I am trying to do a simulation in blender in order to place some objects on a ground. During baking or simulation in real time everything is super slow (less then 1 fps) but CPU usage (6700K i7) is only ca. 30%. GPU 0%. I guess all simulations are calculated by CPU but why Bleder does not use it at 100%?

  • $\begingroup$ Perhaps the simulation is relatively easy to calculate, however it produces lots of data that needs to be written to disk. This could result in an I/O bound task, where the writing speed to disk is the bottleneck. That is complete speculation though. Another reason could be that the implementation uses fewer threads than there are CPU cores $\endgroup$ – Robert Gützkow Nov 28 '19 at 0:30

The simulation is likely only running on one core (with another core being partly used for scene updates).

Many (particularly older) physics simulators are single threaded, or at least some steps are. Multi threading in physics simulators is very hard due to all the possible inter-dependencies between objects, see also: Is Physics calculation multithreading possible?

Blender's rigid body engine is Bullet, which does support some multithreading, but I don't think that part has been integrated with Blender, and from what I hear, still has a fair bit of overhead.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.