File: itkGPUGradientAnisotropicDiffusionImageFilterTest.cxx

package info (click to toggle)
insighttoolkit5 5.4.3-5
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid, trixie
  • size: 704,384 kB
  • sloc: cpp: 783,592; ansic: 628,724; xml: 44,704; fortran: 34,250; python: 22,874; sh: 4,078; pascal: 2,636; lisp: 2,158; makefile: 464; yacc: 328; asm: 205; perl: 203; lex: 146; tcl: 132; javascript: 98; csh: 81
file content (223 lines) | stat: -rw-r--r-- 7,055 bytes parent folder | download
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
/*=========================================================================
 *
 *  Copyright NumFOCUS
 *
 *  Licensed under the Apache License, Version 2.0 (the "License");
 *  you may not use this file except in compliance with the License.
 *  You may obtain a copy of the License at
 *
 *         https://www.apache.org/licenses/LICENSE-2.0.txt
 *
 *  Unless required by applicable law or agreed to in writing, software
 *  distributed under the License is distributed on an "AS IS" BASIS,
 *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 *  See the License for the specific language governing permissions and
 *  limitations under the License.
 *
 *=========================================================================*/

/**
 * Test program for GPUGradientAnisotropicDiffusionImageFilter class
 */
#include "itkImageFileReader.h"
#include "itkImageFileWriter.h"
#include "itkGradientAnisotropicDiffusionImageFilter.h"
#include "itkTimeProbe.h"
#include "itkImageRegionIterator.h"
#include "itkTestingMacros.h"

#include "itkOpenCLUtil.h"
#include "itkGPUImage.h"
#include "itkGPUKernelManager.h"
#include "itkGPUContextManager.h"
#include "itkGPUImageToImageFilter.h"
#include "itkGPUGradientAnisotropicDiffusionImageFilter.h"

template <unsigned int VImageDimension>
int
runGPUGradientAnisotropicDiffusionImageFilterTest(const std::string & inFile, const std::string & outFile)
{
  using InputPixelType = float;
  using OutputPixelType = float;

  using InputImageType = itk::GPUImage<InputPixelType, VImageDimension>;
  using OutputImageType = itk::GPUImage<OutputPixelType, VImageDimension>;

  using ReaderType = itk::ImageFileReader<InputImageType>;
  using WriterType = itk::ImageFileWriter<OutputImageType>;

  auto reader = ReaderType::New();
  auto writer = WriterType::New();

  reader->SetFileName(inFile);
  writer->SetFileName(outFile);

  // Create CPU/GPU anistorpic diffusion filter
  using CPUAnisoDiffFilterType = itk::GradientAnisotropicDiffusionImageFilter<InputImageType, OutputImageType>;
  using GPUAnisoDiffFilterType = itk::GPUGradientAnisotropicDiffusionImageFilter<InputImageType, OutputImageType>;

  auto CPUFilter = CPUAnisoDiffFilterType::New();
  auto GPUFilter = GPUAnisoDiffFilterType::New();

  ITK_EXERCISE_BASIC_OBJECT_METHODS(
    GPUFilter, GPUGradientAnisotropicDiffusionImageFilter, GPUAnisotropicDiffusionImageFilter);

  reader->Update();

  // -------

  // test 1~8 threads for CPU
  std::vector<int> nThreadVec;
  nThreadVec.push_back(1);
  nThreadVec.push_back(2);
  nThreadVec.push_back(4);
  nThreadVec.push_back(8);
  nThreadVec.push_back(128);

  for (unsigned int idx = 0; idx < nThreadVec.size(); ++idx)
  {
    int            numberOfWorkUnits = nThreadVec[idx];
    itk::TimeProbe cputimer;
    cputimer.Start();

    CPUFilter->SetNumberOfWorkUnits(numberOfWorkUnits);

    CPUFilter->SetInput(reader->GetOutput());
    CPUFilter->SetNumberOfIterations(10);
    CPUFilter->SetTimeStep(0.0625); // 125 );
    CPUFilter->SetConductanceParameter(3.0);
    CPUFilter->UseImageSpacingOn();
    CPUFilter->Update();

    cputimer.Stop();

    std::cout << "CPU Anisotropic diffusion took " << cputimer.GetMean() << " seconds with "
              << CPUFilter->GetNumberOfWorkUnits() << " work units.\n"
              << std::endl;

