with a multiplayer RPG. You may want to start with a simpler project such as recreating something like tetris, pacman, or even pong so you can get an idea of all that goes into creating a game.
Flash is great if you are set on doing an online game, Java plus the Jogl open-gl wrappers are also a great option.
Personally I would suggest using Microsoft Xna. C# is similar enough to Java that you should be able to pick it up fairly quickly and Xna does a good job of abstracting away some of the lower level details involved in graphics programming. The community is also very active and helpful.
Don't go the java applet route.
If you want to make a quick casual game, make it Flash; else, develop a full-blown java app and run it via java web start.
Try Tower to see how far you can go using java as a platform, and also how Java Web Start works.
As for progress-bar thing, I recommend you to implement those files' loading and actually use them before you go for progress-bar eye-candy :)
Responding to question 3. If you are really keen to start developing games, I recommend you read this short article by Jeff Howland.
Q1 = 1% of your time.
Q2 = 40% of your time.
Q3 = 59% of your time.
You might as well be asking how could I write Halo3 for a web browser.
Aim a little lower to start with....
opencourseware class on game theory and design. It also comes with a reading list.
Since I can't add a comment, I'll add an answer. I'd recommend NOT using threads for moving characters, etc. For file loading, playing sounds, and even server side networking it works well but any sort of AI/Graphics, it tends to cause endless headaches.
Also, as other people have mentioned, Java may not be the best choice. The simple reason that people say this is that it's difficult. The amount of support/tutorials for Java game development is rather lacking when compared to C++ (GameDev.net is great resource, but it's moved its focus more towards professional development over time), Flash (decent tutorial on Kongregate as well as being a spot that could host the game for you), or even C# (using XNA). Java game development has improved over the years and you can do some amazing things with it, but it may end up being more difficult going that route. No matter which language you choose, definitely look into using a graphics/game engine as it will cut down on the amount of work on your end and allow you to hopefully create a better game.
If you're looking for a gentle entry path to game development and don't mind learning C#, XNA has an entire community built around learning how to write games. It has samples of all sorts of things -- the Game State Management sample shows a good way to implement screen transitions and loading, and there's a similar Network Game State Management sample for networking stuff. They also have entire games available as source code to download.
Whether or not your long-term goals are to make a Java applet, Flash game, or you want to work in C/C++ with OpenGL, XNA is a great way to ignore all the hairy implementation details of rendering/sound/etc. That's very very helpful when you're just trying to make your first game, believe me.
Don't go with the java applet approach; applets are generally annoying, and slow.
If you really want it to be playable from the browser, consider Flash (actionscript), or maybe silverlight (I don't know much about games in silverlight).
Gamedev is a great resource for general game programming stuff
And ...
I'll shamelessly steal the link from yx's answer:
MIT Open Course Ware: Game Theory and Mechanism Design
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 7 years ago.
Improve this question
I understand that C++ is generally considered better than Java for games (at least larger-scale games). Why is this?
What is keeping Java from being competitive in this field? Which reasons against using Java for game programming have the most basis, and which ones are myths?
EDIT: Am a bit unfamiliar with C/C++, and did not think to differentiate between the two at line 1 >.<
The reason Java (and C#/.NET) is not a viable option for AAA titles at this point is the established game engines and their toolchains are written in C++. Game development is all about getting a title on the market in the shortest amount of time, and budgets don't allow for piddling in things like a new language/engine when several are already available, work well, and have an extensive set of editors and tools backing them.
Moving to Java (or C#) would also require a new performance-driven JVM (or CLI) across the big-3 (PC, X360, PS3) or big-5 (add Wii, iPhone). It's technically doable, but not financially viable.
Edit: Anyone with low-level knowledge of both virtual machines and the current state of game engines can tell you that a JVM or CLI could unquestionably be implemented with a new game engine to beat the performance of current C++ engines. The preventing factor is time and money, nothing more and nothing less.
High performance and the inertia of C and C++ traditionally being used for games.
Choosing based on performance isn't that big a priority unless you are making a 3D extravaganza.
Because:
Java is not compiled to native code, meaning that there is a performance hit the first time the code is run.
Java does not give you a predictable memory model (console games need this)
Java does not give you deterministic object finalization.
Java is not as close to the hardware as C is, an essential for a lot of professional 3D game programming.
Console programmers likely don't have a JVM that runs on the PS3, X-Box, etc.
