| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #include "krymova_k_lsd_sort_merge_double/stl/include/ops_stl.hpp" | ||
| 2 | |||
| 3 | #include <algorithm> | ||
| 4 | #include <cstdint> | ||
| 5 | #include <cstring> | ||
| 6 | #include <thread> | ||
| 7 | #include <vector> | ||
| 8 | |||
| 9 | #include "krymova_k_lsd_sort_merge_double/common/include/common.hpp" | ||
| 10 | |||
| 11 | namespace krymova_k_lsd_sort_merge_double { | ||
| 12 | |||
| 13 |
1/2✓ Branch 1 taken 96 times.
✗ Branch 2 not taken.
|
96 | KrymovaKLsdSortMergeDoubleSTL::KrymovaKLsdSortMergeDoubleSTL(const InType &in) { |
| 14 | SetTypeOfTask(GetStaticTypeOfTask()); | ||
| 15 |
1/2✓ Branch 1 taken 96 times.
✗ Branch 2 not taken.
|
96 | GetInput() = in; |
| 16 | 96 | GetOutput() = OutType(); | |
| 17 | 96 | } | |
| 18 | |||
| 19 | 96 | bool KrymovaKLsdSortMergeDoubleSTL::ValidationImpl() { | |
| 20 | 96 | return !GetInput().empty(); | |
| 21 | } | ||
| 22 | |||
| 23 | 96 | bool KrymovaKLsdSortMergeDoubleSTL::PreProcessingImpl() { | |
| 24 | 96 | GetOutput() = GetInput(); | |
| 25 | 96 | return true; | |
| 26 | } | ||
| 27 | |||
| 28 | ✗ | uint64_t KrymovaKLsdSortMergeDoubleSTL::DoubleToULL(double d) { | |
| 29 | uint64_t ull = 0; | ||
| 30 | std::memcpy(&ull, &d, sizeof(double)); | ||
| 31 |
1/4✗ Branch 0 not taken.
✓ Branch 1 taken 900160 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
|
900160 | if ((ull & 0x8000000000000000ULL) != 0U) { |
| 32 | ✗ | ull = ~ull; | |
| 33 | } else { | ||
| 34 | 900160 | ull |= 0x8000000000000000ULL; | |
| 35 | } | ||
| 36 | ✗ | return ull; | |
| 37 | } | ||
| 38 | |||
| 39 | ✗ | double KrymovaKLsdSortMergeDoubleSTL::ULLToDouble(uint64_t ull) { | |
| 40 |
1/4✓ Branch 0 taken 900160 times.
✗ Branch 1 not taken.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
|
900160 | if ((ull & 0x8000000000000000ULL) != 0U) { |
| 41 | 900160 | ull &= 0x7FFFFFFFFFFFFFFFULL; | |
| 42 | } else { | ||
| 43 | ✗ | ull = ~ull; | |
| 44 | } | ||
| 45 | double d = 0.0; | ||
| 46 | std::memcpy(&d, &ull, sizeof(double)); | ||
| 47 | ✗ | return d; | |
| 48 | } | ||
| 49 | |||
| 50 | 368 | void KrymovaKLsdSortMergeDoubleSTL::LSDSortDoubleSequential(double *arr, int size) { | |
| 51 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 368 times.
|
368 | if (size <= 1) { |
| 52 | ✗ | return; | |
| 53 | } | ||
| 54 | const int k_bits_per_pass = 8; | ||
| 55 | const int k_radix = 1 << k_bits_per_pass; | ||
| 56 | const int k_passes = static_cast<int>(sizeof(double)) * 8 / k_bits_per_pass; | ||
| 57 | 368 | std::vector<uint64_t> ull_arr(size); | |
| 58 |
1/4✓ Branch 1 taken 368 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
368 | std::vector<uint64_t> ull_tmp(size); |
| 59 |
2/2✓ Branch 0 taken 900160 times.
✓ Branch 1 taken 368 times.
|
900528 | for (int i = 0; i < size; ++i) { |
| 60 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 900160 times.
|
1800320 | ull_arr[i] = DoubleToULL(arr[i]); |
| 61 | } | ||
| 62 |
1/4✓ Branch 1 taken 368 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
368 | std::vector<unsigned int> count(k_radix, 0U); |
| 63 |
2/2✓ Branch 0 taken 2944 times.
✓ Branch 1 taken 368 times.
|
3312 | for (int pass = 0; pass < k_passes; ++pass) { |
| 64 | 2944 | int shift = pass * k_bits_per_pass; | |
| 65 |
2/2✓ Branch 0 taken 753664 times.
