A Distributed Architecture for Simulation Environments Based on Game Engine Systems - Visual Computing: Scientific Visualization and Imaging Systems - page 42

Fig. 2.1 Simple coupled

model

by feeding the update stage with a time parameter. Figures 2.2 and 2.3 illustrate
these models, respectively.
By using these models, the application has a chance to adjust its execution with
time, so the system can run the same way in different machines. More powerful
machines will be able to run the simulation more smoothly, while less powerful
ones will still be able to provide some experience to the user. This requirement is
also important due the fact that hardware configuration change fast, while the
simulator may have a longer time of life.
Although these are working solutions, time measuring may vary greatly in
different machines due to many reasons (such as process load), making it difficult
do reproduce it faithfully. For example, some training sessions may require a scene
replay feature [ 5 ], which may not be trivial to implement if it is not possible to run
some part of the loop sequence in a deterministic way. Other features as network
module implementation and program debugging [ 5 ] may be easier to implement if
the loop uses a deterministic model. Another issue is that running some simula-
tions too frequently, like AI and the game logic, may not yield better results.