Parametric Dependencies

Parametric dependencies define the finest-grain component of the Skeletons Framework. These entities are represented as nodes in the generated dataflow graphs in a given environment. More...

Collaboration diagram for Parametric Dependencies:

Classes
class	stapl::skeletons::skeletons_impl::pre_broadcast_pd< Arity, Op, Span, SetResult >
	Creates a set_result parametric dependency given a op. Basically this parametric dependency is only used for sinking values of skeletons which couldn't currently set their results directly without using sink_value skeleton. More...
class	stapl::skeletons::skeletons_impl::set_result_pd< Arity, Op, Span, SetResult >
	Creates a set_result parametric dependency given a op. Basically this parametric dependency is only used for sinking values of skeletons which couldn't currently set their results directly without using sink_value skeleton. More...

Modules
	Parametric Dependencies Internal
	Parametric Dependencies of
	Algorithm Specific Parametric
	Input Resizing Parametric

Functions
skeletons_impl::bitreversal_pd	stapl::skeletons::bitreversal_pd (void)
	Creates a biterversal parametric dependency.
template<typename ContCond >
skeletons_impl::do_while_pd< ContCond >	stapl::skeletons::do_while_pd (ContCond const &continuation_cond)
	Creates a do while parametric dependency given a continuation_cond functor. More...
template<typename Generator >
skeletons_impl::farm_pd< Generator >	stapl::skeletons::farm_pd (Generator const &generator)
	Creates a farm parametric dependency given a generator. More...
template<int NotifCount = 1, typename Op , typename F = skeletons::no_filter>
skeletons_impl::notify_map_pd< Op, NotifCount, F >	stapl::skeletons::notify_map_pd (Op const &op, F const &f=F())
	Creates a zip parametric dependency given a op. This method is used whenever filtering is needed on the input edge. More...
template<std::size_t Arity, typename Span , bool SetResult, typename Op >
skeletons_impl::pre_broadcast_pd< Arity, Op, Span, SetResult >	stapl::skeletons::pre_broadcast_pd (Op const &op)
	Creates a set_result parametric dependency given a op. Basically this parametric dependency is only used for sinking values of skeletons which couldn't currently set their results directly without using sink_value skeleton. More...
template<int i, typename Op >
skeletons_impl::serial_pd< Op, i >	stapl::skeletons::serial_pd (Op const &op, std::size_t number_of_sets)
	Creates a serial_pd parametric dependency. More...
template<std::size_t Arity, typename Span , bool SetResult, typename Op >
skeletons_impl::set_result_pd< Arity, Op, Span, SetResult >	stapl::skeletons::set_result_pd (Op const &op)
	Creates a set_result parametric dependency given a op. Basically this parametric dependency is only used for sinking values of skeletons which couldn't currently set their results directly without using sink_value skeleton. More...
template<int i = 2, typename Op , typename T >
skeletons_impl::shifted_first_pd< Op, T, i >	stapl::skeletons::shifted_first_pd (Op const &op, T neutral_value)
	Creates a shifted first input parametric dependency given a op and number of input flows specified by i, and the default value specified by neutral_value. More...
template<int i = 2, std::size_t dims = 1, bool SetResult = false, typename Op , typename F = skeletons::no_filter, typename Mappers = skeletons::no_mapper>
skeletons_impl::zip_pd< Op, i, dims, F, Mappers, SetResult >	stapl::skeletons::zip_pd (Op const &op, F const &f=F(), Mappers const &mappers=Mappers())
	Creates a zip parametric dependency given a op. This method is used whenever filtering is needed on the input edge. More...

Detailed Description

Parametric dependencies define the finest-grain component of the Skeletons Framework. These entities are represented as nodes in the generated dataflow graphs in a given environment.

Function Documentation

do_while_pd()

template<typename ContCond >

skeletons_impl::do_while_pd<ContCond> stapl::skeletons::do_while_pd

(

ContCond const &

continuation_cond

)

Creates a do while parametric dependency given a continuation_cond functor.

