Robot executor slows down simulation

I have a simulation with a robot. The robot receives the location for his moves via other components and the instructions are processed in a python script.
At the start, the simulation can run at x10 speed without problems: GPU utilization ±40% CPU 30%

After 1h of simulation time, the performance of the simulation slows down. This is only the case for the robot movements, when the robot is idle, the other actions are fast. The framerate is reduced a lot during robot actions
After 4h of simulation time, it is even impossible to have the simulation running at 1x speed during robot movements.

  • GPU utilization 7% during robot motion, 40% when robot is idle
  • CPU utilization 12%-13% (1core at 100% during robot motion), >20% with robot idle
    In task manager, it is really clear when the robot is idle:

Additional information:
Visual Components 4.6 Premium

Does anyone has an idea what could cause the reduction in performance and how to solve it?
From my analysis it seems like some internal behaviour of the robot executor is single threaded and takes longer to calculate.

I guess in general you could try to figure out and eliminate whether some objects or data accumulate as your simulation progresses. Some examples could be e.g. components on top of each other, geometry like traces, objects in Python scripts like event handlers, statistics dashboard data like robot joint values or other variables related to your robot.

Thanks for the quick reply.
I checked most of the options but could not find the cause:

  • components are removed after processing (max 5 at the same time in a model, no relation between speed and number of products)
  • Traces are not enabled
  • No statistics dashboards active
  • No event handlers linked to products or robot