#include <deal.II/matrix_free/task_info.h>

Public Member Functions
	TaskInfo ()

void	clear ()

void	loop (MFWorkerInterface &worker) const

void	collect_boundary_cells (const unsigned int n_active_cells, const unsigned int n_active_and_ghost_cells, const unsigned int vectorization_length, std::vector< unsigned int > &boundary_cells)

void	create_blocks_serial (const std::vector< unsigned int > &boundary_cells, const std::vector< unsigned int > &cells_close_to_boundary, const unsigned int dofs_per_cell, const std::vector< unsigned int > &cell_vectorization_categories, const bool cell_vectorization_categories_strict, std::vector< unsigned int > &renumbering, std::vector< unsigned char > &incompletely_filled_vectorization)

void	initial_setup_blocks_tasks (const std::vector< unsigned int > &boundary_cells, std::vector< unsigned int > &renumbering, std::vector< unsigned char > &incompletely_filled_vectorization)

void	guess_block_size (const unsigned int dofs_per_cell)

void	make_thread_graph_partition_color (DynamicSparsityPattern &connectivity, std::vector< unsigned int > &renumbering, std::vector< unsigned char > &irregular_cells, const bool hp_bool)

void	make_thread_graph_partition_partition (const std::vector< unsigned int > &cell_active_fe_index, DynamicSparsityPattern &connectivity, std::vector< unsigned int > &renumbering, std::vector< unsigned char > &irregular_cells, const bool hp_bool)

void	make_thread_graph (const std::vector< unsigned int > &cell_active_fe_index, DynamicSparsityPattern &connectivity, std::vector< unsigned int > &renumbering, std::vector< unsigned char > &irregular_cells, const bool hp_bool)

void	make_connectivity_cells_to_blocks (const std::vector< unsigned char > &irregular_cells, const DynamicSparsityPattern &connectivity_cells, DynamicSparsityPattern &connectivity_blocks) const

void	make_coloring_within_partitions_pre_blocked (const DynamicSparsityPattern &connectivity, const unsigned int partition, const std::vector< unsigned int > &cell_partition, const std::vector< unsigned int > &partition_list, const std::vector< unsigned int > &partition_size, std::vector< unsigned int > &partition_color_list)

void	make_partitioning_within_partitions_post_blocked (const DynamicSparsityPattern &connectivity, const std::vector< unsigned int > &cell_active_fe_index, const unsigned int partition, const unsigned int cluster_size, const bool hp_bool, const std::vector< unsigned int > &cell_partition, const std::vector< unsigned int > &partition_list, const std::vector< unsigned int > &partition_size, std::vector< unsigned int > &partition_partition_list, std::vector< unsigned char > &irregular_cells)

void	make_partitioning (const DynamicSparsityPattern &connectivity, const unsigned int cluster_size, std::vector< unsigned int > &cell_partition, std::vector< unsigned int > &partition_list, std::vector< unsigned int > &partition_size, unsigned int &partition) const

void	update_task_info (const unsigned int partition)

void	create_flow_graph ()

std::size_t	memory_consumption () const

template<typename StreamType >
void	print_memory_statistics (StreamType &out, std::size_t data_length) const

Public Attributes
unsigned int	n_active_cells

unsigned int	n_ghost_cells

unsigned int	vectorization_length

unsigned int	block_size

unsigned int	n_blocks

TasksParallelScheme	scheme

std::vector< unsigned int >	partition_row_index

std::vector< unsigned int >	cell_partition_data

std::vector< unsigned int >	face_partition_data

std::vector< unsigned int >	boundary_partition_data

std::vector< unsigned int >	ghost_face_partition_data

std::vector< unsigned int >	refinement_edge_face_partition_data

std::vector< unsigned int >	partition_evens

std::vector< unsigned int >	partition_odds

std::vector< unsigned int >	partition_n_blocked_workers

std::vector< unsigned int >	partition_n_workers

unsigned int	evens

unsigned int	odds

unsigned int	n_blocked_workers

unsigned int	n_workers

std::vector< unsigned char >	task_at_mpi_boundary

MPI_Comm	communicator

unsigned int	my_pid

unsigned int	n_procs

Detailed Description

A struct that collects all information related to parallelization with threads: The work is subdivided into tasks that can be done independently.

