| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #include "krymova_k_lsd_sort_merge_double/all/include/ops_all.hpp" | ||
| 2 | |||
| 3 | #include <mpi.h> | ||
| 4 | #include <omp.h> | ||
| 5 | |||
| 6 | #include <algorithm> | ||
| 7 | #include <cstdint> | ||
| 8 | #include <cstring> | ||
| 9 | #include <utility> | ||
| 10 | #include <vector> | ||
| 11 | |||
| 12 | #include "krymova_k_lsd_sort_merge_double/common/include/common.hpp" | ||
| 13 | |||
| 14 | namespace krymova_k_lsd_sort_merge_double { | ||
| 15 | |||
| 16 |
1/2✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
|
24 | KrymovaKLsdSortMergeDoubleALL::KrymovaKLsdSortMergeDoubleALL(const InType &in) { |
| 17 | SetTypeOfTask(GetStaticTypeOfTask()); | ||
| 18 |
1/2✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
|
24 | GetInput() = in; |
| 19 | 24 | GetOutput() = OutType(); | |
| 20 | 24 | } | |
| 21 | |||
| 22 | 24 | bool KrymovaKLsdSortMergeDoubleALL::ValidationImpl() { | |
| 23 | 24 | return true; | |
| 24 | } | ||
| 25 | |||
| 26 | 24 | bool KrymovaKLsdSortMergeDoubleALL::PreProcessingImpl() { | |
| 27 | 24 | return true; | |
| 28 | } | ||
| 29 | |||
| 30 | ✗ | uint64_t KrymovaKLsdSortMergeDoubleALL::DoubleToULL(double d) { | |
| 31 | uint64_t ull = 0U; | ||
| 32 | std::memcpy(&ull, &d, sizeof(double)); | ||
| 33 | ✗ | return ((ull & 0x8000000000000000ULL) != 0U) ? ~ull : (ull | 0x8000000000000000ULL); | |
| 34 | } | ||
| 35 | |||
| 36 | ✗ | double KrymovaKLsdSortMergeDoubleALL::ULLToDouble(uint64_t ull) { | |
| 37 | ✗ | if ((ull & 0x8000000000000000ULL) != 0U) { | |
| 38 | ✗ | ull &= 0x7FFFFFFFFFFFFFFFULL; | |
| 39 | } else { | ||
| 40 | ✗ | ull = ~ull; | |
| 41 | } | ||
| 42 | double d = 0.0; | ||
| 43 | std::memcpy(&d, &ull, sizeof(double)); | ||
| 44 | ✗ | return d; | |
| 45 | } | ||
| 46 | |||
| 47 | 23 | void KrymovaKLsdSortMergeDoubleALL::LSDSort(double *arr, int size) { | |
| 48 |
2/2✓ Branch 0 taken 1 times.
✓ Branch 1 taken 22 times.
|
23 | if (size <= 1) { |
| 49 | 1 | return; | |
| 50 | } | ||
| 51 | |||
| 52 | const int k_bits_per_pass = 8; | ||
| 53 | const int k_radix = 1 << k_bits_per_pass; | ||
| 54 | const int k_passes = 8; | ||
| 55 | |||
| 56 | 22 | std::vector<uint64_t> ull_arr(size); | |
| 57 |
1/4✓ Branch 1 taken 22 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
22 | std::vector<uint64_t> ull_tmp(size); |
| 58 |
1/4✓ Branch 1 taken 22 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
22 | std::vector<unsigned int> count(k_radix, 0U); |
| 59 | |||
| 60 | 22 | #pragma omp parallel for default(none) shared(ull_arr, arr, size) | |
| 61 | for (int i = 0; i < size; ++i) { | ||
| 62 | ull_arr[i] = DoubleToULL(arr[i]); | ||
| 63 | } | ||
| 64 | |||
| 65 |
2/2✓ Branch 0 taken 176 times.
✓ Branch 1 taken 22 times.
|
198 | for (int pass = 0; pass < k_passes; ++pass) { |
| 66 |
1/2✓ Branch 0 taken 176 times.
✗ Branch 1 not taken.
|
176 | int shift = pass * k_bits_per_pass; |
| 67 | std::ranges::fill(count, 0U); | ||
| 68 | |||
| 69 |
2/2✓ Branch 0 taken 900160 times.
