When we start to speed up our programs by introducing concepts such as concurrency and parallelism, we start to face new challenges that must be thought about and addressed appropriately. One of the biggest challenges we start to face is the speed at which we can access data. It's important to note at this stage that if we cannot access data fast enough, then this becomes a bottleneck for our programs, and no matter how expertly we design our systems, we'll never see any performance gains.
Computer designers have been increasingly looking for ways to improve the ease with which we can develop new parallel solutions to problems. One of the ways they have managed to improve things is by providing a single physical address space that all of our multiple cores can access within a processor. This removes a certain amount of complexity away from us, as programmers, and allows us to instead focus on ensuring that our code is thread safe. There are a number of...