TrioCFD 1.9.9_beta
TrioCFD documentation
Loading...
Searching...
No Matches
Partitionneur_base_32_64< _SIZE_ > Class Template Referenceabstract

Base class for domain partitioners (for splitting a mesh before a parallel computation). More...

#include <Partitionneur_base.h>

Inheritance diagram for Partitionneur_base_32_64< _SIZE_ >:
[legend]
Collaboration diagram for Partitionneur_base_32_64< _SIZE_ >:
[legend]

Public Types

using int_t = _SIZE_
using ArrOfInt_t = ArrOfInt_T<_SIZE_>
using IntVect_t = IntVect_T<_SIZE_>
using IntTab_t = IntTab_T<_SIZE_>
using Domaine_t = Domaine_32_64<_SIZE_>
using Static_Int_Lists_t = Static_Int_Lists_32_64<_SIZE_>
using Bord_t = Bord_32_64<_SIZE_>
using BigArrOfInt_ = TRUSTArray<int, _SIZE_>
using BigIntVect_ = TRUSTVect<int, _SIZE_>

Public Member Functions

int lire_motcle_non_standard (const Motcle &, Entree &) override
 Reads non-simple-type parameters of an Objet_U from an input stream.
virtual void associer_domaine (const Domaine_t &domaine)=0
virtual void construire_partition (BigIntVect_ &elem_part, int &nb_parts_tot) const =0
Public Member Functions inherited from Objet_U
 ~Objet_U () override
 Destructor. Removes the object from the list of objects registered in "memory".
int numero () const
 Returns the index of the object in Memoire::data.
virtual int duplique () const =0
virtual SortieprintOn (Sortie &) const
 Writes the object to an output stream. Virtual method to override.
virtual unsigned taille_memoire () const =0
virtual int est_egal_a (const Objet_U &) const
 Returns 1 if x and *this are the same instance (same memory address).
virtual const Nomle_nom () const
 Returns the name of the Objet_U. Virtual method to override: returns "neant" in this implementation.
virtual void nommer (const Nom &)
 Assigns a name to the Objet_U. Virtual method to override.
virtual int reprendre (Entree &)
 Restores an Objet_U from an input stream. Virtual method to override.
virtual int sauvegarder (Sortie &) const
 Saves an Objet_U to an output stream. Virtual method to override.
int get_object_id () const
 Returns the unique identifier of the object (object_id_).
virtual const Type_infoget_info () const
 Returns type information for the Objet_U.
const Nomque_suis_je () const
 Returns the string identifying the class.
const char * le_type () const
 Returns the type name of the Objet_U.
virtual int change_num (const int *const)
 Changes the internal number of the Objet_U.
virtual int associer_ (Objet_U &)
 Associates the Objet_U with another Objet_U. Virtual method to override.
const Interpreteinterprete () const
Interpreteinterprete ()
Public Member Functions inherited from Process
virtual ~Process ()