✓ Branch 1 taken 176 times.
|
900336 | for (int i = 0; i < size; ++i) { |
| 70 | 900160 | unsigned int digit = (ull_arr[i] >> shift) & (k_radix - 1); | |
| 71 | 900160 | ++count[digit]; | |
| 72 | } | ||
| 73 | |||
| 74 |
2/2✓ Branch 0 taken 44880 times.
✓ Branch 1 taken 176 times.
|
45056 | for (int i = 1; i < k_radix; ++i) { |
| 75 | 44880 | count[i] += count[i - 1]; | |
| 76 | } | ||
| 77 | |||
| 78 |
2/2✓ Branch 0 taken 900160 times.
✓ Branch 1 taken 176 times.
|
900336 | for (int i = size - 1; i >= 0; --i) { |
| 79 | 900160 | unsigned int digit = (ull_arr[i] >> shift) & (k_radix - 1); | |
| 80 | 900160 | ull_tmp[--count[digit]] = ull_arr[i]; | |
| 81 | } | ||
| 82 | |||
| 83 | ull_arr.swap(ull_tmp); | ||
| 84 | } | ||
| 85 | |||
| 86 |
1/2✓ Branch 0 taken 22 times.
✗ Branch 1 not taken.
|
22 | #pragma omp parallel for default(none) shared(arr, ull_arr, size) |
| 87 | for (int i = 0; i < size; ++i) { | ||
| 88 | arr[i] = ULLToDouble(ull_arr[i]); | ||
| 89 | } | ||
| 90 | } | ||
| 91 | |||
| 92 | 11 | std::vector<double> KrymovaKLsdSortMergeDoubleALL::SimpleMerge(const std::vector<double> &a, | |
| 93 | const std::vector<double> &b) { | ||
| 94 |
1/2✓ Branch 1 taken 11 times.
✗ Branch 2 not taken.
|
11 | std::vector<double> res; |
| 95 |
1/2✓ Branch 1 taken 11 times.
✗ Branch 2 not taken.
|
11 | res.reserve(a.size() + b.size()); |
| 96 | size_t i = 0; | ||
| 97 | size_t j = 0; | ||
| 98 |
4/4✓ Branch 0 taken 10 times.
✓ Branch 1 taken 56261 times.
✓ Branch 2 taken 1 times.
✓ Branch 3 taken 56260 times.
|
56271 | while (i < a.size() && j < b.size()) { |
| 99 |
2/2✓ Branch 0 taken 56210 times.
✓ Branch 1 taken 50 times.
|
56260 | res.push_back(a[i] <= b[j] ? a[i++] : b[j++]); |
| 100 | } | ||
| 101 |
2/2✓ Branch 0 taken 50 times.
✓ Branch 1 taken 11 times.
|
61 | while (i < a.size()) { |
| 102 |
1/2✓ Branch 0 taken 50 times.
✗ Branch 1 not taken.
|
50 | res.push_back(a[i++]); |
| 103 | } | ||
| 104 |
2/2✓ Branch 0 taken 56210 times.
✓ Branch 1 taken 11 times.
|
56221 | while (j < b.size()) { |
| 105 |
1/2✓ Branch 0 taken 56210 times.
✗ Branch 1 not taken.
|
56210 | res.push_back(b[j++]); |
| 106 | } | ||
| 107 | 11 | return res; | |
| 108 | } | ||
| 109 | |||
| 110 | ✗ | bool KrymovaKLsdSortMergeDoubleALL::RunSmallDataset(int total_size) { | |
| 111 | ✗ | if (total_size > 0) { | |
| 112 | ✗ | GetOutput() = GetInput(); | |
| 113 | ✗ | LSDSort(GetOutput().data(), total_size); | |
| 114 | } else { | ||
| 115 | GetOutput().clear(); | ||
| 116 | } | ||
| 117 | ✗ | return true; | |
| 118 | } | ||
| 119 | |||
| 120 | 24 | void KrymovaKLsdSortMergeDoubleALL::ComputeDistribution(int total_size, int size_comm, std::vector<int> &send_counts, | |
| 121 | std::vector<int> &offsets) { | ||
| 122 | 24 | int chunk = total_size / size_comm; | |
| 123 | 24 | int rem = total_size % size_comm; | |
| 124 | |||
| 125 |
2/2✓ Branch 0 taken 48 times.
