How does Iray works?
I do not quite understand how Iray works, in particular, the connection between complexity of the scene and number of samples.
As I understand it, Iray shells objects with samples getting the result. More samples - longer render.
I worked with a 3D scanner, There is the same principle: more rays - longer scan, but more accurate result.
Suppose. I have a small cube on the stage and 15,000 samples. Does this mean that Iray will render it as much as a complex scene, until the entire rays/samples is fired. Does this mean that Iray will render it as much as a complex scene, until the entire sample ammunition is fired?
Although, in this case, the samples fly away into the empty space, where there is nothing to calculate, then the speed of fire increases.
Specialists in the render, explain, please, how it works.

Comments
Not samples but iterations. Iray is a simulation of the physical behavior of light. It will iterate the algoriyhms until it meets one of the completion criteria (e.g. time, iterations, convergnece).
Their is significantly more information on Nvidia's website.
The mark-one eyeball is the best method of deciding if the render is done. If you like it, stop it.