Sponsored Feature: Designing the Framework of a Parallel Game Engine
February 25, 2009 Page 2 of 6
The engine's design focuses on flexibility, allowing for the simple expansion of its functionality. It can be easily modified to accommodate platforms that are constrained by certain factors, such as memory.
The engine is broken up into two distinct pieces: the framework and the managers. The framework contains the parts of the game that are duplicated; that is, there will be multiple instances of them. It also contains items that have to do with execution of the main game loop. The managers are singletons that the game logic depends on.
Figure 3 illustrates the different sections that make up the engine.
Figure 3. The high-level architecture of the engine.
Notice that the game processing functionality, referred to as a system, is treated as a separate entity from the engine. This modularity makes the engine the "glue" for tying all the functionality together. Modularity also allows the systems to be loaded or unloaded as needed.
The interfaces are the means of communication between the engine and the systems: The systems implement the interface so that the engine can access a system's functionality, and the engine implements the interface so that the systems can access the managers.
For more information about this concept, refer to Appendix A.
As described in the Parallel Execution State section, the systems are inherently discrete-systems can run in parallel without interfering with the execution of other systems. This does cause some problems when systems need to communicate with each other, because the data is not guaranteed to be in a stable state. Two reasons for inter-system communication are:
To inform another system of a change made to shared data, such as position or orientation.
To request functionality that is not available within a particular system, such as the AI system asking the geometry or physics system to perform a ray intersection test.
The first communication problem is solved by implementing the state manager mentioned in the previous section. To solve the second problem, a mechanism is included through which a system can provide a service that a different system can use.
The framework ties all the different pieces of the engine together. Engine initialization occurs within the framework, with the exception of the managers, which are globally instantiated. The information about the scene is also stored in the framework. For flexibility, the scene is implemented as what is called a universal scene, which contains universal objects-containers for tying together the different functional parts of a scene.
The game loop is also located within the framework. Figure 4 shows its flow.
Figure 4. The main game loop.
The first step in the game loop is to process all pending OS window messages because the engine operates in a windowed environment. The engine will be unresponsive to the OS if this is not done. The scheduler next issues the systems' tasks with the task manager, and then the changes that the state manager has been keeping track of are distributed to all interested parties. Finally, the framework checks the execution status to see if the engine should quit or perform some other engine execution action, such as go to the next scene. The engine execution status is located in the environment manager, which is described later in this article.
The scheduler holds the master clock for execution, which is set at a pre-determined frequency. The clock can also run at an unlimited rate, for things like benchmarking mode, so that there is no waiting for the clock time to expire before proceeding.
The scheduler submits systems for execution, via the task manager, on a clock tick. For free step mode, the scheduler communicates with the systems to determine how many clock ticks they will need to complete their execution. Then it determines which systems are ready for execution and which ones will be finished by a certain clock tick. The scheduler can adjust this amount if it determines that a system needs more execution time. Lock step mode has all systems start and end on the same clock, so the scheduler will wait for all systems to complete execution.
Universal Scene and Objects
The universal scene and objects are containers for the functionality that is implemented within the systems. By themselves, the universal scene and objects have no functionality other than being able to interact with the engine. They can, however, be extended to include the functionality that is available in a system, which gives them the ability to take on the properties of any available system without having to be tied to a specific system. This is called loose coupling. Loose coupling is important because it allows the systems to be independent of each other and run in parallel.
Figure 5 illustrates the universal scene and object extension of a system.
Figure 5. The universal scene and object extension.
Here's how extensions work: A universal scene is extended to have graphics, physics, and other properties. The graphics scene extension initializes the display and other things, and the physics scene extension sets up the rigid body world, such as gravity. Scenes contain objects, so a universal scene has several universal objects. Similarly, the universal object, contained in the universal scene, is extended to have graphics, physics, and other properties. The graphics object extension draws the object on screen, and the physics object extension is responsible for the rigid body interaction of the object with other rigid bodies.
For a more detailed diagram showing the relationship between the engine and the systems, refer to Appendix B.
The universal scene and universal objects are also responsible for registering their extensions with the state manager, so that the extensions are notified of changes made by other extensions (that is, other systems). For example, the graphics extension would be registered to receive notification of position and orientation changes made by the physics extension.
Page 2 of 6