hpx/parallel/algorithms/for_loop_reduction.hpp#

See Public API for a list of names and headers that are part of the public HPX API.

namespace hpx
namespace experimental

Top-level namespace.

Functions

template<typename T, typename Op>
constexpr hpx::parallel::detail::reduction_helper<T, std::decay_t<Op>> reduction(T &var, T const &identity, Op &&combiner)#

The function template returns a reduction object of unspecified type having a value type and encapsulating an identity value for the reduction, a combiner function object, and a live-out object from which the initial value is obtained and into which the final value is stored.

A parallel algorithm uses reduction objects by allocating an unspecified number of instances, called views, of the reduction’s value type. Each view is initialized with the reduction object’s identity value, except that the live-out object (which was initialized by the caller) comprises one of the views. The algorithm passes a reference to a view to each application of an element-access function, ensuring that no two concurrently-executing invocations share the same view. A view can be shared between two applications that do not execute concurrently, but initialization is performed only once per view.

Modifications to the view by the application of element access functions accumulate as partial results. At some point before the algorithm returns, the partial results are combined, two at a time, using the reduction object’s combiner operation until a single value remains, which is then assigned back to the live-out object.

T shall meet the requirements of CopyConstructible and MoveAssignable. The expression

var = combiner(var, var) 
shall be well formed.

Note

In order to produce useful results, modifications to the view should be limited to commutative operations closely related to the combiner operation. For example if the combiner is plus<T>, incrementing the view would be consistent with the combiner but doubling it or assigning to it would not.

Template Parameters
  • T – The value type to be used by the induction object.

  • Op – The type of the binary function (object) used to perform the reduction operation.

Parameters
  • var – [in,out] The life-out value to use for the reduction object. This will hold the reduced value after the algorithm is finished executing.

  • identity – [in] The identity value to use for the reduction operation.

  • combiner – [in] The binary function (object) used to perform a pairwise reduction on the elements.

Returns

This returns a reduction object of unspecified type having a value type of T. When the return value is used by an algorithm, the reference to var is used as the live-out object, new views are initialized to a copy of identity, and views are combined by invoking the copy of combiner, passing it the two views to be combined.

namespace parallel