Hellooo

Was wondering if there was a better way of caching large crowds in houdini while working and played around with only caching some of the intrinsic attributes and reapplying them on the rest crowd geometry. It seems to mostly work and is pretty light only caching the agentclipname, agentcliptime, and agentclipweight, orient, P, however when an agent has foot locking im not sure where that info is located and how it is applied to the agents.
I see that there are point attributes with agentfoot_locked, agentfoot_lock_pos, agentterrainadaption_hipshift, etc. but seems to do nothing when copied from the sim geo to the rest geo. Is that a deeper thing going on to control the footlocking on the agents? is there any attributes that can drive the foot locking of agents? I know that there is the CHOP channel nodes that are created when doing the agent prep, is it something to do with that?

Cheers in advance!