Hi IT,
I’m not familiar with V-Ray.
https://docs.chaos.com/display/VMAYA/Subdivision+and+Displacement+Quality
You ask a general question, and I believe that is not easy to answer.
I believe that you might get as many different replies if you only ask that question often enough. What is really needed and when, while using perhaps optimizing tools based on the camera view, all of that is needed to be known. That is a skill, not a fixed answer, to operate from in general.
It might be faster if you have a model set in polygons and no other “generator” producing anything. Still, since it is static, it might have too many polygons for most camera settings, slowing down everything. On the other hand, if the renderer has to produce everything all the time newly, that might take a while to find the optimized configuration for a model. My ideal scenario is to start with a good model and don’t rely on SDS too much
There are other options, and if the object tends to be in strong motion blur, why bother?
Set up a scene with some standards of your projects and have a High Polygon object in it, then and low polygon with a fixed Subdivision, and one with a camera view-dependent Subdivision. That will tell you more about your complete setup then anyone can tell you.
All the best