| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #include "viderman_a_sparse_matrix_mult_crs_complex/omp/include/ops_omp.hpp" | ||
| 2 | |||
| 3 | #include <algorithm> | ||
| 4 | #include <cmath> | ||
| 5 | #include <complex> | ||
| 6 | #include <cstddef> | ||
| 7 | #include <vector> | ||
| 8 | |||
| 9 | #include "viderman_a_sparse_matrix_mult_crs_complex/common/include/common.hpp" | ||
| 10 | |||
| 11 | namespace viderman_a_sparse_matrix_mult_crs_complex { | ||
| 12 | namespace { | ||
| 13 | |||
| 14 |
2/2✓ Branch 0 taken 2 times.
✓ Branch 1 taken 38 times.
|
40 | void ProcessRow(const CRSMatrix &a, const CRSMatrix &b, int row, std::vector<std::complex<double>> &accumulator, |
| 15 | std::vector<int> &marker, std::vector<int> ¤t_row_indices, std::vector<int> &dst_cols, | ||
| 16 | std::vector<Complex> &dst_vals) { | ||
| 17 | current_row_indices.clear(); | ||
| 18 | |||
| 19 |
2/2✓ Branch 0 taken 28 times.
✓ Branch 1 taken 40 times.
|
68 | for (int j = a.row_ptr[row]; j < a.row_ptr[row + 1]; ++j) { |
| 20 | 28 | const int col_a = a.col_indices[j]; | |
| 21 | 28 | const std::complex<double> val_a = a.values[j]; | |
| 22 | |||
| 23 |
2/2✓ Branch 0 taken 28 times.
✓ Branch 1 taken 28 times.
|
56 | for (int k = b.row_ptr[col_a]; k < b.row_ptr[col_a + 1]; ++k) { |
| 24 |
1/2✓ Branch 0 taken 28 times.
✗ Branch 1 not taken.
|
28 | const int col_b = b.col_indices[k]; |
| 25 |
1/2✓ Branch 0 taken 28 times.
✗ Branch 1 not taken.
|
28 | accumulator[col_b] += val_a * b.values[k]; |
| 26 | |||
| 27 |
1/2✓ Branch 0 taken 28 times.
✗ Branch 1 not taken.
|
28 | if (marker[col_b] != row) { |
| 28 | current_row_indices.push_back(col_b); | ||
| 29 | 28 | marker[col_b] = row; | |
| 30 | } | ||
| 31 | } | ||
| 32 | } | ||
| 33 | |||
| 34 | std::ranges::sort(current_row_indices); | ||
| 35 | |||
| 36 | dst_cols.clear(); | ||
| 37 | dst_vals.clear(); | ||
| 38 | 40 | dst_cols.reserve(current_row_indices.size()); | |
| 39 | 40 | dst_vals.reserve(current_row_indices.size()); | |
| 40 | |||
| 41 |
2/2✓ Branch 0 taken 28 times.
✓ Branch 1 taken 40 times.
|
68 | for (const int idx : current_row_indices) { |
| 42 |
1/2✓ Branch 0 taken 28 times.
✗ Branch 1 not taken.
|
28 | if (std::abs(accumulator[idx]) > kEpsilon) { |
| 43 | dst_cols.push_back(idx); | ||
| 44 | dst_vals.push_back(accumulator[idx]); | ||
| 45 | } | ||
| 46 | 28 | accumulator[idx] = std::complex<double>(0.0, 0.0); | |
| 47 | } | ||
| 48 | 40 | } | |
| 49 | |||
| 50 | } // namespace | ||
| 51 | |||
| 52 |
1/2✓ Branch 2 taken 24 times.
✗ Branch 3 not taken.
|
24 | VidermanASparseMatrixMultCRSComplexOMP::VidermanASparseMatrixMultCRSComplexOMP(const InType &in) { |
| 53 | SetTypeOfTask(GetStaticTypeOfTask()); | ||
| 54 | GetInput() = in; | ||
| 55 | 24 | GetOutput() = CRSMatrix(); | |
| 56 | 24 | } | |
| 57 | |||
| 58 |
1/2✓ Branch 0 taken 24 times.
✗ Branch 1 not taken.
|
24 | bool VidermanASparseMatrixMultCRSComplexOMP::ValidationImpl() { |
| 59 | const auto &input = GetInput(); | ||
| 60 | const auto &a = std::get<0>(input); | ||
| 61 | const auto &b = std::get<1>(input); | ||
| 62 | |||
| 63 |
2/4✓ Branch 0 taken 24 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 24 times.
✗ Branch 3 not taken.
|
24 | if (!a.IsValid() || !b.IsValid()) { |
| 64 | return false; | ||
| 65 | } | ||
| 66 | |||
| 67 |
2/2✓ Branch 0 taken 4 times.
