When I was porting the mbw to the OSX, I found that sched_getaffinity pthread_setaffinity_np are missing, so that I could not bind a thread to a core. The result is the mbw threads could be running on the same core and compete for per-core resources, although the OSX kernel scheduler is unlikly to do so. I simply simulate these calls by using native OSX calls. The first thing to do is to detect th