Author: Katharina Kormann, Martin Kronbichler, 2011, 2018

Definition at line 91 of file task_info.h.

Constructor & Destructor Documentation

◆ TaskInfo()

internal::MatrixFreeFunctions::TaskInfo::TaskInfo ( )

Constructor.

Definition at line 564 of file task_info.cc.

Member Function Documentation

◆ clear()

void internal::MatrixFreeFunctions::TaskInfo::clear ( )

Clears all the data fields and resets them to zero.

Definition at line 571 of file task_info.cc.

◆ loop()

void internal::MatrixFreeFunctions::TaskInfo::loop ( MFWorkerInterface & worker ) const

Runs the matrix-free loop.

Definition at line 324 of file task_info.cc.

◆ collect_boundary_cells()

void internal::MatrixFreeFunctions::TaskInfo::collect_boundary_cells	(	const unsigned int	n_active_cells,
		const unsigned int	n_active_and_ghost_cells,
		const unsigned int	vectorization_length,
		std::vector< unsigned int > &	boundary_cells
	)

Determines the position of cells with ghosts for distributed-memory calculations.

Definition at line 630 of file task_info.cc.

◆ create_blocks_serial()

void internal::MatrixFreeFunctions::TaskInfo::create_blocks_serial	(	const std::vector< unsigned int > &	boundary_cells,
		const std::vector< unsigned int > &	cells_close_to_boundary,
		const unsigned int	dofs_per_cell,
		const std::vector< unsigned int > &	cell_vectorization_categories,
		const bool	cell_vectorization_categories_strict,
		std::vector< unsigned int > &	renumbering,
		std::vector< unsigned char > &	incompletely_filled_vectorization
	)

Sets up the blocks for running the cell loop based on the options controlled by the input arguments.

Parameters

boundary_cells	A list of cells that need to exchange data prior to performing computations. These will be given a certain id in the partitioning.
cells_close_to_boundary	A list of cells that interact with boundaries between different subdomains and are involved in sending data to neighboring processes.
dofs_per_cell	Gives an expected value for the number of degrees of freedom on a cell, which is used to determine the block size for interleaving cell and face integrals.
cell_vectorization_categories	This set of categories defines the cells that should be grouped together inside the lanes of a vectorized array. This can be the polynomial degree in an hp-element or a user-provided grouping.
cell_vectorization_categories_strict	Defines whether the categories defined by the previous variables should be separated strictly or whether it is allowed to insert lower categories into the next high one(s).
renumbering	When leaving this function, the vector contains a new numbering of the cells that aligns with the grouping stored in this class.
incompletely_filled_vectorization	Given the vectorized layout of this class, some cell batches might have components in the vectorized array (SIMD lanes) that are not used and do not carray valid data. This array indicates the cell batches where this occurs according to the renumbering returned by this function.

Definition at line 701 of file task_info.cc.

◆ initial_setup_blocks_tasks()

void internal::MatrixFreeFunctions::TaskInfo::initial_setup_blocks_tasks	(	const std::vector< unsigned int > &	boundary_cells,
		std::vector< unsigned int > &	renumbering,
		std::vector< unsigned char > &	incompletely_filled_vectorization
	)

First step in the block creation for the task-parallel blocking setup.

Parameters

boundary_cells	A list of cells that need to exchange data prior to performing computations. These will be given a certain id in the partitioning.
renumbering	When leaving this function, the vector contains a new numbering of the cells that aligns with the grouping stored in this class (before actually creating the tasks).
incompletely_filled_vectorization	Given the vectorized layout of this class, some cell batches might have components in the vectorized array (SIMD lanes) that are not used and do not carray valid data. This array indicates the cell batches where this occurs according to the renumbering returned by this function.

