Reorganize clustering benchmarks

Author: aprokop

benchmarks/clustering/CMakeLists.txt

@@ -1,26 +1,27 @@
-set(EXPLICIT_INSTANTIATION_SOURCE_FILES)
-set(TEMPLATE_PARAMETERS 2 3 4 5 6)
-foreach(DIM ${TEMPLATE_PARAMETERS})
-  set(filename ${CMAKE_CURRENT_BINARY_DIR}/dbscan_${DIM}.cpp)
-  file(WRITE ${filename}
-    "#include \"${CMAKE_CURRENT_SOURCE_DIR}/dbscan_timpl.hpp\"\n"
-    "template bool ArborXBenchmark::run<${DIM}>(ArborXBenchmark::Parameters const&);\n"
-  )
-  list(APPEND EXPLICIT_INSTANTIATION_SOURCE_FILES ${filename})
-endforeach()
-
-add_executable(ArborX_Benchmark_DBSCAN.exe
-  ${EXPLICIT_INSTANTIATION_SOURCE_FILES}
+add_library(cluster_benchmark_helpers
   print_timers.cpp
-  dbscan.cpp
 )
+target_compile_features(cluster_benchmark_helpers PRIVATE cxx_std_17)
+target_link_libraries(cluster_benchmark_helpers PRIVATE Kokkos::kokkos)
+
+set(input_file "input.txt")
+configure_file(${CMAKE_CURRENT_SOURCE_DIR}/${input_file} ${CMAKE_CURRENT_BINARY_DIR}/${input_file} COPYONLY)
+
+add_executable(ArborX_Benchmark_DBSCAN.exe dbscan.cpp)
 target_include_directories(ArborX_Benchmark_DBSCAN.exe PRIVATE ${CMAKE_CURRENT_SOURCE_DIR})
-target_link_libraries(ArborX_Benchmark_DBSCAN.exe ArborX::ArborX Boost::program_options)
+target_link_libraries(ArborX_Benchmark_DBSCAN.exe ArborX::ArborX Boost::program_options cluster_benchmark_helpers)
+add_test(NAME ArborX_Benchmark_DBSCAN COMMAND ArborX_Benchmark_DBSCAN.exe --filename=${input_file} --eps=1.4 --verify)
+
+add_executable(ArborX_Benchmark_MST.exe mst.cpp)
+target_include_directories(ArborX_Benchmark_MST.exe PRIVATE ${CMAKE_CURRENT_SOURCE_DIR})
+target_link_libraries(ArborX_Benchmark_MST.exe ArborX::ArborX Boost::program_options cluster_benchmark_helpers)
+add_test(NAME ArborX_Benchmark_HDBSCAN COMMAND ArborX_Benchmark_HDBSCAN.exe --filename=${input_file})
+
+add_executable(ArborX_Benchmark_HDBSCAN.exe hdbscan.cpp)
+target_include_directories(ArborX_Benchmark_HDBSCAN.exe PRIVATE ${CMAKE_CURRENT_SOURCE_DIR})
+target_link_libraries(ArborX_Benchmark_HDBSCAN.exe ArborX::ArborX Boost::program_options cluster_benchmark_helpers)
+add_test(NAME ArborX_Benchmark_MST COMMAND ArborX_Benchmark_MST.exe --filename=${input_file})
 
 add_executable(ArborX_DataConverter.exe converter.cpp)
 target_compile_features(ArborX_DataConverter.exe PRIVATE cxx_std_17)
 target_link_libraries(ArborX_DataConverter.exe Boost::program_options)
-
-set(input_file "input.txt")
-add_test(NAME ArborX_Benchmark_DBSCAN COMMAND ArborX_Benchmark_DBSCAN.exe --filename=${input_file} --eps=1.4 --verify)
-configure_file(${CMAKE_CURRENT_SOURCE_DIR}/${input_file} ${CMAKE_CURRENT_BINARY_DIR}/${input_file} COPYONLY)

benchmarks/clustering/data.hpp

@@ -15,10 +15,13 @@
 #include <ArborX_Point.hpp>
 #include <misc/ArborX_Exception.hpp>
 
+#include <fstream>
 #include <iostream>
 #include <random>
 #include <vector>
 
+#include "parameters.hpp"
+
 using ArborX::Point;
 
 template <int DIM>
@@ -51,6 +54,33 @@ std::vector<Point<DIM>> sampleData(std::vector<Point<DIM>> const &data,
   return sampled_data;
 }
 
+static int getDataDimension(std::string const &filename, bool binary)
+{
+  std::ifstream input;
+  if (!binary)
+    input.open(filename);
+  else
+    input.open(filename, std::ifstream::binary);
+  if (!input.good())
+    throw std::runtime_error("Error reading file \"" + filename + "\"");
+
+  int num_points;
+  int dim;
+  if (!binary)
+  {
+    input >> num_points;
+    input >> dim;
+  }
+  else
+  {
+    input.read(reinterpret_cast<char *>(&num_points), sizeof(int));
+    input.read(reinterpret_cast<char *>(&dim), sizeof(int));
+  }
+  input.close();
+
+  return dim;
+}
+
 template <int DIM>
 std::vector<Point<DIM>> loadData(std::string const &filename,
                                  bool binary = true, int max_num_points = -1,
@@ -298,4 +328,41 @@ std::vector<Point<DIM>> GanTao(int n, bool variable_density = false,
   return points;
 }
 
+template <typename... P, typename T>
+auto vec2view(std::vector<T> const &in, std::string const &label = "")
+{
+  Kokkos::View<T *, P...> out(
+      Kokkos::view_alloc(label, Kokkos::WithoutInitializing), in.size());
+  Kokkos::deep_copy(out, Kokkos::View<T const *, Kokkos::HostSpace,
+                                      Kokkos::MemoryTraits<Kokkos::Unmanaged>>{
+                             in.data(), in.size()});
+  return out;
+}
+
+template <int DIM, typename MemorySpace>
+auto loadData(ArborXBenchmark::Parameters const &params)
+{
+  if (!params.filename.empty())
+  {
+    // Read in data
+    printf("filename          : %s [%s, max_pts = %d]\n",
+           params.filename.c_str(), (params.binary ? "binary" : "text"),
+           params.max_num_points);
+    printf("samples           : %d\n", params.num_samples);
+    return vec2view<MemorySpace>(loadData<DIM>(params.filename, params.binary,
+                                               params.max_num_points,
+                                               params.num_samples),
+                                 "Benchmark::primitives");
+  }
+  else
+  {
+    // Generate data
+    int dim = params.dim;
+    printf("generator         : n = %d, dim = %d, density = %s\n", params.n,
+           dim, (params.variable_density ? "variable" : "constant"));
+    return vec2view<MemorySpace>(GanTao<DIM>(params.n, params.variable_density),
+                                 "Benchmark::primitives");
+  }
+}
+
 #endif