Class to manage the parallelization scheme. More...

#include <parallel.h>

Collaboration diagram for Parallel:

Static Public Member Functions
static void	initialize ()
	initializes MPI, gets the rank number and the total numer of processors

static void	createGeometry (std::array< int, 4 > latticeSize, std::array< int, 4 > subLatticeSize)
	creates the parallel gemometry of the lattice More...

static void	finalize ()
	finalizes MPI

static int	rank ()

static int	numProcs ()

static int	activeProcs ()

static bool	isActive ()

static MPI_Comm	cartCoordComm ()

static std::array< int, 4 > &	subBlocks ()

static std::array< int, 4 > &	rankCoord ()

static std::array< int, 4 > &	latticeSubSize ()

static std::array< int, 4 > &	latticeFullSize ()

static std::array< int, 4 > &	parity ()

static int	getNeighbor (int direction, int sign)
	return the rank of the neighbor along the given direction and sign

static int	getSecondNeighbor (int direction1, int sign1, int direction2, int sign2)
	return the rank of the neighbor along the two given directions and signs

static void	openFile (MPI_File &file, const char *fileName)
	opens a file with MPI

static void	closeFile (MPI_File &file)
	closes a file with MPI

Static Private Member Functions
static void	createNeighborLists ()
	Create castesian coordinates and creates neighbor lists for every processor in every direction.

static void	assignNeighbor (int direction)
	use MPI utilities to find the neighbors in one direction

static void	assignSecondNeighbor (int dir1, int dir2)
	use MPI utilities to find the neighbors in two direction

static void	MyMPI_Cart_shift2 (MPI_Comm comm, int dir1, int shift1, int dir2, int shift2, int &source, int &dest)
	use MPI utilities to find the neighbors in two direction

Static Private Attributes
static int	m_rank

static int	m_numProcs

static int	m_activeProcs

static bool	m_isActive

static MPI_Comm	m_cartCoordComm

static std::array< int, 4 >	m_subBlocks

static std::array< int, 4 >	m_rankCoord

static std::array< int, 4 >	m_latticeSubSize

static std::array< int, 4 >	m_latticeFullSize

static std::array< int, 4 >	m_parity

static MPI_Comm	m_cartCoords

static std::array< std::array< int, 2 >, 4 >	m_neighbor

static std::array< std::array< std::array< std::array< int, 2 >, 4 >, 2 >, 4 >	m_secondNeighbor

Detailed Description

Class to manage the parallelization scheme.

Author: Giovanni Pederiva

Version: 1.0

Date: 2017-2018

Copyright: MIT License.

Utility class to manage the parallel architecture of the program and interface each process with the overall job. Handles local coordinates, nearest neighbors and communicators

Definition at line 56 of file parallel.h.

Member Function Documentation

void Parallel::createGeometry	(	std::array< int, 4 >	latticeSize,
		std::array< int, 4 >	subLatticeSize
	)

static

creates the parallel gemometry of the lattice

Parameters

latticeSize	array containing the total size of the lattice
latticeSize	array containing the size of each sublattice

Definition at line 77 of file parallel.cpp.

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

Static Public Member Functions

Static Private Member Functions

Static Private Attributes

Detailed Description

Member Function Documentation