Earlier in this book, we explored various ways to express parallelism with TBB, such as algorithms, flow graphs, and task groups. In those chapters, we highlighted how TBB delivers strong performance across a range of systems, regardless of core count or memory size. In Chapter 10 , we discussed how the combination of TBB’s global thread pool, arbitrator, arenas, and task scheduler contributes to its composability, allowing it to effectively adapt to typical conditions.