File: vtkDistributedDataFilter.h

package info (click to toggle)
vtk9 9.5.2%2Bdfsg3-4
  • links: PTS, VCS
  • area: main
  • in suites: sid
  • size: 205,916 kB
  • sloc: cpp: 2,336,565; ansic: 327,116; python: 111,200; yacc: 4,104; java: 3,977; sh: 3,032; xml: 2,771; perl: 2,189; lex: 1,787; makefile: 178; javascript: 165; objc: 153; tcl: 59
file content (251 lines) | stat: -rw-r--r-- 8,316 bytes parent folder | download | duplicates (7)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
// SPDX-FileCopyrightText: Copyright (c) Ken Martin, Will Schroeder, Bill Lorensen
// SPDX-FileCopyrightText: Copyright (c) Sandia Corporation
// SPDX-License-Identifier: BSD-3-Clause

/**
 * @class   vtkDistributedDataFilter
 * @brief   Serial stand-in for parallel filter that distributes data among processors
 *
 * This filter is a serial implementation of the vtkPDistributedDataFilter
 * filter with the intent that it can be used in non-MPI builds. This filter
 * acts almost as a "pass-through" filter when run in serial, though when the
 * input is a non-composite dataset, the output will be an vtkUnstructuredGrid
 * to be consistent with the parallel filter. The combination of this filter and
 * its parallel counterpart serves to unify the API for serial and parallel
 * builds.
 *
 * @sa vtkPDistributedDataFilter
 */

#ifndef vtkDistributedDataFilter_h
#define vtkDistributedDataFilter_h

#include "vtkDataObjectAlgorithm.h"
#include "vtkFiltersParallelModule.h" // For export macro

#include <vector> // for vector

VTK_ABI_NAMESPACE_BEGIN
class vtkBSPCuts;
class vtkMultiProcessController;
class vtkPKdTree;

class VTKFILTERSPARALLEL_EXPORT vtkDistributedDataFilter : public vtkDataObjectAlgorithm
{
public:
  vtkTypeMacro(vtkDistributedDataFilter, vtkDataObjectAlgorithm);

  void PrintSelf(ostream& os, vtkIndent indent) override;

  static vtkDistributedDataFilter* New();

  ///@{
  /**
   * Set/Get the communicator object
   */
  void SetController(vtkMultiProcessController* c);
  vtkGetObjectMacro(Controller, vtkMultiProcessController);
  ///@}

  /**
   * Get a pointer to the parallel k-d tree object.  Required for changing
   * default behavior for region assignment, changing default depth of tree,
   * or other tree building default parameters.  See vtkPKdTree and
   * vtkKdTree for more information about these options.
   * NOTE: Changing the tree returned by this method does NOT change
   * the d3 filter. Make sure to call Modified() on the d3 object if
   * you want it to re-execute.
   */

  vtkPKdTree* GetKdtree();

  /**
   * When this filter executes, it creates a vtkPKdTree (K-d tree)
   * data structure in parallel which divides the total distributed
   * data set into spatial regions.  The K-d tree object also creates
   * tables describing which processes have data for which
   * regions.  Only then does this filter redistribute
   * the data according to the region assignment scheme.  By default,
   * the K-d tree structure and it's associated tables are deleted
   * after the filter executes.  If you anticipate changing only the
   * region assignment scheme (input is unchanged) and explicitly
   * re-executing, then RetainKdTreeOn, and the K-d tree structure and
   * tables will be saved.  Then, when you re-execute, this filter will
   * skip the k-d tree build phase and go straight to redistributing
   * the data according to region assignment.  See vtkPKdTree for
   * more information about region assignment.
   */

  vtkBooleanMacro(RetainKdtree, int);
  vtkGetMacro(RetainKdtree, int);
  vtkSetMacro(RetainKdtree, int);

  /**
   * Each cell in the data set is associated with one of the
   * spatial regions of the k-d tree decomposition.  In particular,
   * the cell belongs to the region that it's centroid lies in.
   * When the new vtkUnstructuredGrid is created, by default it
   * is composed of the cells associated with the region(s)
   * assigned to this process.  If you also want it to contain
   * cells that intersect these regions, but have their centroid
   * elsewhere, then set this variable on.  By default it is off.
   */

  vtkBooleanMacro(IncludeAllIntersectingCells, int);
  vtkGetMacro(IncludeAllIntersectingCells, int);
  vtkSetMacro(IncludeAllIntersectingCells, int);

  /**
   * Set this variable if you want the cells of the output
   * vtkUnstructuredGrid to be clipped to the spatial region
   * boundaries.  By default this is off.
   */

  vtkBooleanMacro(ClipCells, int);
  vtkGetMacro(ClipCells, int);
  vtkSetMacro(ClipCells, int);

  enum BoundaryModes
  {
    ASSIGN_TO_ONE_REGION = 0,
    ASSIGN_TO_ALL_INTERSECTING_REGIONS = 1,
    SPLIT_BOUNDARY_CELLS = 2
  };

