|
|
/*
Copyright 2005-2014 Intel Corporation. All Rights Reserved.
This file is part of Threading Building Blocks.
Threading Building Blocks is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License version 2 as published by the Free Software Foundation.
Threading Building Blocks is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with Threading Building Blocks; if not, write to the Free Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
As a special exception, you may use this file as part of a free software library without restriction. Specifically, if other files instantiate templates or use macros or inline functions from this file, or you compile this file and link it with other files to produce an executable, this file does not by itself cause the resulting executable to be covered by the GNU General Public License. This exception does not however invalidate any other reasons why the executable file might be covered by the GNU General Public License.*/
#ifndef __TBB_parallel_scan_H
#define __TBB_parallel_scan_H
#include "task.h"
#include "aligned_space.h"
#include <new>
#include "partitioner.h"
namespace tbb {
//! Used to indicate that the initial scan is being performed.
/** @ingroup algorithms */struct pre_scan_tag { static bool is_final_scan() {return false;}};
//! Used to indicate that the final scan is being performed.
/** @ingroup algorithms */struct final_scan_tag { static bool is_final_scan() {return true;}};
//! @cond INTERNAL
namespace internal {
//! Performs final scan for a leaf
/** @ingroup algorithms */ template<typename Range, typename Body> class final_sum: public task { public: Body my_body; private: aligned_space<Range,1> my_range; //! Where to put result of last subrange, or NULL if not last subrange.
Body* my_stuff_last; public: final_sum( Body& body_ ) : my_body(body_,split()) { poison_pointer(my_stuff_last); } ~final_sum() { my_range.begin()->~Range(); } void finish_construction( const Range& range_, Body* stuff_last_ ) { new( my_range.begin() ) Range(range_); my_stuff_last = stuff_last_; } private: /*override*/ task* execute() { my_body( *my_range.begin(), final_scan_tag() ); if( my_stuff_last ) my_stuff_last->assign(my_body); return NULL; } };
//! Split work to be done in the scan.
/** @ingroup algorithms */ template<typename Range, typename Body> class sum_node: public task { typedef final_sum<Range,Body> final_sum_type; public: final_sum_type *my_incoming; final_sum_type *my_body; Body *my_stuff_last; private: final_sum_type *my_left_sum; sum_node *my_left; sum_node *my_right; bool my_left_is_final; Range my_range; sum_node( const Range range_, bool left_is_final_ ) : my_left_sum(NULL), my_left(NULL), my_right(NULL), my_left_is_final(left_is_final_), my_range(range_) { // Poison fields that will be set by second pass.
poison_pointer(my_body); poison_pointer(my_incoming); } task* create_child( const Range& range_, final_sum_type& f, sum_node* n, final_sum_type* incoming_, Body* stuff_last_ ) { if( !n ) { f.recycle_as_child_of( *this ); f.finish_construction( range_, stuff_last_ ); return &f; } else { n->my_body = &f; n->my_incoming = incoming_; n->my_stuff_last = stuff_last_; return n; } } /*override*/ task* execute() { if( my_body ) { if( my_incoming ) my_left_sum->my_body.reverse_join( my_incoming->my_body ); recycle_as_continuation(); sum_node& c = *this; task* b = c.create_child(Range(my_range,split()),*my_left_sum,my_right,my_left_sum,my_stuff_last); task* a = my_left_is_final ? NULL : c.create_child(my_range,*my_body,my_left,my_incoming,NULL); set_ref_count( (a!=NULL)+(b!=NULL) ); my_body = NULL; if( a ) spawn(*b); else a = b; return a; } else { return NULL; } } template<typename Range_,typename Body_,typename Partitioner_> friend class start_scan;