✓ Branch 1 taken 2944 times.
|
756608 | for (auto &val : count) { |
| 66 | 753664 | val = 0U; | |
| 67 | } | ||
| 68 |
2/2✓ Branch 0 taken 7201280 times.
✓ Branch 1 taken 2944 times.
|
7204224 | for (int i = 0; i < size; ++i) { |
| 69 | 7201280 | unsigned int digit = (ull_arr[i] >> shift) & (k_radix - 1); | |
| 70 | 7201280 | ++count[digit]; | |
| 71 | } | ||
| 72 |
2/2✓ Branch 0 taken 750720 times.
✓ Branch 1 taken 2944 times.
|
753664 | for (int i = 1; i < k_radix; ++i) { |
| 73 | 750720 | count[i] += count[i - 1]; | |
| 74 | } | ||
| 75 |
2/2✓ Branch 0 taken 7201280 times.
✓ Branch 1 taken 2944 times.
|
7204224 | for (int i = size - 1; i >= 0; --i) { |
| 76 | 7201280 | unsigned int digit = (ull_arr[i] >> shift) & (k_radix - 1); | |
| 77 | 7201280 | ull_tmp[--count[digit]] = ull_arr[i]; | |
| 78 | } | ||
| 79 | ull_arr.swap(ull_tmp); | ||
| 80 | } | ||
| 81 |
2/2✓ Branch 0 taken 900160 times.
✓ Branch 1 taken 368 times.
|
900528 | for (int i = 0; i < size; ++i) { |
| 82 |
1/2✓ Branch 0 taken 900160 times.
✗ Branch 1 not taken.
|
1800320 | arr[i] = ULLToDouble(ull_arr[i]); |
| 83 | } | ||
| 84 | } | ||
| 85 | |||
| 86 | 280 | void KrymovaKLsdSortMergeDoubleSTL::MergeSections(double *left, const double *right, int left_size, int right_size) { | |
| 87 | 280 | std::vector<double> temp(left_size); | |
| 88 | 280 | std::copy(left, left + left_size, temp.begin()); | |
| 89 | int left_index = 0; | ||
| 90 | int right_index = 0; | ||
| 91 | int dest_index = 0; | ||
| 92 |
2/2✓ Branch 0 taken 900256 times.
✓ Branch 1 taken 280 times.
|
900536 | while (left_index < left_size && right_index < right_size) { |
| 93 |
2/2✓ Branch 0 taken 899456 times.
✓ Branch 1 taken 800 times.
|
900256 | if (temp[left_index] <= right[right_index]) { |
| 94 | 899456 | left[dest_index++] = temp[left_index++]; | |
| 95 | } else { | ||
| 96 | 800 | left[dest_index++] = right[right_index++]; | |
| 97 | } | ||
| 98 | } | ||
| 99 |
2/2✓ Branch 0 taken 800 times.
✓ Branch 1 taken 280 times.
|
1080 | while (left_index < left_size) { |
| 100 | 800 | left[dest_index++] = temp[left_index++]; | |
| 101 | } | ||
| 102 | 280 | } | |
| 103 | |||
| 104 | 88 | void KrymovaKLsdSortMergeDoubleSTL::SortSectionsParallel(double *arr, int size, int portion, int num_threads) { | |
| 105 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 88 times.
|
88 | int num_blocks = (size + portion - 1) / portion; |
| 106 | int num_threads_used = (std::min)(num_threads, num_blocks); | ||
| 107 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 88 times.
|
88 | if (num_threads_used <= 1) { |
| 108 | ✗ | for (int start = 0; start < size; start += portion) { | |
| 109 | ✗ | int current_size = (std::min)(portion, size - start); | |
| 110 | ✗ | LSDSortDoubleSequential(arr + start, current_size); | |
| 111 | } | ||
| 112 | ✗ | return; | |
| 113 | } | ||
| 114 | 88 | std::vector<std::thread> threads; | |
| 115 | 88 | int blocks_per_thread = (num_blocks + num_threads_used - 1) / num_threads_used; | |
| 116 |
2/2✓ Branch 0 taken 352 times.
✓ Branch 1 taken 88 times.
|
440 | for (int thread_id = 0; thread_id < num_threads_used; ++thread_id) { |
| 117 | 352 | int start_block = thread_id * blocks_per_thread; | |
| 118 |
1/2✓ Branch 1 taken 352 times.
