Question Hello, I've noticed in the prims_simple.h protocol that __threadfence_system is called by non-worker threads to ensure memory ordering. However, according to the CUDA documentation, ...