Splitting a large task into smaller parallel jobs

CPUs are not getting any faster, so manufacturers are adding more cores to the processors. That means that single-threaded applications are not able to leverage the "parallelization" offered by a multi-core processor. But how to put those cores to work?

The concept of parallelization is based on the assumption that often large problems can be divided into smaller ones, which are solved "in parallel". The smaller task execution can be spread through several cores to complete the main task faster.

Concurrent programming is not easy, mostly because of synchronization issues and the pitfalls of shared data. Historically Java has offered excellent support for multi-threaded programming, partially shielding the developer from the complexity of writing code that runs many tasks in parallel.

One of the most useful algorithms to successfully leverage multiple cores is "fork/join". The "fork/join" algorithms essentially divide a problem into many smaller...