In this chapter, we learned how to use parallel collections to improve program performance. We have seen that sequential operations on large collections can be easily parallelized and learned the difference between parallelizable and non-parallelizable collections. We investigated how mutability and side effects impact correctness and determinism of parallel operations, and saw the importance of using associative operators for parallel operations. Finally, we studied how to implement our custom parallel collection class.

We also found, however, that tuning program performance is tricky. Effects such as memory contention, garbage collection, and dynamic compilation may impact the performance of the program in ways that are hard to predict by looking at the source code. Throughout this section, we urged you to confirm suspicions and claims about program performance by experimentally validating them. Understanding the performance characteristics of your program is the first step toward...