Further simplify benchmarks. Add benchmarks for unordered pixel iteration

robomics · robomics · commit 19b73649f1ac · 2024-11-03T15:02:52.000+01:00
diff --git a/benchmark/interaction_fetching/common.hpp b/benchmark/interaction_fetching/common.hpp
@@ -21,21 +21,6 @@
 #include "hictk/chromosome.hpp"
 #include "hictk/reference.hpp"
 
-// NOLINTBEGIN(*-avoid-magic-numbers)
-struct Params {
-  std::string_view label{};
-  bool cis{};
-
-  double avg_height{1.0e6};
-  double avg_width{1.0e6};
-  double height_std{250.0e3};
-  double width_std{250.0e3};
-  std::size_t num_queries{1};
-  hictk::balancing::Method normalization{hictk::balancing::Method::NONE()};
-  std::uint64_t seed{123456789};
-};
-// NOLINTEND(*-avoid-magic-numbers)
-
 [[nodiscard]] inline std::pair<std::string, std::string> generate_query(
     std::mt19937_64& rand_eng, const hictk::Chromosome& chrom1, const hictk::Chromosome& chrom2,
     double avg_height, double avg_width, double height_std, double width_std) {
@@ -158,3 +143,26 @@ template <typename N, typename File>
   // clang-format on
   return i;
 }
+
+template <typename N, typename File>
+[[nodiscard]] inline std::ptrdiff_t count_nnz_unsorted(
+    const File& file, std::string_view range1, std::string_view range2,
+    const hictk::balancing::Method& normalization) {
+  const auto sel = file.fetch(range1, range2, normalization);
+
+  return std::distance(sel.template begin<N>(false), sel.template end<N>());
+}
+
+template <typename N, typename File>
+[[nodiscard]] inline std::ptrdiff_t count_nnz_unsorted(
+    const File& file, std::size_t max_num_pixels, const hictk::balancing::Method& normalization) {
+  const auto sel = file.fetch(normalization);
+  auto first = sel.template begin<N>(false);
+  auto last = sel.template end<N>();
+
+  std::ptrdiff_t i{};
+  // clang-format off
+  while (++first != last && ++i != static_cast<std::ptrdiff_t>(max_num_pixels));  // NOLINT
+  // clang-format on
+  return i;
+}
diff --git a/benchmark/interaction_fetching/cooler_cis_queries.cpp b/benchmark/interaction_fetching/cooler_cis_queries.cpp
@@ -27,38 +27,28 @@ static constexpr std::string_view range_small{"chr2L:5,000,000-5,100,000"};
 static constexpr std::string_view range_medium{"chr2L:6,000,000-7,000,000"};
 static constexpr std::string_view range_large{"chr2L:10,000,000-15,000,000"};
 
-TEST_CASE("cooler::File::fetch (cis; uint32)") {
-  const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
-                                               test_file.string(), resolutions.back()))
-                          .chromosomes();
-  for (const auto& res : resolutions) {
-    for (const auto& range : {range_small, range_medium, range_large}) {
-      BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}bp"), range, res))
-      (Catch::Benchmark::Chronometer meter) {
-        const cooler::File clr(
-            fmt::format(FMT_STRING("{}::/resolutions/{}"), test_file.string(), res));
-        meter.measure([&clr, &range]() {
-          return count_nnz<std::uint32_t>(clr, range, range, balancing::Method::NONE());
-        });
-      };
-    }
-  }
+template <typename N>
+static void run_benchmark(const std::filesystem::path& path, std::uint32_t resolution,
+                          std::string_view range, const balancing::Method& normalization) {
+  BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}bp; {}"), range, resolution,
+                                 std::is_integral_v<N> ? "int" : "fp"))
+  (Catch::Benchmark::Chronometer meter) {
+    const cooler::File clr(
+        fmt::format(FMT_STRING("{}::/resolutions/{}"), path.string(), resolution));
+    meter.measure([&clr, &range, &normalization]() {
+      return count_nnz<N>(clr, range, range, normalization);
+    });
+  };
 }
 
