| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #include "nalitov_d_dijkstras_algorithm/omp/include/ops_omp.hpp" | ||
| 2 | |||
| 3 | #include <omp.h> | ||
| 4 | |||
| 5 | #include <algorithm> | ||
| 6 | #include <cstdint> | ||
| 7 | #include <limits> | ||
| 8 | #include <vector> | ||
| 9 | |||
| 10 | #include "nalitov_d_dijkstras_algorithm/common/include/common.hpp" | ||
| 11 | |||
| 12 | namespace nalitov_d_dijkstras_algorithm { | ||
| 13 | |||
| 14 | namespace { | ||
| 15 | |||
| 16 | inline bool SafeAdd(InType a, InType b, InType &result) { | ||
| 17 | if (a > 0 && b > std::numeric_limits<InType>::max() - a) { | ||
| 18 | return false; | ||
| 19 | } | ||
| 20 | if (a < 0 && b < std::numeric_limits<InType>::min() - a) { | ||
| 21 | return false; | ||
| 22 | } | ||
| 23 | result = a + b; | ||
| 24 | return true; | ||
| 25 | } | ||
| 26 | |||
| 27 | inline InType GetEdgeWeight(InType from, InType to) { | ||
| 28 | if (from == to) { | ||
| 29 | return 0; | ||
| 30 | } | ||
| 31 | return (from > to) ? (from - to) : (to - from); | ||
| 32 | } | ||
| 33 | |||
| 34 | int GetWorkerCount() { | ||
| 35 | int worker_count = 1; | ||
| 36 | 12 | #pragma omp parallel default(none) shared(worker_count) | |
| 37 | { | ||
| 38 | #pragma omp single | ||
| 39 | worker_count = omp_get_num_threads(); | ||
| 40 | } | ||
| 41 | return worker_count; | ||
| 42 | } | ||
| 43 | |||
| 44 | 48 | InType SelectNextVertexOmp(std::vector<InType> &distances, std::vector<char> &processed, InType k_infinity, | |
| 45 | int worker_count) { | ||
| 46 | 48 | const auto vertex_count = static_cast<InType>(distances.size()); | |
| 47 | 48 | std::vector<InType> thread_best_distance(worker_count, k_infinity); | |
| 48 |
1/4✓ Branch 1 taken 48 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
48 | std::vector<InType> thread_best_vertex(worker_count, -1); |
| 49 | InType selected_vertex = -1; | ||
| 50 | |||
| 51 |
1/2✓ Branch 0 taken 48 times.
✗ Branch 1 not taken.
|
48 | #pragma omp parallel default(none) shared(vertex_count, distances, processed, thread_best_distance, \ |
| 52 | thread_best_vertex, worker_count, selected_vertex, k_infinity) | ||
| 53 | { | ||
| 54 | const int tid = omp_get_thread_num(); | ||
| 55 | InType best_distance = std::numeric_limits<InType>::max(); | ||
| 56 | InType best_vertex = -1; | ||
| 57 | |||
| 58 | #pragma omp for nowait | ||
| 59 | for (InType vertex_idx = 0; vertex_idx < vertex_count; ++vertex_idx) { | ||
| 60 | if (processed[vertex_idx] == 0 && distances[vertex_idx] < best_distance) { | ||
| 61 | best_distance = distances[vertex_idx]; | ||
| 62 | best_vertex = vertex_idx; | ||
| 63 | } | ||
| 64 | } | ||
| 65 | |||
| 66 | thread_best_distance[tid] = best_distance; | ||
| 67 | thread_best_vertex[tid] = best_vertex; | ||
| 68 | |||
| 69 | #pragma omp barrier | ||
| 70 | #pragma omp single | ||
| 71 | { | ||
| 72 | InType best_global_distance = std::numeric_limits<InType>::max(); | ||
| 73 | for (int thread_idx = 0; thread_idx < worker_count; ++thread_idx) { | ||
| 74 | if (thread_best_vertex[thread_idx] != -1 && (thread_best_distance[thread_idx] < best_global_distance || | ||
| 75 | (thread_best_distance[thread_idx] == best_global_distance && | ||
| 76 | thread_best_vertex[thread_idx] < selected_vertex))) { | ||
| 77 | best_global_distance = thread_best_distance[thread_idx]; | ||
| 78 | selected_vertex = thread_best_vertex[thread_idx]; | ||
| 79 | } | ||
| 80 | } | ||
| 81 | if (selected_vertex != -1 && best_global_distance != k_infinity) { | ||
| 82 | processed[selected_vertex] = 1; | ||
| 83 | } else { | ||
| 84 | selected_vertex = -1; | ||
| 85 | } | ||
| 86 | } | ||
| 87 | } | ||
| 88 | 48 | return selected_vertex; | |
| 89 | } | ||
| 90 | |||
| 91 | 48 | void UpdateNeighborsOmp(InType anchor_vertex, std::vector<InType> &distances, const std::vector<char> &processed, | |
| 92 | InType k_infinity) { | ||
| 93 | 48 | const auto vertex_count = static_cast<InType>(distances.size()); | |
| 94 | 48 | const InType anchor_distance = distances[anchor_vertex]; | |
| 95 | |||
| 96 | 48 | #pragma omp parallel for default(none) \ | |
| 97 | shared(anchor_vertex, distances, processed, vertex_count, k_infinity, anchor_distance) | ||
| 98 | for (InType neighbor = 0; neighbor < vertex_count; ++neighbor) { | ||
| 99 | if (processed[neighbor] == 0 && neighbor != anchor_vertex) { | ||
| 100 | const InType edge_weight = GetEdgeWeight(anchor_vertex, neighbor); | ||
| 101 | if (anchor_distance == k_infinity) { | ||
| 102 | continue; | ||
| 103 | } | ||
| 104 | InType new_distance = 0; | ||