    // -------

    if (idx == nThreadVec.size() - 1)
    {
      itk::TimeProbe gputimer;
      gputimer.Start();

      GPUFilter->SetInput(reader->GetOutput());
      GPUFilter->SetNumberOfIterations(10);
      GPUFilter->SetTimeStep(0.0625); // 125 );
      GPUFilter->SetConductanceParameter(3.0);
      GPUFilter->UseImageSpacingOn();
      try
      {
        GPUFilter->Update();
      }
      catch (const itk::ExceptionObject & excp)
      {
        std::cout << "Caught exception during GPUFilter->Update() " << excp << std::endl;
        return EXIT_FAILURE;
      }

      try
      {
        GPUFilter->GetOutput()->UpdateBuffers(); // synchronization point
      }
      catch (const itk::ExceptionObject & excp)
      {
        std::cout << "Caught exception during GPUFilter->GetOutput()->UpdateBuffers() " << excp << std::endl;
        return EXIT_FAILURE;
      }


      gputimer.Stop();
      std::cout << "GPU Anisotropic diffusion took " << gputimer.GetMean() << " seconds.\n" << std::endl;

      // ---------------
      // RMS Error check
      // ---------------

      double                                    diff = 0;
      unsigned int                              nPix = 0;
      itk::ImageRegionIterator<OutputImageType> cit(CPUFilter->GetOutput(),
                                                    CPUFilter->GetOutput()->GetLargestPossibleRegion());
      itk::ImageRegionIterator<OutputImageType> git(GPUFilter->GetOutput(),
                                                    GPUFilter->GetOutput()->GetLargestPossibleRegion());

      for (cit.GoToBegin(), git.GoToBegin(); !cit.IsAtEnd(); ++cit, ++git)
      {
        double err = static_cast<double>(cit.Get()) - static_cast<double>(git.Get());
        diff += err * err;
        nPix++;
      }
      writer->SetInput(GPUFilter->GetOutput()); // copy GPU->CPU implicilty

      // execute pipeline filter and write output
      writer->Update();

      if (nPix > 0)
      {
        double RMSError = sqrt(diff / static_cast<double>(nPix));
        std::cout << "RMS Error : " << RMSError << std::endl;
        double RMSThreshold = 10.0;
        if (itk::Math::isnan(RMSError))
        {
          std::cout << "RMS Error is NaN! nPix: " << nPix << std::endl;
          return EXIT_FAILURE;
        }
        if (RMSError > RMSThreshold)
        {
          std::cout << "RMS Error exceeds threshold (" << RMSThreshold << ')' << std::endl;
          return EXIT_FAILURE;
        }
      }
      else
      {
        std::cout << "No pixels in output!" << std::endl;
        return EXIT_FAILURE;
      }
    }
  }

  GPUFilter = nullptr;                                      // explicit GPU object destruction test
  itk::GPUContextManager::GetInstance()->DestroyInstance(); // GPUContextManager singleton destruction test
  return EXIT_SUCCESS;
}

int
itkGPUGradientAnisotropicDiffusionImageFilterTest(int argc, char * argv[])
{
  if (!itk::IsGPUAvailable())
  {
    std::cerr << "OpenCL-enabled GPU is not present." << std::endl;
    return EXIT_FAILURE;
  }

  if (argc < 3)
  {
    std::cerr << "Error: missing arguments" << std::endl;
    std::cerr << "inputfile outputfile [num_dimensions]" << std::endl;
    return EXIT_FAILURE;
  }

  std::string inFile(argv[1]);
  std::string outFile(argv[2]);

  unsigned int dim = 3;
  if (argc >= 4)
  {
    dim = std::stoi(argv[3]);
  }

  if (dim == 2)
  {
    return runGPUGradientAnisotropicDiffusionImageFilterTest<2>(inFile, outFile);
  }
  else if (dim == 3)
  {
    return runGPUGradientAnisotropicDiffusionImageFilterTest<3>(inFile, outFile);
  }
  else
  {
    std::cerr << "Error: only 2 or 3 dimensions allowed, " << dim << " selected." << std::endl;
    return EXIT_FAILURE;
  }
}