-TEST_CASE("cooler::File::fetch (cis; double)") {
+TEST_CASE("cooler::File::fetch (cis)") {
   const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
                                                test_file.string(), resolutions.back()))
                           .chromosomes();
   for (const auto& res : resolutions) {
     for (const auto& range : {range_small, range_medium, range_large}) {
-      BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}bp"), range, res))
-      (Catch::Benchmark::Chronometer meter) {
-        const cooler::File clr(
-            fmt::format(FMT_STRING("{}::/resolutions/{}"), test_file.string(), res));
-        meter.measure([&clr, &range]() {
-          return count_nnz<double>(clr, range, range, balancing::Method::KR());
-        });
-      };
+      run_benchmark<std::uint32_t>(test_file, res, range, balancing::Method::NONE());
+      run_benchmark<double>(test_file, res, range, balancing::Method::KR());
     }
   }
 }
diff --git a/benchmark/interaction_fetching/cooler_gw_queries.cpp b/benchmark/interaction_fetching/cooler_gw_queries.cpp
@@ -21,34 +21,26 @@ static const std::filesystem::path test_file{"test/data/integration_tests/4DNFIZ
 static const std::vector<std::uint32_t> resolutions{1000,   5000,   10000,  25000,   50000,
                                                     100000, 250000, 500000, 1000000, 2500000};
 
-TEST_CASE("cooler::File::fetch (gw; uint32)") {
-  const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
-                                               test_file.string(), resolutions.back()))
-                          .chromosomes();
-  for (const auto& res : resolutions) {
-    BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}bp"), res))
-    (Catch::Benchmark::Chronometer meter) {
-      const cooler::File clr(
-          fmt::format(FMT_STRING("{}::/resolutions/{}"), test_file.string(), res));
-      meter.measure([&clr]() {
-        return count_nnz<std::uint32_t>(clr, 10'000'000, balancing::Method::NONE());
-      });
-    };
-  }
+template <typename N>
+static void run_benchmark(const std::filesystem::path& path, std::uint32_t resolution,
+                          const balancing::Method& normalization) {
+  BENCHMARK_ADVANCED(
+      fmt::format(FMT_STRING("{}bp; {}"), resolution, std::is_integral_v<N> ? "int" : "fp"))
+  (Catch::Benchmark::Chronometer meter) {
+    const cooler::File clr(
+        fmt::format(FMT_STRING("{}::/resolutions/{}"), path.string(), resolution));
+    meter.measure(
+        [&clr, &normalization]() { return count_nnz<N>(clr, 10'000'000, normalization); });
+  };
 }
 
-TEST_CASE("cooler::File::fetch (gw; double)") {
+TEST_CASE("cooler::File::fetch (gw)") {
   const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
                                                test_file.string(), resolutions.back()))
                           .chromosomes();
   for (const auto& res : resolutions) {
-    BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}bp"), res))
-    (Catch::Benchmark::Chronometer meter) {
-      const cooler::File clr(
-          fmt::format(FMT_STRING("{}::/resolutions/{}"), test_file.string(), res));
-      meter.measure(
-          [&clr]() { return count_nnz<std::uint32_t>(clr, 10'000'000, balancing::Method::KR()); });
-    };
+    run_benchmark<std::uint32_t>(test_file, res, balancing::Method::NONE());
+    run_benchmark<double>(test_file, res, balancing::Method::KR());
   }
 }
 // NOLINTEND(*-avoid-magic-numbers, cert-err58-cpp, readability-function-cognitive-complexity)
diff --git a/benchmark/interaction_fetching/cooler_trans_queries.cpp b/benchmark/interaction_fetching/cooler_trans_queries.cpp
@@ -32,38 +32,30 @@ static constexpr std::pair<std::string_view, std::string_view> range_medium{
 static constexpr std::pair<std::string_view, std::string_view> range_large{
     "chr2L:15,000,000-20,000,000", "chrX:15,000,000-20,000,000"};
 
-TEST_CASE("cooler::File::fetch (trans; uint32)") {
-  const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
-                                               test_file.string(), resolutions.back()))
-                          .chromosomes();
-  for (const auto& res : resolutions) {
-    for (const auto& query : {range_small, range_medium, range_large}) {
-      BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}; {}bp"), query.first, query.second, res))
-      (Catch::Benchmark::Chronometer meter) {
-        const cooler::File clr(
-            fmt::format(FMT_STRING("{}::/resolutions/{}"), test_file.string(), res));
-        meter.measure([&clr, range1 = query.first, range2 = query.second]() {
-          return count_nnz<std::uint32_t>(clr, range1, range2, balancing::Method::NONE());
-        });
-      };
-    }
-  }
+template <typename N>
+static void run_benchmark(const std::filesystem::path& path, std::uint32_t resolution,
+                          std::string_view range1, std::string_view range2,
+                          const balancing::Method& normalization) {
+  BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}; {}bp; {}"), range1, range2, resolution,
+                                 std::is_integral_v<N> ? "int" : "fp"))
+  (Catch::Benchmark::Chronometer meter) {
+    const cooler::File clr(
+        fmt::format(FMT_STRING("{}::/resolutions/{}"), path.string(), resolution));
+    meter.measure([&clr, &range1, &range2, &normalization]() {
+      return count_nnz<N>(clr, range1, range2, normalization);
+    });
+  };
 }
 
