Visible to Intel only — GUID: GUID-F9C3A4A2-BF1E-4488-B480-82592D0BC28E
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
Parallel Reduction for rvalues
Description
API
Example
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-F9C3A4A2-BF1E-4488-B480-82592D0BC28E
Parallel Reduction for rvalues
Description
oneAPI Threading Building Blocks (oneTBB) implementation extends the ParallelReduceFunc and ParallelReduceReduction to optimize operating with rvalues using functional form of tbb::parallel_reduce and tbb::parallel_deterministic_reduce algorithms.
API
Header
#include <oneapi/tbb/parallel_reduce.h>
ParallelReduceFunc Requirements: Pseudo-Signature, Semantics
ValueFunc::operator()(constRange&range, Value&&x)const
or
ValueFunc::operator()(constRange&range, constValue&x)const
Accumulates the result for a subrange, starting with initial value x. The Range type must meet the Range requirements <https://spec.oneapi.io/versions/latest/elements/oneTBB/source/named_requirements/algorithms/range.html>_. The Value type must be the same as a corresponding template parameter for the parallel_reduce algorithm.
If both rvalue and lvalue forms are provided, the rvalue is preferred.
ParallelReduceReduction Requirements: Pseudo-Signature, Semantics
ValueReduction::operator()(Value&&x, Value&&y)const
or
ValueReduction::operator()(constValue&x, constValue&y)const
Combines the x and y results. The Value type must be the same as a corresponding template parameter for the parallel_reduce algorithm.
If both rvalue and lvalue forms are provided, the rvalue is preferred.
Example
// C++17
#include <oneapi/tbb/parallel_reduce.h>
#include <oneapi/tbb/blocked_range.h>
#include <vector>
#include <set>
int main() {
std::vector<std::set<int>> sets = ...;
oneapi::tbb::parallel_reduce(oneapi::tbb::blocked_range<size_t>(0, sets.size()),
std::set<int>{}, // identity element - empty set
[&](const oneapi::tbb::blocked_range<size_t>& range, std::set<int>&& value) {
for (size_t i = range.begin(); i < range.end(); ++i) {
// Having value as a non-const rvalue reference allows to efficiently
// transfer nodes from sets[i] without copying/moving the data
value.merge(std::move(sets[i]));
}
return value;
},
[&](std::set<int>&& x, std::set<int>&& y) {
x.merge(std::move(y));
return x;
}
);
}
See also