Static Public Member Functions

static void corriger_elem0_sur_proc0 (BigIntVect_ &elem_part)
 Corrects the partition so that element 0 of the initial domain is on the first sub-domain of the partition.
static int_t calculer_graphe_connexions_periodiques (const Domaine_t &domaine, const Static_Int_Lists_t &som_elem, const int_t my_offset, Static_Int_Lists_t &graph)
 Computes a connectivity graph between elements connected by periodic faces.
static int_t corriger_bords_avec_graphe (const Static_Int_Lists_t &graph_elements_perio, const Static_Int_Lists_t &som_elem, const Domaine_t &domaine, BigIntVect_ &elem_part)
 Corrects elem_part so that element i is on the same partition elem_part[i] as all elements connected to it in the graph.
static void corriger_bords_avec_liste (const Domaine_t &dom, const int_t my_offset, BigIntVect_ &elem_part)
 Computes the periodic element connectivity graphs and calls corriger_periodique_avec_graphe.
static int_t corriger_sommets_bord (const Domaine_t &domaine, const ArrOfInt_t &renum_som_perio, const Static_Int_Lists_t &som_elem, BigIntVect_ &elem_part)
 Modifies elem_part to ensure the following properties: 1) Elements that have a boundary vertex are associated.
static int_t corriger_multiperiodique (const Domaine_t &domaine, const ArrOfInt_t &renum_som_perio, const Static_Int_Lists_t &som_elem, BigIntVect_ &elem_part)
 Applies corrections to elem_part so that multi-periodicity is correct:
Static Public Member Functions inherited from Objet_U
static const Nomnom_du_cas ()
 Returns a constant reference to the case name. This method is static.
static Nomget_set_nom_du_cas ()
 Returns a non-constant reference to the case name (to allow modification). This method is static.
static const Type_infoinfo ()
 Returns type information for the Objet_U (static version).
static const Objet_Uself_cast (const Objet_U &)
 Method added for casting in Python.
static Objet_Uself_cast (Objet_U &)
Static Public Member Functions inherited from Process
static int me ()
 Returns the rank of the local processor in the current communication group. See Comm_Group::rank() and PE_Groups::current_group().
static int nproc ()
 Returns the number of processors in the current group. See Comm_Group::nproc() and PE_Groups::current_group().
static bool is_parallel ()
static void exit (int exit_code=-1)
 Exit routine for TRUST within a Kokkos region.
static double mp_sum (double)
 Computes the sum of x over all processors in the current group.
static float mp_sum (float)
static trustIdType mp_sum (trustIdType)
 Computes the sum of x over all processors in the current group.
static double mp_max (double)
static double mp_min (double)
static int mp_max (int)
 Returns the maximum value of x across all processors in the current group.
static int mp_min (int)
 Returns the minimum value of x across all processors in the current group.
static double mp_sum_as_double (int v)
static trustIdType mppartial_sum (trustIdType i)
 Computes the partial sum of x over processors 0 to me()-1 (returns 0 on processor 0).
template<typename T>
static void mp_sum_for_each (T &arg1, T &arg2)
 C++14 compatible mp_sum_for_each: combine multiple mp_sum calls into one collective operation Usage: mp_sum_for_each(a, b); mp_sum_for_each(a, b, c); mp_sum_for_each(a, b, c, d); mp_sum_for_each(a, b, c, d, e); All arguments must be of the same type (double or int) and are modified in place. Supports 2-5 parameters.
template<typename T>
static void mp_sum_for_each (T &arg1, T &arg2, T &arg3)
template<typename T>
static void mp_sum_for_each (T &arg1, T &arg2, T &arg3, T &arg4)
template<typename T>
static void mp_sum_for_each (T &arg1, T &arg2, T &arg3, T &arg4, T &arg5)
template<typename T>
static void mp_max_for_each (T &arg1, T &arg2)
 C++14 compatible mp_max_for_each: combine multiple mp_max calls into one collective operation.
template<typename T>
static void mp_max_for_each (T &arg1, T &arg2, T &arg3)
template<typename T>
static void mp_max_for_each (T &arg1, T &arg2, T &arg3, T &arg4)
template<typename T>
static void mp_max_for_each (T &arg1, T &arg2, T &arg3, T &arg4, T &arg5)
template<typename T>
static void mp_min_for_each (T &arg1, T &arg2)
 C++14 compatible mp_min_for_each: combine multiple mp_min calls into one collective operation.
template<typename T>
static void mp_min_for_each (T &arg1, T &arg2, T &arg3)
template<typename T>
static void mp_min_for_each (T &arg1, T &arg2, T &arg3, T &arg4)
template<typename T>
static void mp_min_for_each (T &arg1, T &arg2, T &arg3, T &arg4, T &arg5)
template<typename _TYPE_>
static void mp_sum_for_each_item (TRUSTArray< _TYPE_ > &x, int n=-1)
template<typename _TYPE_>
static void mp_max_for_each_item (TRUSTArray< _TYPE_ > &x, int n=-1)
template<typename _TYPE_>
static void mp_min_for_each_item (TRUSTArray< _TYPE_ > &x, int n=-1)
static bool mp_and (bool)
 Computes the logical AND of b across all processors in the current group.
static bool mp_or (bool)
static int check_int_overflow (trustIdType)
static int je_suis_maitre ()
 Returns 1 if on the master processor of the current group (i.e. me() == 0), 0 otherwise.
static KOKKOS_INLINE_FUNCTION void Kokkos_exit (const char *)
 Exit routine for TRUST within a Kokkos region.
static int node_master ()
 Returns 1 if on the NUMA node master processor, 0 otherwise.
static void exit (const Nom &message, int exit_code=-1)
static bool is_sequential ()
static void barrier ()
 Synchronizes all processors in the current group (waits until all processors have reached the barrier).
static void abort ()
 Abort routine for TRUST on a fatal error.
static SortieJournal (int message_level=0)
 Returns a static Sortie object used as an event journal.
static double ram_processeur ()
static void imprimer_ram_totale (int all_process=0)
static bool force_single_file (const int ranks, const Nom &filename)