Runtime performance penalties.
You will never be able to squeeze as much performance out of a Java app as you can with a C++ app.
There are probably more reasons, like the fact that they are using pre-existing code that was written in C or C++.
EDIT: As an aside, I don't think that many modern games are written in C. OOP lends itself to game development, and C++ is the de facto language of choice.
Also, I won't add it to my list, but as others have mentioned there is a lot of pre-existing code that works very well that is used in the game industry. It would not be practical to rewrite all/most of your tools just to switch to a new language especially when that switch could cause you a lot of headaches.
I would say, despite the other answers pointing to a lack in speed caused mainly by the JVM, that the real reason people don't code games in Java is the lack of support for environments such as DirectX and OpenGL (which actually remove the need for your code to be close to he hardware as it was suggested by some answers). They are the base frameworks that people generally use to code games, especially nowadays with 3D games being everywhere - and lack of support for them is the reason why Java is not considered as a language for game development.
To emphasize my point, I would suggest you take a look at Microsoft's XNA which is currently optimized for coding in C# via the .NET framework (which like Java is Just-In-Time-Compiled and doesn't run natively per se). The XNA framework interfaces with DirectX which talks to the hardware and so it is very fast.
EDIT
#Ed Swangren's comment made me realize yet another distinction between .NET vs Java when considered for game development. I think another strong point to .NET is that if you do need to be able to squeeze out that last bit of performance and do some pointer math or implement a sophisticated high-performance algorithm it's a lot easier thanks to the unsafe mode. Of course you can even go beyond that and write native libraries to be used by your C# code which is made pretty simple thanks to P/Invoke.
Leaving C(++) aside for the moment... I am inclined to say much of the reason is that Java lacks anything like XNA. What advantages does Java actually have over a language such as C++ when it comes to game development? You have to consider that several of it's typical advantages disappear for the specific area of game development, while C++ gains several.
XNA is what made C# a highly popular language for amateur game development, and contrary to common belief, a quite viable option for commercial development too. C#/.NET being a parallel to Java in so many ways (and arguably a better framework nowadays), when people now have the option for game development with a higher-level language, C# would seem like the much more appealing one, unless cross-platform support is essential (then again, we have Mono and OpenGL for .NET).
C (or rather, C++) has long been the language of game development due to their low-level nature (thus performance benefits) and the host of graphics frameworks (DirectX, OpenGL) and engines that primarily target them. It's usage is embedded in game development and been used virtually since the inception of the industry - and won't disappear any time soon, I suspect.
Java doesn't have as controllable performance
Its highly reversable so harder to protect
Many Games employ scripting languages like lua or python to get "higher" level programming
the API of most systems is C oriented.
Java can be used for back end server systems that games connect to
Flash games seems to of taken the niche Java games could have had.
Java can be optimized to be very fast, as is evident by an interview I had recently with a high-frequency trading company, where they do use Java, as well as C++.
Java has OpenGL bindings, as others have pointed out, so getting to the hardware isn't such a problem, esp since not all games need that, some commercial games have been written for Java3D.
You can use Scala or F# if you want some more performance for multi-threaded or numerically intensive operations, and just tie those in with the GUI.
But, as others have mentioned, the tools that are used tend to be written for C++, and some companies feel more comfortable doing some optimizations in assembly, but, given the fact that the new cpus are very complicated, with multicores, it is unlikely you will get any performance increase over the optimizations from the compilers, but, as long as companies feel these optimizations are still needed, they will stay with C++.
If some developers wanted to write commercial-grade tools for Java or .NET, there could be a market opportunity there, but it will be a great deal of work to make it as good as what is already out there.
I think C or C++ is a better language for building many types of games because it is closer to the hardware and likely to be the first language implemented on any new hardware. Not only that the libraries for accessing many of the advanced features of today's hardware are likely to be implemented in C.
Your typically general purpose higher level language has no easy way to access features of the hardware unless it uses some type of binding layer to call the libraries which are written in C.
For instance how do you write code to access a GPU, or write a custom Shader, or write code that run well on a Cell chip, or run on an Iphone, or on a Blackberry in a high level language. Even when these things are supported, they come out well after other people are able to write games in C that use these features in games.
One compromise you can make is to use a higher level language like Java for most things and C where its needed. You will limit the types of platforms you support though.
Java might also be good for client/server games where the server is written in Java.
It's very hard to write a program that runs in constant memory without garbage collection in Java.
