What overhead is involved in 2 threads accessing an object that is created in a 3rd thread - each thread has a different CPU affinity (c++ and linux)?

Question

Suppose thread A creates an Object (Oa) to be used by two other threads. thread B writes to Oa and thread C reades from Oa. Threads A,B, and C are locked into CPUs 1,2,3 respectively using cpu affinity.

What overhead/activities are involved with executing the methods in Oa since it was created in on one CPU but accessed by two other CPUs?

This question pertains more about executing the instruction vs. accessing the data (which is pinned)

Answer 1

I think you are asking about Cache Coherence, which is a CPU specific. Basically, each CPU would have a cached value of that data. Any changes to the data would be propagated to the other CPUs.