✓ Branch 1 taken 24 times.
|
72 | for (int i = 0; i < size_comm; ++i) { |
| 126 |
4/4✓ Branch 0 taken 46 times.
✓ Branch 1 taken 2 times.
✓ Branch 2 taken 24 times.
✓ Branch 3 taken 24 times.
|
94 | send_counts[i] = chunk + (i < rem ? 1 : 0); |
| 127 |
2/2✓ Branch 0 taken 24 times.
✓ Branch 1 taken 24 times.
|
48 | offsets[i] = (i == 0) ? 0 : offsets[i - 1] + send_counts[i - 1]; |
| 128 | } | ||
| 129 | 24 | } | |
| 130 | |||
| 131 | 24 | void KrymovaKLsdSortMergeDoubleALL::ScatterData(int rank, const std::vector<int> &send_counts, | |
| 132 | const std::vector<int> &offsets, std::vector<double> &local_data) { | ||
| 133 |
2/2✓ Branch 0 taken 12 times.
✓ Branch 1 taken 12 times.
|
24 | if (rank == 0) { |
| 134 | 12 | MPI_Scatterv(GetInput().data(), send_counts.data(), offsets.data(), MPI_DOUBLE, local_data.data(), | |
| 135 | send_counts[rank], MPI_DOUBLE, 0, MPI_COMM_WORLD); | ||
| 136 |
2/2✓ Branch 0 taken 11 times.
✓ Branch 1 taken 1 times.
|
12 | } else if (send_counts[rank] > 0) { |
| 137 | 11 | MPI_Scatterv(nullptr, send_counts.data(), offsets.data(), MPI_DOUBLE, local_data.data(), send_counts[rank], | |
| 138 | MPI_DOUBLE, 0, MPI_COMM_WORLD); | ||
| 139 | } | ||
| 140 | 24 | } | |
| 141 | |||
| 142 | 24 | void KrymovaKLsdSortMergeDoubleALL::GatherResults(int rank, int size_comm, const std::vector<int> &send_counts, | |
| 143 | std::vector<double> &local_data) { | ||
| 144 |
2/2✓ Branch 0 taken 12 times.
✓ Branch 1 taken 12 times.
|
24 | if (rank == 0) { |
| 145 | 12 | std::vector<double> result = local_data; | |
| 146 |
2/2✓ Branch 0 taken 12 times.
✓ Branch 1 taken 12 times.
|
24 | for (int i = 1; i < size_comm; ++i) { |
| 147 |
2/2✓ Branch 0 taken 11 times.
✓ Branch 1 taken 1 times.
|
12 | if (send_counts[i] > 0) { |
| 148 |
2/6✓ Branch 1 taken 11 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 11 times.
✗ Branch 5 not taken.
✗ Branch 6 not taken.
✗ Branch 7 not taken.
|
11 | std::vector<double> recv_buf(send_counts[i]); |
| 149 |
1/2✓ Branch 1 taken 11 times.
✗ Branch 2 not taken.
|
11 | MPI_Recv(recv_buf.data(), send_counts[i], MPI_DOUBLE, i, 0, MPI_COMM_WORLD, MPI_STATUS_IGNORE); |
| 150 |
2/6✓ Branch 1 taken 11 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 11 times.
✗ Branch 4 not taken.
✗ Branch 5 not taken.
✗ Branch 6 not taken.
|
22 | result = SimpleMerge(result, recv_buf); |
| 151 | } | ||
| 152 | } | ||
| 153 | GetOutput() = std::move(result); | ||
| 154 |
2/2✓ Branch 0 taken 11 times.