If you want to read statements from game studios who are happy to use java for serious game development then search on twitter for #javaforgames
https://twitter.com/search?q=javaforgames
By studying the Java developers tweets conclude that Java is good for game development if you use it in combination with a high level game engine API such as libgdx and jmonkeyengine that both ease development effort and makes the game tuned to take advantage of the OpenGL/ES hardware acceleration provided by modern GPU hardware.
Java do allow cross-platform goals using said engines on mobile, console and desktop.
Some game developers prefer java for the design phase of the game to engineer in-house tooling.
Java game studios and developers who seek the best performance can archive it by tightly controlling OpenGL by using JOGL or LWJGL directly. Also performance and flexibility archive this by writing their own engine, if time and money allows it.
I wouldn't say that it is always true that C is better than Java for game programming.
For example, if you want to write a game client which can be hosted in a web page, Java could be better.
When writing games which produce amazing 3-D graphics (e.g., Halo on the X-box), usually there is barely enough computing speed to generate all the pixels for each frame. In this kind of game, C would be preferable to Java because it allows the programmer to write faster programs -- at a large expense in terms of difficulty in extracting that speed.
I think it's mostly because c lets developers squeeze every last little bit of performance out of hardware, whereas Java doesn't, it's not low level enough for things like high end 3d video renderers. Basically c lets you squeeze out a couple more frames per second in your next gen shooter.
Inertia mainly... Although it's a bit old now, you can look at Jake2 (which is a pure Java port of Quake 2 with jogl as the openGL lib).
It can perform up to 85% as fast as the C++ original, which means it's fine for most games; especially modern ones which are more social and game play based rather than the limited "hard core" games.
I'd also suggest that most of the answers you get here are coming from a gaming geek/"I want the coolest hardware and games". For [these] hard core 3D games and gamers, that final 15% is hugely important, as that's what separates the $150 graphics card from the $500 one they just bought.
As John Carmack is reported to have said (something along the lines of): "If I were to enter games programming now, I'd program for the iPhone". e.g. it's not nearly as much fun to make the fastest game with the best 3d engine as it is to make the best game.
The console game market is about 10X larger (financially) than the PC game market. For current generation consoles, all the major programming is in C or C++ with some little bits of intrinsics or assembler. If you want to be a professional Game Programmer and work at any major game company, then learn C and C++.
More to the point, all the commercial cross platform engines that are that starting points for many commercial games are in C or C++ right now as well.
You may be able to work on small games in java and flash or even casual Windows games in C#. You can even code C# for XNA games but if you want to make REAL (pressed on DVD) XBOX 360 games you need to learn C++.
Speaking as a game developer at WBGames Chicago. We would not hire a programmer who didn't have strong C++ skills. The same is true at every other game studio that I know of.
The short answer is because almost no gaming platforms support Java, or if they do there is a much better alternative.
On the game consoles, seasoned game developers are using C because that's what they've been doing for 20 years. There's no compelling reason for Java games to exist there.
On the desktop, there are a few Java games, but they can't compete with the market leaders such as World of Warcraft.
On the web, Applets are long dead, and if you're going to make a casual game it's going to be in Flash.
In the mobile space, JME has had it's time in the sun but it's on its way to obsolescence in the face of far superior platforms such as the iPhone.
The perfect world scenario for Java to be a heavy hitter in the gaming world would be for Sun/Oracle to deliver a gaming platform like the Wii or the iPhone - a must-have device that everybody owns. Then, there is a good reason for developers to invest time into experimenting with new technology. But given Sun and Oracle's apparent lack of interest in the consumer market, the chances of them making a gaming rig are slim to none.
It's not a technical issue but purely a "client optics" issue for this.
Good Java code can run just as fast as good C++ code. It can access OpenGL just as fast as native C++ via projects like LWJGL and JMonkeyEngine.
The real reason is that Sun never put any effort into highlighting Java for games and they've never made any deployment particularly easy for Java games. It's a LOT of work and effort of the developer / team to create a seamless / smooth installation (and launch?) experience for the player.
Customers just don't "trust" Java games and bail during the install process when they see the logo.
Although old, no one here seems to point out
All the games and apps on Android are programmed in Java
Because not only is Android written in Java, so are most all of its apps/games/etc...
I will answer in a very different way.
If Java can nevre write a device driver like C, how can you Imagine that it would be faster than the language that the driver is already written in ?