template<typename Range_,typename Body_> friend class finish_scan; };
//! Combine partial results
/** @ingroup algorithms */ template<typename Range, typename Body> class finish_scan: public task { typedef sum_node<Range,Body> sum_node_type; typedef final_sum<Range,Body> final_sum_type; final_sum_type** const my_sum; sum_node_type*& my_return_slot; public: final_sum_type* my_right_zombie; sum_node_type& my_result;
/*override*/ task* execute() { __TBB_ASSERT( my_result.ref_count()==(my_result.my_left!=NULL)+(my_result.my_right!=NULL), NULL ); if( my_result.my_left ) my_result.my_left_is_final = false; if( my_right_zombie && my_sum ) ((*my_sum)->my_body).reverse_join(my_result.my_left_sum->my_body); __TBB_ASSERT( !my_return_slot, NULL ); if( my_right_zombie || my_result.my_right ) { my_return_slot = &my_result; } else { destroy( my_result ); } if( my_right_zombie && !my_sum && !my_result.my_right ) { destroy(*my_right_zombie); my_right_zombie = NULL; } return NULL; }
finish_scan( sum_node_type*& return_slot_, final_sum_type** sum_, sum_node_type& result_ ) : my_sum(sum_), my_return_slot(return_slot_), my_right_zombie(NULL), my_result(result_) { __TBB_ASSERT( !my_return_slot, NULL ); } };
//! Initial task to split the work
/** @ingroup algorithms */ template<typename Range, typename Body, typename Partitioner=simple_partitioner> class start_scan: public task { typedef sum_node<Range,Body> sum_node_type; typedef final_sum<Range,Body> final_sum_type; final_sum_type* my_body; /** Non-null if caller is requesting total. */ final_sum_type** my_sum; sum_node_type** my_return_slot; /** Null if computing root. */ sum_node_type* my_parent_sum; bool my_is_final; bool my_is_right_child; Range my_range; typename Partitioner::partition_type my_partition; /*override*/ task* execute(); public: start_scan( sum_node_type*& return_slot_, start_scan& parent_, sum_node_type* parent_sum_ ) : my_body(parent_.my_body), my_sum(parent_.my_sum), my_return_slot(&return_slot_), my_parent_sum(parent_sum_), my_is_final(parent_.my_is_final), my_is_right_child(false), my_range(parent_.my_range,split()), my_partition(parent_.my_partition,split()) { __TBB_ASSERT( !*my_return_slot, NULL ); }
start_scan( sum_node_type*& return_slot_, const Range& range_, final_sum_type& body_, const Partitioner& partitioner_) : my_body(&body_), my_sum(NULL), my_return_slot(&return_slot_), my_parent_sum(NULL), my_is_final(true), my_is_right_child(false), my_range(range_), my_partition(partitioner_) { __TBB_ASSERT( !*my_return_slot, NULL ); }
static void run( const Range& range_, Body& body_, const Partitioner& partitioner_ ) { if( !range_.empty() ) { typedef internal::start_scan<Range,Body,Partitioner> start_pass1_type; internal::sum_node<Range,Body>* root = NULL; typedef internal::final_sum<Range,Body> final_sum_type; final_sum_type* temp_body = new(task::allocate_root()) final_sum_type( body_ ); start_pass1_type& pass1 = *new(task::allocate_root()) start_pass1_type( /*my_return_slot=*/root, range_, *temp_body, partitioner_ ); task::spawn_root_and_wait( pass1 ); if( root ) { root->my_body = temp_body; root->my_incoming = NULL; root->my_stuff_last = &body_; task::spawn_root_and_wait( *root ); } else { body_.assign(temp_body->my_body); temp_body->finish_construction( range_, NULL ); temp_body->destroy(*temp_body); } } } };
template<typename Range, typename Body, typename Partitioner> task* start_scan<Range,Body,Partitioner>::execute() { typedef internal::finish_scan<Range,Body> finish_pass1_type; finish_pass1_type* p = my_parent_sum ? static_cast<finish_pass1_type*>( parent() ) : NULL; // Inspecting p->result.left_sum would ordinarily be a race condition.