✗ Branch 2 not taken.
|
352 | int end_block = (std::min)(start_block + blocks_per_thread, num_blocks); |
| 119 |
1/2✓ Branch 1 taken 352 times.
✗ Branch 2 not taken.
|
352 | threads.emplace_back([&, start_block, end_block, portion, size, arr]() { |
| 120 |
2/2✓ Branch 0 taken 368 times.
✓ Branch 1 taken 352 times.
|
720 | for (int block = start_block; block < end_block; ++block) { |
| 121 | 368 | int start_pos = block * portion; | |
| 122 | 368 | int current_size = (std::min)(portion, size - start_pos); | |
| 123 | 368 | LSDSortDoubleSequential(arr + start_pos, current_size); | |
| 124 | } | ||
| 125 | 352 | }); | |
| 126 | } | ||
| 127 |
2/2✓ Branch 0 taken 352 times.
✓ Branch 1 taken 88 times.
|
440 | for (auto &thr : threads) { |
| 128 |
1/2✓ Branch 1 taken 352 times.
✗ Branch 2 not taken.
|
352 | thr.join(); |
| 129 | } | ||
| 130 | 88 | } | |
| 131 | |||
| 132 | 88 | void KrymovaKLsdSortMergeDoubleSTL::IterativeMergeSort(double *arr, int size, int portion, int num_threads) { | |
| 133 |
1/2✓ Branch 0 taken 88 times.
✗ Branch 1 not taken.
|
88 | if (size <= 1) { |
| 134 | return; | ||
| 135 | } | ||
| 136 | 88 | SortSectionsParallel(arr, size, portion, num_threads); | |
| 137 |
2/2✓ Branch 0 taken 192 times.
✓ Branch 1 taken 88 times.
|
280 | for (int merge_size = portion; merge_size < size; merge_size *= 2) { |
| 138 |
2/2✓ Branch 0 taken 312 times.
✓ Branch 1 taken 192 times.
|
504 | for (int start_pos = 0; start_pos < size; start_pos += 2 * merge_size) { |
| 139 | int left_size = merge_size; | ||
| 140 |
2/2✓ Branch 0 taken 32 times.
✓ Branch 1 taken 280 times.
|
312 | int right_size = (std::min)(merge_size, size - (start_pos + merge_size)); |
| 141 |
2/2✓ Branch 0 taken 32 times.
✓ Branch 1 taken 280 times.
|
312 | if (right_size <= 0) { |
| 142 | 32 | continue; | |
| 143 | } | ||
| 144 | 280 | double *left = arr + start_pos; | |
| 145 | 280 | const double *right = arr + start_pos + left_size; | |
| 146 | 280 | MergeSections(left, right, left_size, right_size); | |
| 147 | } | ||
| 148 | } | ||
| 149 | } | ||
| 150 | |||
| 151 |
2/2✓ Branch 0 taken 88 times.
✓ Branch 1 taken 8 times.
|
96 | bool KrymovaKLsdSortMergeDoubleSTL::RunImpl() { |
| 152 | OutType &output = GetOutput(); | ||
| 153 | 96 | int size = static_cast<int>(output.size()); | |
| 154 |
2/2✓ Branch 0 taken 88 times.
✓ Branch 1 taken 8 times.
|
96 | if (size <= 1) { |
| 155 | return true; | ||
| 156 | } | ||
| 157 | 88 | int num_threads = (std::max)(1, static_cast<int>(std::thread::hardware_concurrency())); | |
| 158 | 88 | num_threads = (std::min)(num_threads, size); | |
| 159 | 88 | int portion = (std::max)(1, size / num_threads); | |
| 160 | 88 | IterativeMergeSort(output.data(), size, portion, num_threads); | |
| 161 | return true; | ||
| 162 | } | ||
| 163 | |||
| 164 | 96 | bool KrymovaKLsdSortMergeDoubleSTL::PostProcessingImpl() { | |
| 165 | const OutType &output = GetOutput(); | ||
| 166 |
2/2✓ Branch 0 taken 900072 times.
✓ Branch 1 taken 96 times.
|
900168 | for (size_t i = 1; i < output.size(); ++i) { |
| 167 |
1/2✓ Branch 0 taken 900072 times.
✗ Branch 1 not taken.
|
900072 | if (output[i] < output[i - 1]) { |
| 168 | return false; | ||
| 169 | } | ||
| 170 | } | ||
| 171 | return true; | ||
| 172 | } | ||
| 173 | |||
| 174 | } // namespace krymova_k_lsd_sort_merge_double | ||
| 175 |