Definition at line 886 of file task_info.cc.

◆ guess_block_size()

void internal::MatrixFreeFunctions::TaskInfo::guess_block_size ( const unsigned int dofs_per_cell )

This helper function determines a block size if the user decided not to force a block size through MatrixFree::AdditionalData. This is computed based on the number of hardware threads on the system and the number of macro cells that we should work on.

Definition at line 945 of file task_info.cc.

◆ make_thread_graph_partition_color()

void internal::MatrixFreeFunctions::TaskInfo::make_thread_graph_partition_color	(	DynamicSparsityPattern &	connectivity,
		std::vector< unsigned int > &	renumbering,
		std::vector< unsigned char > &	irregular_cells,
		const bool	hp_bool
	)

This method goes through all cells that have been filled into dof_indices and finds out which cells can be worked on independently and which ones are neighboring and need to be done at different times when used in parallel.

The strategy is based on a two-level approach. The outer level is subdivided into partitions similar to the type of neighbors in Cuthill-McKee, and the inner level is subdivided via colors (for chunks within the same color, can work independently). One task is represented by a chunk of cells. The cell chunks are formed before subdivision into partitions and colors.

Parameters

cell_active_fe_index	The active FE index corresponding to the individual indices in the list of all cell indices, in order to be able to not place cells with different indices into the same cell batch with vectorization.
connectivity	(in/out) Determines whether cells `i` and `j` are conflicting, expressed by an entry in position (i,j).
renumbering	(in/out) At output, the element j of this variable gives the original number of the cell that is reordered to place j by the ordering due to the thread graph.
irregular_cells	(in/out) Informs the current function whether some SIMD lanes in VectorizedArray would not be filled for a given cell batch index.
hp_bool	Defines whether we are in hp mode or not

Definition at line 975 of file task_info.cc.

◆ make_thread_graph_partition_partition()

void internal::MatrixFreeFunctions::TaskInfo::make_thread_graph_partition_partition	(	const std::vector< unsigned int > &	cell_active_fe_index,
		DynamicSparsityPattern &	connectivity,
		std::vector< unsigned int > &	renumbering,
		std::vector< unsigned char > &	irregular_cells,
		const bool	hp_bool
	)

This function goes through all cells that have been filled into dof_indices and finds out which cells can be worked on independently and which ones are neighboring and need to be done at different times when used in parallel.

The strategy is based on a two-level approach. The outer level is subdivided into partitions similar to the type of neighbors in Cuthill-McKee, and the inner level is again subdivided into Cuthill- McKee-like partitions (partitions whose level differs by more than 2 can be worked on independently). One task is represented by a chunk of cells. The cell chunks are formed after subdivision into the two levels of partitions.

Parameters

cell_active_fe_index	The active FE index corresponding to the individual indices in the list of all cell indices, in order to be able to not place cells with different indices into the same cell batch with vectorization.
connectivity	(in/out) Determines whether cells `i` and `j` are conflicting, expressed by an entry in position (i,j).
renumbering	(in/out) At output, the element j of this variable gives the original number of the cell that is reordered to place j by the ordering due to the thread graph.
irregular_cells	(in/out) Informs the current function whether some SIMD lanes in VectorizedArray would not be filled for a given cell batch index.
hp_bool	Defines whether we are in hp mode or not

Definition at line 1284 of file task_info.cc.

◆ make_thread_graph()

void internal::MatrixFreeFunctions::TaskInfo::make_thread_graph	(	const std::vector< unsigned int > &	cell_active_fe_index,
		DynamicSparsityPattern &	connectivity,
		std::vector< unsigned int > &	renumbering,
		std::vector< unsigned char > &	irregular_cells,
		const bool	hp_bool
	)

Either calls make_thread_graph_partition_color() or make_thread_graph_partition_partition() accessible from the outside, depending on the setting in the data structure.