// But we inspect it only if we are not a stolen task, in which case we
// know that task assigning to p->result.left_sum has completed.
bool treat_as_stolen = my_is_right_child && (is_stolen_task() || my_body!=p->my_result.my_left_sum); if( treat_as_stolen ) { // Invocation is for right child that has been really stolen or needs to be virtually stolen
p->my_right_zombie = my_body = new( allocate_root() ) final_sum_type(my_body->my_body); my_is_final = false; } task* next_task = NULL; if( (my_is_right_child && !treat_as_stolen) || !my_range.is_divisible() || my_partition.should_execute_range(*this) ) { if( my_is_final ) (my_body->my_body)( my_range, final_scan_tag() ); else if( my_sum ) (my_body->my_body)( my_range, pre_scan_tag() ); if( my_sum ) *my_sum = my_body; __TBB_ASSERT( !*my_return_slot, NULL ); } else { sum_node_type* result; if( my_parent_sum ) result = new(allocate_additional_child_of(*my_parent_sum)) sum_node_type(my_range,/*my_left_is_final=*/my_is_final); else result = new(task::allocate_root()) sum_node_type(my_range,/*my_left_is_final=*/my_is_final); finish_pass1_type& c = *new( allocate_continuation()) finish_pass1_type(*my_return_slot,my_sum,*result); // Split off right child
start_scan& b = *new( c.allocate_child() ) start_scan( /*my_return_slot=*/result->my_right, *this, result ); b.my_is_right_child = true; // Left child is recycling of *this. Must recycle this before spawning b,
// otherwise b might complete and decrement c.ref_count() to zero, which
// would cause c.execute() to run prematurely.
recycle_as_child_of(c); c.set_ref_count(2); c.spawn(b); my_sum = &result->my_left_sum; my_return_slot = &result->my_left; my_is_right_child = false; next_task = this; my_parent_sum = result; __TBB_ASSERT( !*my_return_slot, NULL ); } return next_task; } } // namespace internal
//! @endcond
// Requirements on Range concept are documented in blocked_range.h
/** \page parallel_scan_body_req Requirements on parallel_scan body
Class \c Body implementing the concept of parallel_scan body must define: - \code Body::Body( Body&, split ); \endcode Splitting constructor. Split \c b so that \c this and \c b can accumulate separately - \code Body::~Body(); \endcode Destructor - \code void Body::operator()( const Range& r, pre_scan_tag ); \endcode Preprocess iterations for range \c r - \code void Body::operator()( const Range& r, final_scan_tag ); \endcode Do final processing for iterations of range \c r - \code void Body::reverse_join( Body& a ); \endcode Merge preprocessing state of \c a into \c this, where \c a was created earlier from \c b by b's splitting constructor**/
/** \name parallel_scan
See also requirements on \ref range_req "Range" and \ref parallel_scan_body_req "parallel_scan Body". **///@{
//! Parallel prefix with default partitioner
/** @ingroup algorithms **/template<typename Range, typename Body>void parallel_scan( const Range& range, Body& body ) { internal::start_scan<Range,Body,__TBB_DEFAULT_PARTITIONER>::run(range,body,__TBB_DEFAULT_PARTITIONER());}
//! Parallel prefix with simple_partitioner
/** @ingroup algorithms **/template<typename Range, typename Body>void parallel_scan( const Range& range, Body& body, const simple_partitioner& partitioner ) { internal::start_scan<Range,Body,simple_partitioner>::run(range,body,partitioner);}
//! Parallel prefix with auto_partitioner
/** @ingroup algorithms **/template<typename Range, typename Body>void parallel_scan( const Range& range, Body& body, const auto_partitioner& partitioner ) { internal::start_scan<Range,Body,auto_partitioner>::run(range,body,partitioner);}//@}
} // namespace tbb
#endif /* __TBB_parallel_scan_H */
|
