As you have learned, when operating upon shared writable data, the critical section must be protected in some manner. Locking is perhaps the most common technology used to effect this protection. It's not all rosy, though, as performance can suffer. To realize why, consider a few analogies to a lock: one would be a funnel, with the stem of the funnel just wide enough to allow one thread at a time to flow through, no more. Another is a single toll booth on a busy highway or a traffic light at a busy intersection. These analogies help us visualize and understand why locking can cause bottlenecks, slowing performance down to a crawl in some drastic cases. Worse, these adverse effects can be multiplied on high-end multicore systems with a few hundred cores; in effect, locking doesn't scale well.Â
Another issue is that of lock contention; how often is a particular lock being acquired? Increasing...