Parameters

cell_active_fe_index	The active FE index corresponding to the individual indices in the list of all cell indices, in order to be able to not place cells with different indices into the same cell batch with vectorization.
connectivity	(in/out) Determines whether cells `i` and `j` are conflicting, expressed by an entry in position (i,j).
renumbering	(in/out) At output, the element j of this variable gives the original number of the cell that is reordered to place j by the ordering due to the thread graph.
irregular_cells	(in/out) Informs the current function whether some SIMD lanes in VectorizedArray would not be filled for a given cell batch index.
hp_bool	Defines whether we are in hp mode or not

Definition at line 1110 of file task_info.cc.

◆ make_connectivity_cells_to_blocks()

void internal::MatrixFreeFunctions::TaskInfo::make_connectivity_cells_to_blocks	(	const std::vector< unsigned char > &	irregular_cells,
		const DynamicSparsityPattern &	connectivity_cells,
		DynamicSparsityPattern &	connectivity_blocks
	)		const

This function computes the connectivity between blocks of cells from the connectivity between the individual cells.

Definition at line 1345 of file task_info.cc.

◆ make_coloring_within_partitions_pre_blocked()

void internal::MatrixFreeFunctions::TaskInfo::make_coloring_within_partitions_pre_blocked	(	const DynamicSparsityPattern &	connectivity,
		const unsigned int	partition,
		const std::vector< unsigned int > &	cell_partition,
		const std::vector< unsigned int > &	partition_list,
		const std::vector< unsigned int > &	partition_size,
		std::vector< unsigned int > &	partition_color_list
	)

Function to create coloring on the second layer within each partition.

Definition at line 1678 of file task_info.cc.

◆ make_partitioning_within_partitions_post_blocked()

void internal::MatrixFreeFunctions::TaskInfo::make_partitioning_within_partitions_post_blocked	(	const DynamicSparsityPattern &	connectivity,
		const std::vector< unsigned int > &	cell_active_fe_index,
		const unsigned int	partition,
		const unsigned int	cluster_size,
		const bool	hp_bool,
		const std::vector< unsigned int > &	cell_partition,
		const std::vector< unsigned int > &	partition_list,
		const std::vector< unsigned int > &	partition_size,
		std::vector< unsigned int > &	partition_partition_list,
		std::vector< unsigned char > &	irregular_cells
	)

Function to create partitioning on the second layer within each partition.

Definition at line 1387 of file task_info.cc.

◆ make_partitioning()

void internal::MatrixFreeFunctions::TaskInfo::make_partitioning	(	const DynamicSparsityPattern &	connectivity,
		const unsigned int	cluster_size,
		std::vector< unsigned int > &	cell_partition,
		std::vector< unsigned int > &	partition_list,
		std::vector< unsigned int > &	partition_size,
		unsigned int &	partition
	)		const

This function creates partitions according to the provided connectivity graph.

Parameters

connectivity	Connectivity between (blocks of cells)
cluster_size	The number of cells in each partition should be a multiple of cluster_size (for blocking later on)
cell_partition	Saves of each (block of cells) to which partition the block belongs
partition_list	partition_list[j] gives the old number of the block that should be renumbered to j due to the partitioning
partition_size	Vector pointing to start of each partition (on output)
partition	number of partitions created

Definition at line 1752 of file task_info.cc.

◆ update_task_info()

void internal::MatrixFreeFunctions::TaskInfo::update_task_info ( const unsigned int partition )

Update fields of task info for task graph set up in make_thread_graph.

Definition at line 1973 of file task_info.cc.

◆ create_flow_graph()

void internal::MatrixFreeFunctions::TaskInfo::create_flow_graph ( )

Creates a task graph from a connectivity structure.

◆ memory_consumption()

std::size_t internal::MatrixFreeFunctions::TaskInfo::memory_consumption ( ) const

Returns the memory consumption of the class.

Definition at line 614 of file task_info.cc.

◆ print_memory_statistics()

template<typename StreamType >

template void internal::MatrixFreeFunctions::TaskInfo::print_memory_statistics< ConditionalOStream >	(	StreamType &	out,
		std::size_t	data_length
	)		const

Prints minimum, average, and maximal memory consumption over the MPI processes.