  ///@{
  /**
   * Handling of ClipCells and IncludeAllIntersectingCells.
   */
  void SetBoundaryMode(int mode);
  void SetBoundaryModeToAssignToOneRegion()
  {
    this->SetBoundaryMode(vtkDistributedDataFilter::ASSIGN_TO_ONE_REGION);
  }
  void SetBoundaryModeToAssignToAllIntersectingRegions()
  {
    this->SetBoundaryMode(vtkDistributedDataFilter::ASSIGN_TO_ALL_INTERSECTING_REGIONS);
  }
  void SetBoundaryModeToSplitBoundaryCells()
  {
    this->SetBoundaryMode(vtkDistributedDataFilter::SPLIT_BOUNDARY_CELLS);
  }
  int GetBoundaryMode();
  ///@}

  /**
   * Ensure previous filters don't send up ghost cells
   */
  int RequestUpdateExtent(vtkInformation*, vtkInformationVector**, vtkInformationVector*) override;

  /**
   * This class does a great deal of all-to-all communication
   * when exchanging portions of data sets and building new sub
   * grids.
   * By default it will do fast communication.  It can instead
   * use communication routines that use the least possible
   * amount of memory, but these are slower.  Set this option
   * ON to choose these latter routines.
   */

  vtkBooleanMacro(UseMinimalMemory, int);
  vtkGetMacro(UseMinimalMemory, int);
  vtkSetMacro(UseMinimalMemory, int);

  /**
   * The minimum number of ghost levels to add to each processor's output. If
   * the pipeline also requests ghost levels, the larger value will be used.
   */
  vtkGetMacro(MinimumGhostLevel, int);
  vtkSetMacro(MinimumGhostLevel, int);

  /**
   * Turn on collection of timing data
   */

  vtkBooleanMacro(Timing, int);
  vtkSetMacro(Timing, int);
  vtkGetMacro(Timing, int);

  /**
   * You can set the k-d tree decomposition, rather than
   * have D3 compute it.  This allows you to divide a dataset using
   * the decomposition computed for another dataset.  Obtain a description
   * of the k-d tree cuts this way:

   * vtkBSPCuts *cuts = D3Object1->GetCuts()

   * And set it this way:

   * D3Object2->SetCuts(cuts)
   */
  vtkBSPCuts* GetCuts() { return this->UserCuts; }
  void SetCuts(vtkBSPCuts* cuts);

  /**
   * vtkBSPCuts doesn't have information about process assignments for the cuts.
   * Typically D3 filter simply reassigns the processes for each cut. However,
   * that may not always work, sometimes the processes have be pre-assigned and
   * we want to preserve that partitioning. In that case, one sets the region
   * assignments explicitly. Look at vtkPKdTree::AssignRegions for details about
   * the arguments. Calling SetUserRegionAssignments(nullptr, 0) will revert to
   * default behavior i.e. letting the KdTree come up with the assignments.
   */
  void SetUserRegionAssignments(const int* map, int numRegions);

protected:
  vtkDistributedDataFilter();
  ~vtkDistributedDataFilter() override;

  /**
   * Build a vtkUnstructuredGrid to store the input.
   */
  int RequestData(vtkInformation*, vtkInformationVector**, vtkInformationVector*) override;
  int RequestInformation(vtkInformation*, vtkInformationVector**, vtkInformationVector*) override;
  int FillInputPortInformation(int port, vtkInformation* info) override;

  /**
   * Overridden to create the correct type of data output. If input is dataset,
   * output is vtkUnstructuredGrid. If input is composite dataset, output is
   * vtkMultiBlockDataSet.
   */
  int RequestDataObject(vtkInformation*, vtkInformationVector**, vtkInformationVector*) override;

  vtkPKdTree* Kdtree;
  vtkMultiProcessController* Controller;

  int NumProcesses;
  int MyId;

  int* Target;
  int* Source;

  int NumConvexSubRegions;
  double* ConvexSubRegionBounds;

  // User-adjustable minimum number of ghost levels.
  int MinimumGhostLevel;

  // Actual number of ghost levels used during execution.
  int GhostLevel;

  int RetainKdtree;
  int IncludeAllIntersectingCells;
  int ClipCells;

  int Timing;

  int NextProgressStep;
  double ProgressIncrement;

  int UseMinimalMemory;

  vtkBSPCuts* UserCuts;

  std::vector<int> UserRegionAssignments;

private:
  vtkDistributedDataFilter(const vtkDistributedDataFilter&) = delete;
  void operator=(const vtkDistributedDataFilter&) = delete;
};
VTK_ABI_NAMESPACE_END
#endif