Fine-grained locking
Fine-grained locking is a synchronization technique that protects smaller portions of shared state with separate locks, allowing independent portions to be accessed concurrently.
C++ Concurrency in Action, Second Edition
Fine-grained locking protects smaller portions of shared data with separate locks rather than one broad lock.
- How it is used
- Uses separate head and tail locks to increase concurrency beyond a queue protected by one mutex.
- How deep it goes
- Works through a linked-queue baseline, dummy-node redesign, two-mutex implementation, and condition-variable waiting.
- Examples
- Compares single-mutex and linked queues, then uses push, try_pop, and wait_and_pop implementations to expose lock boundaries.
- What you can do after
- Build it from scratch as a C++ queue whose producers and consumers can work at opposite ends under separate locks.
Read how C++ Concurrency in Action, Second Edition explains fine-grained locking →
Latency
Fine-grained locking is a synchronization technique that protects smaller portions of shared work or state with more numerous locks.
- How it is used
- Uses smaller lock scopes to increase concurrency and reduce contention in thread-based low-latency systems.
- How deep it goes
- Explains the coarse-versus-fine trade-off and convoying mitigation, but stops short of implementation details or correctness techniques.
- Examples
- Uses thread-based concurrency, kernel scheduling, context switching, and lock convoying as practical performance contexts.
- What you can do after
- Apply it to reduce lock contention and critical-section duration, while recognizing added complexity and synchronization overhead.
Related concepts
- ConcurrentDictionary<TKey,TValue>used by · ConcurrentDictionary<TKey,TValue> is a .NET thread-safe generic key-value collection that supports concurrent access and atomic conditional updates.
- concurrent collectionused by · A concurrent collection is a collection designed to support safe, defined access or modification by multiple concurrent computations.
- concurrencyproduces · Concurrency is the structuring or execution of multiple computational activities so their progress overlaps in time, whether or not they run simultaneously.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.