Modern C++ concurrency stops being intimidating when you understand what the language is actually doing.
Writing fast multithreaded software is not simply a matter of starting more threads. Real performance comes from understanding how threads communicate, how shared data is protected, when atomics make sense, where locks become bottlenecks, and how modern C++ helps you coordinate work safely.
Modern C++ Concurrency gives C++ developers a practical path from the fundamentals of multithreading to the techniques used in high-performance concurrent applications.
Instead of burying you in theory, Adrian Colebrook explains concurrency through clear examples, practical design patterns, and real-world situations that show not only how each tool works, but when you should use it.
Inside you'll learn how to:
- Create and manage threads with std:: thread and std:: jthread
- Prevent race conditions, deadlocks, and unsafe shared state
- Use mutexes, scoped locks, condition variables, futures, and promises
- Build cleaner asynchronous workflows
- Work with C++20 semaphores, latches, barriers, and stop tokens
- Understand atomics and the C++ memory model without drowning in theory
- Use acquire/release semantics and memory ordering correctly
- Design thread-safe queues, stacks, caches, and shared resources
- Understand lock-free programming and compare-and-exchange operations
- Build thread pools, worker queues, and task-based systems
- Use parallel algorithms and execution policies
- Reduce contention, false sharing, and unnecessary synchronization
- Diagnose race conditions, deadlocks, starvation, and timing-dependent bugs
- Design concurrent software that remains maintainable as it grows
Whether you're moving beyond basic C++, modernizing an existing application, preparing for systems-level development, or trying to understand why your multithreaded code behaves unpredictably, this book gives you the tools to build concurrent software with greater confidence.
Stop treating concurrency like black magic. Learn how the pieces fit together-and start writing modern C++ that can actually use the hardware beneath it.