✓ Branch 1 taken 1 times.
|
12 | } else if (send_counts[rank] > 0) { |
| 155 | 11 | MPI_Send(local_data.data(), send_counts[rank], MPI_DOUBLE, 0, 0, MPI_COMM_WORLD); | |
| 156 | } | ||
| 157 | 24 | } | |
| 158 | |||
| 159 | 24 | void KrymovaKLsdSortMergeDoubleALL::BroadcastResult(int rank) { | |
| 160 | 24 | int out_size = static_cast<int>(GetOutput().size()); | |
| 161 | 24 | MPI_Bcast(&out_size, 1, MPI_INT, 0, MPI_COMM_WORLD); | |
| 162 | |||
| 163 |
3/4✓ Branch 0 taken 12 times.
✓ Branch 1 taken 12 times.
✓ Branch 2 taken 12 times.
✗ Branch 3 not taken.
|
24 | if ((rank != 0) && (out_size > 0)) { |
| 164 | 12 | GetOutput().resize(out_size); | |
| 165 | 12 | MPI_Bcast(GetOutput().data(), out_size, MPI_DOUBLE, 0, MPI_COMM_WORLD); | |
| 166 |
1/2✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
|
12 | } else if ((rank == 0) && (out_size > 0)) { |
| 167 | 12 | MPI_Bcast(GetOutput().data(), out_size, MPI_DOUBLE, 0, MPI_COMM_WORLD); | |
| 168 | } | ||
| 169 | 24 | } | |
| 170 | |||
| 171 | 24 | bool KrymovaKLsdSortMergeDoubleALL::RunImpl() { | |
| 172 | 24 | int rank = 0; | |
| 173 | 24 | int size_comm = 0; | |
| 174 | 24 | MPI_Comm_rank(MPI_COMM_WORLD, &rank); | |
| 175 | 24 | MPI_Comm_size(MPI_COMM_WORLD, &size_comm); | |
| 176 | |||
| 177 | 24 | int total_size = static_cast<int>(GetInput().size()); | |
| 178 | |||
| 179 | 24 | MPI_Bcast(&total_size, 1, MPI_INT, 0, MPI_COMM_WORLD); | |
| 180 | |||
| 181 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 24 times.
|
24 | if (total_size == 0) { |
| 182 | GetOutput().clear(); | ||
| 183 | ✗ | return true; | |
| 184 | } | ||
| 185 | |||
| 186 | 24 | std::vector<int> send_counts(size_comm); | |
| 187 |
1/4✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
24 | std::vector<int> offsets(size_comm); |
| 188 | 24 | ComputeDistribution(total_size, size_comm, send_counts, offsets); | |
| 189 | |||
| 190 |
1/4✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
24 | std::vector<double> local_data(send_counts[rank]); |
| 191 | |||
| 192 |
1/2✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
|
24 | ScatterData(rank, send_counts, offsets, local_data); |
| 193 | |||
| 194 |
2/2✓ Branch 0 taken 23 times.
✓ Branch 1 taken 1 times.
|
24 | if (send_counts[rank] > 0) { |
| 195 |
1/2✓ Branch 1 taken 23 times.
✗ Branch 2 not taken.
|
23 | LSDSort(local_data.data(), send_counts[rank]); |
| 196 | } | ||
| 197 | |||
| 198 |
1/2✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
|
24 | GatherResults(rank, size_comm, send_counts, local_data); |
| 199 | |||
| 200 |
1/2✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
|
24 | BroadcastResult(rank); |
| 201 | |||
| 202 | return true; | ||
| 203 | } | ||
| 204 | |||
| 205 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 24 times.
|
24 | bool KrymovaKLsdSortMergeDoubleALL::PostProcessingImpl() { |
| 206 | const OutType &output = GetOutput(); | ||
| 207 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 24 times.
|
24 | if (output.empty()) { |
| 208 | return true; | ||
| 209 | } | ||
| 210 |
2/2✓ Branch 0 taken 24 times.
✓ Branch 1 taken 225018 times.
|
225042 | for (size_t i = 1; i < output.size(); ++i) { |
| 211 |
1/2✓ Branch 0 taken 225018 times.
✗ Branch 1 not taken.
|
225018 | if (output[i] < output[i - 1]) { |
| 212 | return false; | ||
| 213 | } | ||
| 214 | } | ||
| 215 | return true; | ||
| 216 | } | ||
| 217 | |||
| 218 | } // namespace krymova_k_lsd_sort_merge_double | ||
| 219 |