Visible to Intel only — GUID: GUID-B8161478-0D60-4D6E-9BB6-35C4E175B674
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
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-B8161478-0D60-4D6E-9BB6-35C4E175B674
Cancel a Graph Explicitly
To cancel a graph execution without an exception, you can create the graph using an explicit task_group_context, and then call cancel_group_execution() on that object. This is done in the example below:
task_group_context t;
graph g(t);
function_node< int, int > f1( g, 1, []( int i ) { return i; } );
function_node< int, int > f2( g, 1,
[]( const int i ) -> int {
cout << "Begin " << i << "\n";
spin_for(0.2);
cout << "End " << i << "\n";
return i;
} );
function_node< int, int > f3( g, 1, []( int i ) { return i; } );
make_edge( f1, f2 );
make_edge( f2, f3 );
f1.try_put(1);
f1.try_put(2);
spin_for(0.1);
t.cancel_group_execution();
g.wait_for_all();
When a graph execution is canceled, any node that has already started to execute will execute to completion, but any node that has not started to execute will not start. So in the example above, f2 will print both the Begin and End message for input 1, but will not receive the input 2.
You can also get the task_group_context that a node belongs to from within the node body and use it to cancel the execution of the graph it belongs to:
graph g;
function_node< int, int > f1( g, 1, []( int i ) { return i; } );
function_node< int, int > f2( g, 1,
[]( const int i ) -> int {
cout << "Begin " << i << "\n";
spin_for(0.2);
cout << "End " << i << "\n";
task::self().group()->cancel_group_execution();
return i;
} );
function_node< int, int > f3( g, 1, []( int i ) { return i; } );
make_edge( f1, f2 );
make_edge( f2, f3 );
f1.try_put(1);
f1.try_put(2);
g.wait_for_all();
You can get the task_group_context from a node’s body even if the graph was not explicitly passed one at construction time.