| 105 | if (!SafeAdd(anchor_distance, edge_weight, new_distance)) { | ||
| 106 | continue; | ||
| 107 | } | ||
| 108 | distances[neighbor] = std::min(new_distance, distances[neighbor]); | ||
| 109 | } | ||
| 110 | } | ||
| 111 | 48 | } | |
| 112 | |||
| 113 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | OutType AccumulateReachableDistanceOmp(const std::vector<InType> &distances, InType k_infinity) { |
| 114 | int64_t aggregated_distance = 0; | ||
| 115 | 12 | const auto vertex_count = static_cast<InType>(distances.size()); | |
| 116 | 12 | #pragma omp parallel for reduction(+ : aggregated_distance) default(none) shared(distances, vertex_count, k_infinity) | |
| 117 | for (InType vertex_idx = 0; vertex_idx < vertex_count; ++vertex_idx) { | ||
| 118 | if (distances[vertex_idx] != k_infinity) { | ||
| 119 | aggregated_distance += distances[vertex_idx]; | ||
| 120 | } | ||
| 121 | } | ||
| 122 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | if (aggregated_distance < 0 || aggregated_distance > std::numeric_limits<OutType>::max()) { |
| 123 | ✗ | return -1; | |
| 124 | } | ||
| 125 | return static_cast<OutType>(aggregated_distance); | ||
| 126 | } | ||
| 127 | |||
| 128 | } // namespace | ||
| 129 | |||
| 130 | 12 | NalitovDDijkstrasAlgorithmOmp::NalitovDDijkstrasAlgorithmOmp(const InType &in) { | |
| 131 | SetTypeOfTask(GetStaticTypeOfTask()); | ||
| 132 | 12 | GetInput() = in; | |
| 133 | GetOutput() = 0; | ||
| 134 | 12 | } | |
| 135 | |||
| 136 | 12 | bool NalitovDDijkstrasAlgorithmOmp::ValidationImpl() { | |
| 137 | 12 | const InType n = GetInput(); | |
| 138 | |||
| 139 | constexpr InType kMaxVertices = 10000; | ||
| 140 |
1/2✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
|
12 | if (n <= 0 || n > kMaxVertices) { |
| 141 | return false; | ||
| 142 | } | ||
| 143 | |||
| 144 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | if (GetOutput() != 0) { |
| 145 | ✗ | return false; | |
| 146 | } | ||
| 147 | |||
| 148 | return true; | ||
| 149 | } | ||
| 150 | |||
| 151 | 12 | bool NalitovDDijkstrasAlgorithmOmp::PreProcessingImpl() { | |
| 152 | 12 | const InType n = GetInput(); | |
| 153 | 12 | const InType k_infinity = std::numeric_limits<InType>::max(); | |
| 154 | 12 | distances_.assign(n, k_infinity); | |
| 155 |
1/2✓ Branch 1 taken 12 times.
✗ Branch 2 not taken.
|
12 | processed_.assign(n, 0); |
| 156 |
1/2✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
|
12 | if (!distances_.empty()) { |
| 157 | 12 | distances_[0] = 0; | |
| 158 | } | ||
| 159 | 12 | GetOutput() = 0; | |
| 160 | 12 | return true; | |
| 161 | } | ||
| 162 | |||
| 163 | 12 | bool NalitovDDijkstrasAlgorithmOmp::RunImpl() { | |
| 164 | 12 | const InType n = GetInput(); | |
| 165 | |||
| 166 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | if (n <= 0) { |
| 167 | return false; | ||
| 168 | } | ||
| 169 | |||
| 170 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | if (n == 1) { |
| 171 | ✗ | GetOutput() = 0; | |
| 172 | ✗ | return true; | |
| 173 | } | ||
| 174 | |||
| 175 | if (n < 2) { | ||
| 176 | return false; | ||
| 177 | } | ||
| 178 | |||
| 179 | const InType k_infinity = std::numeric_limits<InType>::max(); | ||
| 180 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | if (distances_.empty()) { |
| 181 | return false; | ||
| 182 | } | ||
| 183 | const int worker_count = GetWorkerCount(); | ||
| 184 | |||
| 185 |
2/2✓ Branch 0 taken 48 times.
✓ Branch 1 taken 12 times.
|
60 | for (InType iteration = 0; iteration < n; ++iteration) { |
| 186 | 48 | const InType current_vertex = SelectNextVertexOmp(distances_, processed_, k_infinity, worker_count); | |
| 187 | |||
| 188 |
2/4✓ Branch 0 taken 48 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 48 times.
✗ Branch 3 not taken.
|
48 | if (current_vertex == -1 || distances_[current_vertex] == k_infinity) { |
| 189 | break; | ||
| 190 | } | ||
| 191 | |||
| 192 | 48 | UpdateNeighborsOmp(current_vertex, distances_, processed_, k_infinity); | |
| 193 | } | ||
| 194 | |||
| 195 | 12 | const OutType total_sum = AccumulateReachableDistanceOmp(distances_, k_infinity); | |
| 196 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | if (total_sum < 0) { |
| 197 | return false; | ||
| 198 | } | ||
| 199 | |||
| 200 | 12 | GetOutput() = total_sum; | |
| 201 | 12 | return true; | |
| 202 | } | ||
| 203 | |||
| 204 | 12 | bool NalitovDDijkstrasAlgorithmOmp::PostProcessingImpl() { | |
| 205 | 12 | return GetOutput() >= 0; | |
| 206 | } | ||
| 207 | |||
| 208 | } // namespace nalitov_d_dijkstras_algorithm | ||
| 209 |