The performance of the application may vary significantly depending on the characteristics of your system. This chapter describes what are the factors that have an impact in a greater or lesser degree in the performance of the Gaia Sky and explains how to tweak them. It is organised in two parts, namely GPU performance (graphics performance) and CPU performance.
Refer to the Graphics performance chapter.
The CPU also plays an obvious role in updating the scene state (positions, orientations, etc.), managing the input and events, executing the scripts and calling and running the rendering subsystem, which streams all the texturing and geometric information to the GPU for rendering. This section describes what are the elements that can cause a major impact in CPU performance and explains how to tune them.
When multithreading is enabled the work of the update loop will be distributed to a number of threads (usually this number is defined by the number of cores/threads of the main CPU). In our tests, multithreading starts having a positive impact when more than 4 threads are available. Using a CPU which supports 8 threads and enabling multithreading in Gaia Sky results in a significant performance boost. However, in the case of 4 threads or less, our tests yield that the multithreading overhead penalty is larger than the gain, resulting in lower FPS. Usually, with newer CPUs it is a good practice to enable multithreading.
You can modify the magnitude limit by setting the property
in the configuration file. This will prevent the loading of stars whose magnitude
is higher (they are fainter) than the specified magnitude, thus relieving the
CPU of some processing. Also, take a look at the
Data properties section.
Draw distance (levels of detail)¶
These settings apply only when using a catalog with levels of detail
TGAS. We can configure whether we want Smooth transitions between
the levels (fade-outs and fade-ins) and also the draw distance, which is
represented by a range slider. The left knob represents the view angle
above which octants are rendered.
Basically, the slider sets the view angle above which a particular octree node (axis aligned cubic volume) is marked as observed and thus its stars are processed and drawn.
Set the knob to the right to lower the draw distance and increase performance.
Set the knob to the left to higher the draw distance at the expense of performance.
This setting controls whether particles fade in and out depending on the octree view angle. This will prevent pop-ins when using a catalog backed by an octree but it will have a hit on peformance due to the opacity information being sent to the GPU continuously. If smooth transitions are enabled, there is a fade-in between the draw distance angle angle and the draw distance angle + 0.4 rad.