Additional Inherited Members

Static Public Attributes inherited from Objet_U
static double precision_geom = 1e-10
static constexpr bool HAS_POINTER = false
static int dimension =0
static int format_precision_geom =11
static int axi =0
static int bidim_axi =0
static int DEACTIVATE_SIGINT_CATCH =0
static Type_info info_obj
static bool disable_TU =false
 Flag to disable or not the writing of the .TU files.
static bool stat_per_proc_perf_log =false
 Flag to enable the writing of the statistics detailed per processor in _csv.TU file.
Static Public Attributes inherited from Process
static int exception_sur_exit =0
static int multiple_files =5120
Protected Member Functions inherited from Objet_U_With_Params
EntreereadOn (Entree &) override final
 Reading of the input stream (dataset) with the helper class Param.
virtual void validate_params () const
 Called in the readOn of Objet_U_With_Params, after reading the params.
Protected Member Functions inherited from Objet_U
 Objet_U ()
 Default constructor: assigns a unique identifier to the object (object_id_) and registers the object in "memory" by giving it a num_obj number.
 Objet_U (const Objet_U &)
 Copy constructor.
const Objet_Uoperator= (const Objet_U &)
 Assignment operator: does nothing (the number and identifier are preserved).
virtual void set_param (Param &) const

Detailed Description

template<typename _SIZE_>
class Partitionneur_base_32_64< _SIZE_ >

Base class for domain partitioners (for splitting a mesh before a parallel computation).

This class describes the interface common to all partitioners. Example:

// Create an instance of the class: Partitionneur_xxx part; // Associate the computational domain to be split part.associer_domaine(domaine); // Initialise the parameters specific to the derived class! fichier_data >> part; // Read parameters from the data file or part.initialiser(nb_parties); // Direct initialisation (depends on the class) // If needed: part.declarer_bords_periodiques(liste_noms_bords_perio); // Build the partition array: ArrOfInt elem_part; part.construire_partition(elem_part);

WARNING: in principle, the implementation of construire_partition() must ensure that the partition is valid (applying corrections for periodic boundaries and others...)

Definition at line 51 of file Partitionneur_base.h.

Member Typedef Documentation

◆ ArrOfInt_t

template<typename _SIZE_>
using Partitionneur_base_32_64< _SIZE_ >::ArrOfInt_t = ArrOfInt_T<_SIZE_>

Definition at line 57 of file Partitionneur_base.h.

◆ BigArrOfInt_

template<typename _SIZE_>
using Partitionneur_base_32_64< _SIZE_ >::BigArrOfInt_ = TRUSTArray<int, _SIZE_>

Definition at line 64 of file Partitionneur_base.h.

◆ BigIntVect_

template<typename _SIZE_>
using Partitionneur_base_32_64< _SIZE_ >::BigIntVect_ = TRUSTVect<int, _SIZE_>

Definition at line 65 of file Partitionneur_base.h.

◆ Bord_t

template<typename _SIZE_>
using Partitionneur_base_32_64< _SIZE_ >::Bord_t = Bord_32_64<_SIZE_>

Definition at line 62 of file Partitionneur_base.h.

◆ Domaine_t

template<typename _SIZE_>
using Partitionneur_base_32_64< _SIZE_ >::Domaine_t = Domaine_32_64<_SIZE_>

Definition at line 60 of file Partitionneur_base.h.

◆ int_t

template<typename _SIZE_>
using Partitionneur_base_32_64< _SIZE_ >::int_t = _SIZE_

Definition at line 56 of file Partitionneur_base.h.

◆ IntTab_t

template<typename _SIZE_>
using Partitionneur_base_32_64< _SIZE_ >::IntTab_t = IntTab_T<_SIZE_>

Definition at line 59 of file Partitionneur_base.h.

◆ IntVect_t

template<typename _SIZE_>
using Partitionneur_base_32_64< _SIZE_ >::IntVect_t = IntVect_T<_SIZE_>

Definition at line 58 of file Partitionneur_base.h.

◆ Static_Int_Lists_t

template<typename _SIZE_>
using Partitionneur_base_32_64< _SIZE_ >::Static_Int_Lists_t = Static_Int_Lists_32_64<_SIZE_>

Definition at line 61 of file Partitionneur_base.h.

Member Function Documentation

◆ associer_domaine()

◆ calculer_graphe_connexions_periodiques()

template<typename _SIZE_>
Partitionneur_base_32_64< _SIZE_ >::int_t Partitionneur_base_32_64< _SIZE_ >::calculer_graphe_connexions_periodiques ( const Domaine_t & domaine,
const Static_Int_Lists_t & som_elem,
const int_t my_offset,
Static_Int_Lists_t & graph )
static

Computes a connectivity graph between elements connected by periodic faces.

If element i is a neighbour of element j through a periodic face, then there exists k such that graph(i,k)==j and there exists k2 such that graph(j,k2)==i.