Definition at line 599 of file task_info.cc.

Member Data Documentation

◆ n_active_cells

unsigned int internal::MatrixFreeFunctions::TaskInfo::n_active_cells

Number of physical cells in the mesh, not cell batches after vectorization

Definition at line 402 of file task_info.h.

◆ n_ghost_cells

unsigned int internal::MatrixFreeFunctions::TaskInfo::n_ghost_cells

Number of physical ghost cells in the mesh which are subject to special treatment and should not be included in loops

Definition at line 408 of file task_info.h.

◆ vectorization_length

unsigned int internal::MatrixFreeFunctions::TaskInfo::vectorization_length

Number of lanes in the SIMD array that are used for vectorization

Definition at line 413 of file task_info.h.

◆ block_size

unsigned int internal::MatrixFreeFunctions::TaskInfo::block_size

Block size information for multithreading

Definition at line 418 of file task_info.h.

◆ n_blocks

unsigned int internal::MatrixFreeFunctions::TaskInfo::n_blocks

Number of blocks for multithreading

Definition at line 423 of file task_info.h.

◆ scheme

TasksParallelScheme internal::MatrixFreeFunctions::TaskInfo::scheme

Parallel scheme applied by multithreading

Definition at line 428 of file task_info.h.

◆ partition_row_index

std::vector<unsigned int> internal::MatrixFreeFunctions::TaskInfo::partition_row_index

The blocks are organized by a vector-of-vector concept, and this data field partition_row_index stores the distance from one 'vector' to the next within the linear storage of all data to the two-level partitioning.

Definition at line 436 of file task_info.h.

◆ cell_partition_data

std::vector<unsigned int> internal::MatrixFreeFunctions::TaskInfo::cell_partition_data

This is a linear storage of all partitions, building a range of indices of the form cell_partition_data[idx] to cell_partition_data[idx+1] within the integer list of all cells in MatrixFree, subdivided into chunks by partition_row_index.

Definition at line 444 of file task_info.h.

◆ face_partition_data

std::vector<unsigned int> internal::MatrixFreeFunctions::TaskInfo::face_partition_data

This is a linear storage of all partitions of inner faces, building a range of indices of the form face_partition_data[idx] to face_partition_data[idx+1] within the integer list of all interior faces in MatrixFree, subdivided into chunks by partition_row_index.

Definition at line 453 of file task_info.h.

◆ boundary_partition_data

std::vector<unsigned int> internal::MatrixFreeFunctions::TaskInfo::boundary_partition_data

This is a linear storage of all partitions of boundary faces, building a range of indices of the form boundary_partition_data[idx] to boundary_partition_data[idx+1] within the integer list of all boundary faces in MatrixFree, subdivided into chunks by partition_row_index.

Definition at line 462 of file task_info.h.

◆ ghost_face_partition_data

std::vector<unsigned int> internal::MatrixFreeFunctions::TaskInfo::ghost_face_partition_data

This is a linear storage of all partitions of interior faces on boundaries to other processors that are not locally used, building a range of indices of the form ghost_face_partition_data[idx] to ghost_face_partition_data[idx+1] within the integer list of all such faces in MatrixFree, subdivided into chunks by partition_row_index.

Definition at line 472 of file task_info.h.

◆ refinement_edge_face_partition_data

std::vector<unsigned int> internal::MatrixFreeFunctions::TaskInfo::refinement_edge_face_partition_data

This is a linear storage of all partitions of faces for multigrid levels that have a coarser neighbor and are only included in certain residual computations but not in smoothing, building a range of indices of the form refinement_edge_face_partition_data[idx] to refinement_edge_face_partition_data[idx+1] within the integer list of all such faces in MatrixFree, subdivided into chunks by partition_row_index.

Definition at line 483 of file task_info.h.