-TEST_CASE("cooler::File::fetch (trans; double)") {
+TEST_CASE("cooler::File::fetch (trans)") {
   const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
                                                test_file.string(), resolutions.back()))
                           .chromosomes();
   for (const auto& res : resolutions) {
     for (const auto& query : {range_small, range_medium, range_large}) {
-      BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}; {}bp"), query.first, query.second, res))
-      (Catch::Benchmark::Chronometer meter) {
-        const cooler::File clr(
-            fmt::format(FMT_STRING("{}::/resolutions/{}"), test_file.string(), res));
-        meter.measure([&clr, range1 = query.first, range2 = query.second]() {
-          return count_nnz<double>(clr, range1, range2, balancing::Method::KR());
-        });
-      };
+      run_benchmark<std::uint32_t>(test_file, res, query.first, query.second,
+                                   balancing::Method::NONE());
+      run_benchmark<double>(test_file, res, query.first, query.second, balancing::Method::KR());
     }
   }
 }
diff --git a/benchmark/interaction_fetching/file_cis_queries.cpp b/benchmark/interaction_fetching/file_cis_queries.cpp
@@ -30,41 +30,28 @@ static constexpr std::string_view range_small{"chr2L:5,000,000-5,100,000"};
 static constexpr std::string_view range_medium{"chr2L:6,000,000-7,000,000"};
 static constexpr std::string_view range_large{"chr2L:10,000,000-15,000,000"};
 
-TEST_CASE("File::fetch (cis; uint32)") {
-  const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
-                                               test_file1.string(), resolutions.back()))
-                          .chromosomes();
-
-  for (const auto& path : {test_file1, test_file2, test_file3}) {
-    for (const auto& res : resolutions) {
-      for (const auto& range : {range_small, range_medium, range_large}) {
-        BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}; {}bp"), path.extension(), range, res))
-        (Catch::Benchmark::Chronometer meter) {
-          const File f(path.string(), res);
-          meter.measure([&f, &range]() {
-            return count_nnz<std::uint32_t>(f, range, range, balancing::Method::NONE());
-          });
-        };
-      }
-    }
-  }
+template <typename N>
+static void run_benchmark(const std::filesystem::path& path, std::uint32_t resolution,
+                          std::string_view range, const balancing::Method& normalization) {
+  BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}; {}bp; {}"), path.extension(), range,
+                                 resolution, std::is_integral_v<N> ? "int" : "fp"))
+  (Catch::Benchmark::Chronometer meter) {
+    const File f(path.string(), resolution);
+    meter.measure(
+        [&f, &range, &normalization]() { return count_nnz<N>(f, range, range, normalization); });
+  };
 }
 
-TEST_CASE("File::fetch (cis; double)") {
+TEST_CASE("File::fetch (cis)") {
   const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
                                                test_file1.string(), resolutions.back()))
                           .chromosomes();
 
   for (const auto& path : {test_file1, test_file2, test_file3}) {
     for (const auto& res : resolutions) {
       for (const auto& range : {range_small, range_medium, range_large}) {
-        BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}; {}bp"), path.extension(), range, res))
-        (Catch::Benchmark::Chronometer meter) {
-          const File f(path.string(), res);
-          meter.measure([&f, &range]() {
-            return count_nnz<double>(f, range, range, balancing::Method::KR());
-          });
-        };
+        run_benchmark<std::uint32_t>(path, res, range, balancing::Method::NONE());
+        run_benchmark<double>(path, res, range, balancing::Method::KR());
       }
     }
   }
diff --git a/benchmark/interaction_fetching/file_gw_queries.cpp b/benchmark/interaction_fetching/file_gw_queries.cpp
@@ -24,36 +24,26 @@ static const std::filesystem::path test_file3{"test/data/hic/4DNFIZ1ZVXC8.hic9"}
 static const std::vector<std::uint32_t> resolutions{1000,   5000,   10000,  25000,   50000,
                                                     100000, 250000, 500000, 1000000, 2500000};
 