✓ Branch 1 taken 20 times.
|
24 | if (a.cols != b.rows) { |
| 68 | 4 | return false; | |
| 69 | } | ||
| 70 | |||
| 71 | return true; | ||
| 72 | } | ||
| 73 | |||
| 74 | 20 | bool VidermanASparseMatrixMultCRSComplexOMP::PreProcessingImpl() { | |
| 75 | const auto &input = GetInput(); | ||
| 76 | |||
| 77 | 20 | A_ = &std::get<0>(input); | |
| 78 | 20 | B_ = &std::get<1>(input); | |
| 79 | |||
| 80 | 20 | return true; | |
| 81 | } | ||
| 82 | |||
| 83 | 20 | void VidermanASparseMatrixMultCRSComplexOMP::Multiply(const CRSMatrix &a, const CRSMatrix &b, CRSMatrix &c) { | |
| 84 | 20 | c.rows = a.rows; | |
| 85 | 20 | c.cols = b.cols; | |
| 86 |
1/2✗ Branch 1 not taken.
✓ Branch 2 taken 20 times.
|
20 | c.row_ptr.assign(a.rows + 1, 0); |
| 87 | c.col_indices.clear(); | ||
| 88 | c.values.clear(); | ||
| 89 | |||
| 90 | 20 | std::vector<std::vector<int>> row_cols(a.rows); | |
| 91 |
1/2✓ Branch 1 taken 20 times.
✗ Branch 2 not taken.
|
20 | std::vector<std::vector<Complex>> row_vals(a.rows); |
| 92 | |||
| 93 | 20 | #pragma omp parallel default(none) shared(a, b, row_cols, row_vals, kEpsilon) | |
| 94 | { | ||
| 95 | std::vector<std::complex<double>> accumulator(b.cols, std::complex<double>(0.0, 0.0)); | ||
| 96 | std::vector<int> marker(b.cols, -1); | ||
| 97 | std::vector<int> current_row_indices; | ||
| 98 | |||
| 99 | #pragma omp for schedule(static) | ||
| 100 | for (int i = 0; i < a.rows; ++i) { | ||
| 101 | auto &dst_cols = row_cols[i]; | ||
| 102 | auto &dst_vals = row_vals[i]; | ||
| 103 | ProcessRow(a, b, i, accumulator, marker, current_row_indices, dst_cols, dst_vals); | ||
| 104 | } | ||
| 105 | } | ||
| 106 | |||
| 107 |
2/2✓ Branch 0 taken 40 times.
✓ Branch 1 taken 20 times.
|
60 | for (int i = 0; i < a.rows; ++i) { |
| 108 | 40 | c.row_ptr[i + 1] = c.row_ptr[i] + static_cast<int>(row_cols[i].size()); | |
| 109 | } | ||
| 110 | |||
| 111 |
1/2✓ Branch 1 taken 20 times.
✗ Branch 2 not taken.
|
20 | c.col_indices.reserve(static_cast<std::size_t>(c.row_ptr[a.rows])); |
| 112 |
1/2✓ Branch 1 taken 20 times.
✗ Branch 2 not taken.
|
20 | c.values.reserve(static_cast<std::size_t>(c.row_ptr[a.rows])); |
| 113 |
2/2✓ Branch 0 taken 40 times.
✓ Branch 1 taken 20 times.
|
60 | for (int i = 0; i < a.rows; ++i) { |
| 114 |
2/4✓ Branch 1 taken 40 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 40 times.
✗ Branch 5 not taken.
|
40 | c.col_indices.insert(c.col_indices.end(), row_cols[i].begin(), row_cols[i].end()); |
| 115 | 40 | c.values.insert(c.values.end(), row_vals[i].begin(), row_vals[i].end()); | |
| 116 | } | ||
| 117 | 20 | } | |
| 118 | |||
| 119 | 20 | bool VidermanASparseMatrixMultCRSComplexOMP::RunImpl() { | |
| 120 |
2/4✓ Branch 0 taken 20 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 20 times.
✗ Branch 3 not taken.
|
20 | if (A_ == nullptr || B_ == nullptr) { |
| 121 | return false; | ||
| 122 | } | ||
| 123 | |||
| 124 | CRSMatrix &c = GetOutput(); | ||
| 125 | 20 | Multiply(*A_, *B_, c); | |
| 126 | |||
| 127 | 20 | return true; | |
| 128 | } | ||
| 129 | |||
| 130 | 20 | bool VidermanASparseMatrixMultCRSComplexOMP::PostProcessingImpl() { | |
| 131 | CRSMatrix &c = GetOutput(); | ||
| 132 | 20 | return c.IsValid(); | |
| 133 | } | ||
| 134 | |||
| 135 | } // namespace viderman_a_sparse_matrix_mult_crs_complex | ||
| 136 |