◆ partition_evens

std::vector<unsigned int> internal::MatrixFreeFunctions::TaskInfo::partition_evens

Thread information (which chunk to start 'even' partitions from) to be handed to the dynamic task scheduler

Definition at line 489 of file task_info.h.

◆ partition_odds

std::vector<unsigned int> internal::MatrixFreeFunctions::TaskInfo::partition_odds

Thread information (which chunk to start 'odd' partitions from) to be handed to the dynamic task scheduler

Definition at line 495 of file task_info.h.

◆ partition_n_blocked_workers

std::vector<unsigned int> internal::MatrixFreeFunctions::TaskInfo::partition_n_blocked_workers

Thread information regarding the dependencies for partitions handed to the dynamic task scheduler

Definition at line 501 of file task_info.h.

◆ partition_n_workers

std::vector<unsigned int> internal::MatrixFreeFunctions::TaskInfo::partition_n_workers

Thread information regarding the dependencies for partitions handed to the dynamic task scheduler

Definition at line 507 of file task_info.h.

◆ evens

unsigned int internal::MatrixFreeFunctions::TaskInfo::evens

Number of even partitions accumulated over the field partitions_even

Definition at line 513 of file task_info.h.

◆ odds

unsigned int internal::MatrixFreeFunctions::TaskInfo::odds

Number of odd partitions accumulated over the field partitions_odd

Definition at line 519 of file task_info.h.

◆ n_blocked_workers

unsigned int internal::MatrixFreeFunctions::TaskInfo::n_blocked_workers

Number of blocked workers accumulated over the field partition_n_blocked_workers

Definition at line 525 of file task_info.h.

◆ n_workers

unsigned int internal::MatrixFreeFunctions::TaskInfo::n_workers

Number of workers accumulated over the field partition_n_workers

Definition at line 530 of file task_info.h.

◆ task_at_mpi_boundary

std::vector<unsigned char> internal::MatrixFreeFunctions::TaskInfo::task_at_mpi_boundary

Stores whether a particular task is at an MPI boundary and needs data exchange

Definition at line 536 of file task_info.h.

◆ communicator

MPI_Comm internal::MatrixFreeFunctions::TaskInfo::communicator

MPI communicator

Definition at line 541 of file task_info.h.

◆ my_pid

unsigned int internal::MatrixFreeFunctions::TaskInfo::my_pid

Rank of MPI process

Definition at line 546 of file task_info.h.

◆ n_procs

unsigned int internal::MatrixFreeFunctions::TaskInfo::n_procs

Number of MPI rank for the current communicator

Definition at line 551 of file task_info.h.

The documentation for this struct was generated from the following files:

deal.II/matrix_free/task_info.h
source/matrix_free/task_info.cc

Public Member Functions

Public Attributes

Detailed Description

Constructor & Destructor Documentation

◆ TaskInfo()

Member Function Documentation

◆ clear()

◆ loop()

◆ collect_boundary_cells()

◆ create_blocks_serial()

◆ initial_setup_blocks_tasks()

◆ guess_block_size()

◆ make_thread_graph_partition_color()

◆ make_thread_graph_partition_partition()

◆ make_thread_graph()

◆ make_connectivity_cells_to_blocks()

◆ make_coloring_within_partitions_pre_blocked()

◆ make_partitioning_within_partitions_post_blocked()

◆ make_partitioning()

◆ update_task_info()

◆ create_flow_graph()

◆ memory_consumption()

◆ print_memory_statistics()

Member Data Documentation

◆ n_active_cells

◆ n_ghost_cells

◆ vectorization_length

◆ block_size

◆ n_blocks

◆ scheme

◆ partition_row_index

◆ cell_partition_data

◆ face_partition_data

◆ boundary_partition_data

◆ ghost_face_partition_data

◆ refinement_edge_face_partition_data

◆ partition_evens

◆ partition_odds

◆ partition_n_blocked_workers

◆ partition_n_workers

◆ evens

◆ odds

◆ n_blocked_workers

◆ n_workers

◆ task_at_mpi_boundary

◆ communicator

◆ my_pid

◆ n_procs