I was just wondering why c++ ist that powerful and performant for developing games. I wrote a lot of games in c# and delphi, always using the timer component to make objects "move". Another option for the movement were loops, but they are definitely not performant.
So what technique does c++ use that users can develop performant games?
To a large extent this is based on two things:
For many years, C++ was the only* choice for high-performance applications of any kind, including games. As a result, there were significant investments into game engines and libraries that were also written in C++.
Even today, most frameworks for gaming are written in C++. XNA does allow C#, but even Microsoft keep pushing C++ as their language of choice.
*yeah, it's subjective...
nothing stops you to use a timer to write your c++ games.
You can look into C++ Boost thread::sleep method to do that.
C++ give you finer grain of control over the actual hardware and bit pushing. For common business needs, a third generation language such as Java or C# is quicker to program and takes worries like pointers and garbage collection off the hands of the developer. This is at a cost of lack of ability to optimize and fine tune how memory and data structures are used.
You can also take a hybrid approach of breaking the game into a higher level language for scripting etc and then "drop" into C++ land for parts that require the speed and optimizations.
OpenGL and DirectX are both very compatible with C++. Add in the fact that you have more fine grained control over memory, memory allocation, and you can use just enough features of C++ to make development "easier" than straight C to get some of the reasons.
So what technique does c++ use that users can develop performant games?
None that are not (also) found in other languages.
The advantage of C++ over other languages (subjective to a degree) is not in the techniques used.
It is a combination of a few factors, making it both low-level enough and high-level enough for what is needed:
it is deterministic (you can determine what gets executed when). This is not true for garbage collected languages and it is not (or partially) true for languages run in virtual machines.
it offers high performance (you can get C-like performance and even drop in some assembly code if you feel like it).
it offers enough abstraction to be higher level than other fast languages (like C for example).
In the end it is something close to both low-level languages and high-level languages.
From what I've always understood, and the rule applies to sports as well.
Speed kills.
All about speed.
There is a solid codebase for games, and it's mostly in C++. Most game engines are implemented in C++ and that's because as someone previously mentioned also OpenGL and DirectX are implemented in C++. Further on, games always push the limits of hardware and that's why the code needs to be as fast and optimal as possible..
On the opposite side there are the office/business applications that simply need faster development cycles, stability and to be as extensible as possible.
The strength of C++ when it comes to game development is the ability to exactly layout the data-structures that your software will use. When performant real-time systems (such as games) started growing, it was the most commonly supported and most developed programming language. As such it was a natural choice for games that try to push platforms to their limit. This has in term lead most developers to go along with this choice even if they need it or not.
C++ provides the ability to override important performance bottlenecks such as memory allocation. It has the ability to structure and place things exactly where they want in the memory. On top of this it's a flexible programming language that provides a decent development velocity.
C++ has the combination of efficiency and abstraction that makes it still the only practical choice for large-scale systems that require efficient responses[1].
Hardware interface layers are, as a general rule, in C or C++ as well, meaning a foreign language interface does not have to be set up to talk to the hardware abstraction layers.
Further, it has the advantage of a large community experienced in writing games, as well as numerous frameworks.
On a personal note, I have used games written in other languages, and they have consistently been slower.
[1] Ada or Delphi may also be usable in this area, but their popularity level means it's not as pragmatic to choose them unless you have a niche need.
Games is performance critical software that requires 100% usage of the hardware user has, and C++ is only popular language that gives you such abilities:
For this reason you will find C++ not only in game development but also in many other areas that require from you best performance, lowest resource usage and good ability to optimize like, for example, data bases, system programming and many other areas.
Some other aspects that I think weren't mentioned:
As machines become more powerful and more of the workload is shifted to hardware (graphics and even physics, in some cases), there are more system resources available to perform other tasks. As such, even though C++ is still the industry standard, the down-to-the-metal optimizations it enables are no longer the main reason it's used. The main reason it's used is that the majority of APIs for game development are C++ APIs. You rarely want to write your own engine and even if you write your own engine, you're likely reusing standard low-level components, such as NVidia PhysX, DirectX, OpenGL, etc. Some of these now have Managed code interfaces, but the core APIs are for C++, since it is so commonly used.
Another trend worth noting is that a lot of games these days use scripting languages, such as Python or Lua, to control much of the game logic. This gives your content developers and modders flexibility to provide rich behaviors without having to code in C++ or build. Scripting is not super high-performance, but it's extremely convenient and modern PCs have plenty of horsepower to handle the scripting.
There are a number of reasons here.
Portability
Does one really think that Sony will use C# in one of their consoles? Will Nintendo deploy a JVM on the Wii2?
No.
What if Microsoft goes C# with the next Xbox? Guess what, they won't. That would double the Codebase for all 3rd Party Developers, rendering Microsofts new console much less attractive to develop for.
So C++ is the obvious choice here because one just needs the compiler for the appropriate architecture.
Performance
Also the core of 3D engines will most likely always be written in C/C++ with part of it in ASM, simply due to the fact that you want to squeeze out every bit of performance on Consoles where your environment is basically hammered in stone.
Development Time
Before all the licensable engines came up, there was a time when the fist generation of software titles took extremely long to develop, just because you had to write a whole new engine for your Game on the new Platform, which takes time. Taken into account that the development of a big game already takes years, this costs lots of money. So it was always easier to take existing code and "upgrade" it for the next generation. Then over time you would gradually improve it. Bearing that in mind, it's easy to see that none of the console manufacturers wanted to make their developers face even longer development times by switching their console to something different than C/C++.
So to wrap it all up. Portability, performance and legacy code bases are the reasons that C/C++ is still THE major player in game development.
The only technique that C++ has over most other high level languages is that "you don't pay for what you don't use". The language doesn't have features behind the scenes like garbage collection, reference counting, synchronisation, reflection, introspection, bounds-checking, run-time type checking, meta-classes, and so on, each of which can impose a cost on general operation.
Of course, most of those features are very useful, so in C++ you have to spend a lot more coder time to get the same results. Whether it's worth the effort or not tends to come down to a matter of how much legacy code you have to work with.
I'd like to point out a few things that people are missing. Firstly C++ has it's dominance in the industry because most studios have an existing code C++ 3D engine. Secondly they built it or have been working with it for years. Just those two points alone make it reasonable to stay with C++. Thirdly native C++ will always be slightly faster than managed languages because it doesn't have the overhead of being.. well, managed.
Now be aware however this is for existing game studios, the indie scene is a little difference and I think that here is where XNA is actually picking up speed. Firstly it lets gives you access to three devices (pc, xbox, zune based) and that's not even including mono xna. The speed differences are almost never worth worrying about at this level and it's just more productive.
I hope I've given you something to consider.