A do_while parametric dependency represents an element in the dependence graph that controls if the next iteration of a do-while loop should be spawned or not. It does that by wrapping user's ContCond by do_while_stub with and sending memento_stack to it.

farm_pd()

template<typename Generator >

skeletons_impl::farm_pd<Generator> stapl::skeletons::farm_pd

(

Generator const &

generator

)

Creates a farm parametric dependency given a generator.

The given generator is passed a farm instance using which it can add the initial seeds of the computation.

This parametric dependency is used in farms skeleton to generate the initial seeds of computation.

notify_map_pd()

template<int NotifCount = 1, typename Op , typename F = skeletons::no_filter>

skeletons_impl::notify_map_pd<Op, NotifCount, F> stapl::skeletons::notify_map_pd	(	Op const &	op,
		F const &	f = F()
	)

Creates a zip parametric dependency given a op. This method is used whenever filtering is needed on the input edge.

A notify_map parametric dependency is similar to a map_pd, however, it waits for NotifCount notifications from the first NotifCount flows.

pre_broadcast_pd()

template<std::size_t Arity, typename Span , bool SetResult, typename Op >

skeletons_impl::pre_broadcast_pd<Arity, Op, Span, SetResult> stapl::skeletons::pre_broadcast_pd

(

Op const &

op

)

Creates a set_result parametric dependency given a op. Basically this parametric dependency is only used for sinking values of skeletons which couldn't currently set their results directly without using sink_value skeleton.

Template Parameters

Arity	the arity of set_result
Span	the span for the previous skeleton that is passed to set result for detecting if the corresponding task is spanned or not
SetResult	whether put the task result on the result container or not

Parameters

op	the workfunction to be used in each set_result parametric dependency

serial_pd()

template<int i, typename Op >

skeletons_impl::serial_pd<Op, i> stapl::skeletons::serial_pd	(	Op const &	op,
		std::size_t	number_of_sets
	)

Creates a serial_pd parametric dependency.

Template Parameters

i	number of inputs.

Parameters

op	operation to be applied at each coordinate.
number_of_sets	The number of sets to form.

set_result_pd()

template<std::size_t Arity, typename Span , bool SetResult, typename Op >

skeletons_impl::set_result_pd<Arity, Op, Span, SetResult> stapl::skeletons::set_result_pd

(

Op const &

op

)

Creates a set_result parametric dependency given a op. Basically this parametric dependency is only used for sinking values of skeletons which couldn't currently set their results directly without using sink_value skeleton.

Template Parameters

Arity	the number of inputs are passed to the skeleton
SetResult	whether the skeleton should set the task results on the pg edge container or not
Span	the span for the previous skeleton that is passed to set result_pd for detecting if the corresponding corresponding task is spanned or not

Parameters

op	the workfunction to be used in each set_result parametric dependency

shifted_first_pd()

template<int i = 2, typename Op , typename T >

skeletons_impl::shifted_first_pd<Op, T, i> stapl::skeletons::shifted_first_pd	(	Op const &	op,
		T	neutral_value
	)

Creates a shifted first input parametric dependency given a op and number of input flows specified by i, and the default value specified by neutral_value.

Parameters

op	the operation that will be used for the spawned element
neutral_value	the value to be used for non-existing indices

zip_pd()

template<int i = 2, std::size_t dims = 1, bool SetResult = false, typename Op , typename F = skeletons::no_filter, typename Mappers = skeletons::no_mapper>

skeletons_impl::zip_pd<Op, i, dims, F, Mappers, SetResult> stapl::skeletons::zip_pd	(	Op const &	op,
		F const &	f = F(),
		Mappers const &	mappers = Mappers()
	)

Creates a zip parametric dependency given a op. This method is used whenever filtering is needed on the input edge.

Template Parameters

SetResult	whether the skeleton should set the task results on the pg edge container or not
F	the filter to be used when consuming from a task A zip parametric dependency is usually used in zip skeleton. It sends input with the same index of each flow to the given Op.