Visible to Intel only — GUID: GUID-AA1C47E6-A062-4D76-A600-727D215B70B0
Package Contents
Parallelizing Simple Loops
Parallelizing Complex Loops
Parallelizing Data Flow and Dependence Graphs
Work Isolation
Exceptions and Cancellation
Containers
Mutual Exclusion
Timing
Memory Allocation
The Task Scheduler
Design Patterns
Migrating from Threading Building Blocks (TBB)
Constrained APIs
Invoke a Callable Object
Appendix A Costs of Time Slicing
Appendix B Mixing With Other Threading Packages
References
parallel_for_each Body semantics and requirements
parallel_sort ranges interface extension
TBB_malloc_replacement_log Function
Type-specified message keys for join_node
Scalable Memory Pools
Helper Functions for Expressing Graphs
concurrent_lru_cache
task_group extensions
The customizing mutex type for concurrent_hash_map
Visible to Intel only — GUID: GUID-AA1C47E6-A062-4D76-A600-727D215B70B0
Concurrent Queue Classes
Template class concurrent_queue<T,Alloc> implements a concurrent queue with values of type T. Multiple threads may simultaneously push and pop elements from the queue. The queue is unbounded and has no blocking operations. The fundamental operations on it are push and try_pop. The push operation works just like push for a std::queue. The operation try_pop pops an item if it is available. The check and popping have to be done in a single operation for sake of thread safety.
For example, consider the following serial code:
extern std::queue<T> MySerialQueue;
T item;
if( !MySerialQueue.empty() ) {
item = MySerialQueue.front();
MySerialQueue.pop_front();
... process item...
}Even if each std::queue method were implemented in a thread-safe manner, the composition of those methods as shown in the example would not be thread safe if there were other threads also popping from the same queue. For example, MySerialQueue.empty() might return true just before another thread snatches the last item from MySerialQueue.
The equivalent thread-safe oneAPI Threading Building Blocks (oneTBB) code is:
extern concurrent_queue<T> MyQueue;
T item;
if( MyQueue.try_pop(item) ) {
...process item...
}In a single-threaded program, a queue is a first-in first-out structure. But if multiple threads are pushing and popping concurrently, the definition of “first” is uncertain. Use of concurrent_queue guarantees that if a thread pushes two values, and another thread pops those two values, they will be popped in the same order that they were pushed.
Template class concurrent_queue is unbounded and has no methods that wait. It is up to the user to provide synchronization to avoid overflow, or to wait for the queue to become non-empty. Typically this is appropriate when the synchronization has to be done at a higher level.
Template class concurrent_bounded_queue<T,Alloc> is a variant that adds blocking operations and the ability to specify a capacity. The methods of particular interest on it are:
pop(item) waits until it can succeed.
push(item) waits until it can succeed without exceeding the queue’s capacity.
try_push(item) pushes item only if it would not exceed the queue’s capacity.
size() returns a signed integer.
The value of concurrent_queue::size() is defined as the number of push operations started minus the number of pop operations started. If pops outnumber pushes, size() becomes negative. For example, if a concurrent_queue is empty, and there are n pending pop operations, size() returns -n. This provides an easy way for producers to know how many consumers are waiting on the queue. Method empty() is defined to be true if and only if size() is not positive.
By default, a concurrent_bounded_queue is unbounded. It may hold any number of values, until memory runs out. It can be bounded by setting the queue capacity with method set_capacity.Setting the capacity causes push to block until there is room in the queue. Bounded queues are slower than unbounded queues, so if there is a constraint elsewhere in your program that prevents the queue from becoming too large, it is better not to set the capacity. If you do not need the bounds or the blocking pop, consider using concurrent_queue instead.