Preceding example 2 clearly illustrated this point: when running a multithreaded app on modern multicore hardware (high-end enterprise class servers can have in excess of 700 CPU cores!), the underlying OS will take care of optimally scheduling threads onto available CPU cores; the app developers need not concern themselves with this. Effectively, the Linux kernel will try and ensure perfect SMP scalability whenever possible, which will result in higher throughput and, ultimately, speed gains. (Again, dear reader, we're being optimistic here: the reality is that with heavy parallelism and CPU cores also comes the heavy downsides of concurrency concerns; we shall discuss all of this in more detail in upcoming chapters.)