Parameters
domaineThe domain to process.
som_elemVertex-element connectivity for the given domain. WARNING: periodic boundary faces are assumed to be ordered according to the periodic boundary convention. See check_faces_periodiques().
my_offsetGlobal element offset for this process.
graphWhere the result is stored. Return value: number of elements in the graph (equal to the number of periodic faces).

Definition at line 128 of file Partitionneur_base.cpp.

◆ construire_partition()

◆ corriger_bords_avec_graphe()

template<typename _SIZE_>
Partitionneur_base_32_64< _SIZE_ >::int_t Partitionneur_base_32_64< _SIZE_ >::corriger_bords_avec_graphe ( const Static_Int_Lists_t & graph_elements_perio,
const Static_Int_Lists_t & som_elem,
const Domaine_t & domaine,
BigIntVect_ & elem_part )
static

Corrects elem_part so that element i is on the same partition elem_part[i] as all elements connected to it in the graph.

(elements with indices graph_elements_perio(i, j) for all j).

Parameters
graph_elements_perioGraph computed by calculer_graphe_connexions_periodiques.
som_elemVertex-element connectivity.
domaineThe domain.
elem_partFor each element, which part it belongs to. Return value: number of elements whose partition was corrected.

Definition at line 604 of file Partitionneur_base.cpp.

◆ corriger_bords_avec_liste()

template<typename _SIZE_>
void Partitionneur_base_32_64< _SIZE_ >::corriger_bords_avec_liste ( const Domaine_t & dom,
const int_t my_offset,
BigIntVect_ & elem_part )
static

Computes the periodic element connectivity graphs and calls corriger_periodique_avec_graphe.

(Method to use when the connectivity graph is not yet available; if the graph is already at hand, call corriger_periodique_avec_graphe directly.)

Definition at line 656 of file Partitionneur_base.cpp.

◆ corriger_elem0_sur_proc0()

template<typename _SIZE_>
void Partitionneur_base_32_64< _SIZE_ >::corriger_elem0_sur_proc0 ( BigIntVect_ & elem_part)
static

Corrects the partition so that element 0 of the initial domain is on the first sub-domain of the partition.

The first sub-domain and the one containing element 0 are swapped.

Definition at line 49 of file Partitionneur_base.cpp.

◆ corriger_multiperiodique()

template<typename _SIZE_>
Partitionneur_base_32_64< _SIZE_ >::int_t Partitionneur_base_32_64< _SIZE_ >::corriger_multiperiodique ( const Domaine_t & domaine,
const ArrOfInt_t & renum_som_perio,
const Static_Int_Lists_t & som_elem,
BigIntVect_ & elem_part )
static

Applies corrections to elem_part so that multi-periodicity is correct:

If a vertex belongs to several periodic boundaries, all adjacent elements are assigned to the same processor.

Definition at line 395 of file Partitionneur_base.cpp.

◆ corriger_sommets_bord()

template<typename _SIZE_>
Partitionneur_base_32_64< _SIZE_ >::int_t Partitionneur_base_32_64< _SIZE_ >::corriger_sommets_bord ( const Domaine_t & domaine,
const ArrOfInt_t & renum_som_perio,
const Static_Int_Lists_t & som_elem,
BigIntVect_ & elem_part )
static

Modifies elem_part to ensure the following properties: 1) Elements that have a boundary vertex are associated.

with a processor that owns an adjacent boundary face.

2) If a processor owns a real periodic vertex, it necessarily also owns the associated renum_som_perio (hence an element that has this vertex and a periodic face). This property is essential for periodicity (existence of renum_som_perio for all vertices). For other boundaries, this correction may be unnecessary, but this is not certain. Without this correction, isolated boundary vertices can exist (a processor owns a boundary vertex but no face). If boundary vertices are searched by scanning boundary faces, the result is wrong. With this correction, that algorithm is correct.

Definition at line 236 of file Partitionneur_base.cpp.

◆ lire_motcle_non_standard()

template<typename _SIZE_>
int Partitionneur_base_32_64< _SIZE_ >::lire_motcle_non_standard ( const Motcle & motlu,
Entree & is )
inlineoverridevirtual

Reads non-simple-type parameters of an Objet_U from an input stream.

Parameters
motluThe name of the keyword to read.
isAn input stream.
Returns
Negative value if the keyword is not understood, otherwise positive value.

Reimplemented from Objet_U.

Reimplemented in Partitionneur_Metis_32_64< _SIZE_ >, Partitionneur_Parmetis, and Partitionneur_Ptscotch.

Definition at line 67 of file Partitionneur_base.h.


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