-TEST_CASE("File::fetch (gw; uint32)") {
-  const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
-                                               test_file1.string(), resolutions.back()))
-                          .chromosomes();
-
-  for (const auto& path : {test_file1, test_file2, test_file3}) {
-    for (const auto& res : resolutions) {
-      BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}bp"), path.extension(), res))
-      (Catch::Benchmark::Chronometer meter) {
-        const File f(path.string(), res);
-        meter.measure(
-            [&f]() { return count_nnz<std::uint32_t>(f, 10'000'000, balancing::Method::NONE()); });
-      };
-    }
-  }
+template <typename N>
+static void run_benchmark(const std::filesystem::path& path, std::uint32_t resolution,
+                          const balancing::Method& normalization) {
+  BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}bp; {}"), path.extension(), resolution,
+                                 std::is_integral_v<N> ? "int" : "fp"))
+  (Catch::Benchmark::Chronometer meter) {
+    const File f(path.string(), resolution);
+    meter.measure([&f, &normalization]() { return count_nnz<N>(f, 10'000'000, normalization); });
+  };
 }
 
-TEST_CASE("File::fetch (gw; double)") {
+TEST_CASE("File::fetch (gw)") {
   const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
                                                test_file1.string(), resolutions.back()))
                           .chromosomes();
 
   for (const auto& path : {test_file1, test_file2, test_file3}) {
     for (const auto& res : resolutions) {
-      BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}bp"), path.extension(), res))
-      (Catch::Benchmark::Chronometer meter) {
-        const File f(path.string(), res);
-        meter.measure(
-            [&f]() { return count_nnz<std::uint32_t>(f, 10'000'000, balancing::Method::KR()); });
-      };
+      run_benchmark<std::uint32_t>(path, res, balancing::Method::NONE());
+      run_benchmark<double>(path, res, balancing::Method::KR());
     }
   }
 }
diff --git a/benchmark/interaction_fetching/file_trans_queries.cpp b/benchmark/interaction_fetching/file_trans_queries.cpp
@@ -35,41 +35,31 @@ static constexpr std::pair<std::string_view, std::string_view> range_medium{
 static constexpr std::pair<std::string_view, std::string_view> range_large{
     "chr2L:15,000,000-20,000,000", "chrX:15,000,000-20,000,000"};
 
-TEST_CASE("File::fetch (trans; uint32)") {
-  const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
-                                               test_file1.string(), resolutions.back()))
-                          .chromosomes();
-
-  for (const auto& path : {test_file1, test_file2, test_file3}) {
-    for (const auto& res : resolutions) {
-      for (const auto& query : {range_small, range_medium, range_large}) {
-        BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}; {}bp"), query.first, query.second, res))
-        (Catch::Benchmark::Chronometer meter) {
-          const File f(path.string(), res);
-          meter.measure([&f, range1 = query.first, range2 = query.second]() {
-            return count_nnz<std::uint32_t>(f, range1, range2, balancing::Method::NONE());
-          });
-        };
-      }
-    }
-  }
+template <typename N>
+static void run_benchmark(const std::filesystem::path& path, std::uint32_t resolution,
+                          std::string_view range1, std::string_view range2,
+                          const balancing::Method& normalization) {
+  BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}; {}; {}bp; {}"), path.extension(), range1,
+                                 range2, resolution, std::is_integral_v<N> ? "int" : "fp"))
+  (Catch::Benchmark::Chronometer meter) {
+    const File f(path.string(), resolution);
+    meter.measure([&f, &range1, &range2, &normalization]() {
+      return count_nnz<N>(f, range1, range2, normalization);
+    });
+  };
 }
 
-TEST_CASE("File::fetch (trans; double)") {
+TEST_CASE("File::fetch (trans)") {
   const auto chroms = cooler::File(fmt::format(FMT_STRING("{}::/resolutions/{}"),
                                                test_file1.string(), resolutions.back()))
                           .chromosomes();
 
   for (const auto& path : {test_file1, test_file2, test_file3}) {
     for (const auto& res : resolutions) {
       for (const auto& query : {range_small, range_medium, range_large}) {
-        BENCHMARK_ADVANCED(fmt::format(FMT_STRING("{}; {}; {}bp"), query.first, query.second, res))
-        (Catch::Benchmark::Chronometer meter) {
-          const File f(path.string(), res);
-          meter.measure([&f, range1 = query.first, range2 = query.second]() {
-            return count_nnz<double>(f, range1, range2, balancing::Method::KR());
-          });
-        };
+        run_benchmark<std::uint32_t>(path, res, query.first, query.second,
+                                     balancing::Method::NONE());
+        run_benchmark<double>(path, res, query.first, query.second, balancing::Method::KR());
       }
     }
   }
diff --git a/benchmark/interaction_fetching/hic_cis_queries.cpp b/benchmark/interaction_fetching/hic_cis_queries.cpp
diff --git a/benchmark/interaction_fetching/hic_gw_queries.cpp b/benchmark/interaction_fetching/hic_gw_queries.cpp
diff --git a/benchmark/interaction_fetching/hic_trans_queries.cpp b/benchmark/interaction_fetching/hic_trans_queries.cpp
