GCC Code Coverage Report

Directory: ./
File: tasks/dilshodov_a_spmm_double_css/all/src/ops_all.cpp
Date: 2026-06-04 20:25:32
Exec Total Coverage
Lines: 102 104 98.1%
Functions: 10 10 100.0%
Branches: 76 134 56.7%
Line Branch Exec Source
1 #include "dilshodov_a_spmm_double_css/all/include/ops_all.hpp"
2
3 #include <mpi.h>
4 #include <omp.h>
5
6 #include <algorithm>
7 #include <cmath>
8 #include <cstddef>
9 #include <cstdint>
10 #include <utility>
11 #include <vector>
12
13 #include "dilshodov_a_spmm_double_css/common/include/common.hpp"
14
15 namespace dilshodov_a_spmm_double_css {
16
17 namespace {
18 constexpr double kEps = 1e-10;
19
20 bool HasValidDimensions(const SparseMatrixCCS &m) {
21
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 return m.rows_count > 0 && m.cols_count > 0;
22 }
23
24
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 bool HasValidContainers(const SparseMatrixCCS &m) {
25
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 if (m.col_ptrs.size() != static_cast<size_t>(m.cols_count) + 1) {
26 return false;
27 }
28
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 if (m.row_indices.size() != m.values.size()) {
29 return false;
30 }
31
2/4
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 12 times.
✗ Branch 3 not taken.
 12 if (m.col_ptrs.empty() || m.col_ptrs.front() != 0) {
32 return false;
33 }
34
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 if (m.col_ptrs.back() < 0) {
35 return false;
36 }
37
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
 12 if (static_cast<size_t>(m.col_ptrs.back()) != m.values.size()) {
38 ✗ return false;
39 }
40 return true;
41 }
42
43 12 bool HasValidColumnOrdering(const SparseMatrixCCS &m) {
44
2/2
✓ Branch 0 taken 30 times.
✓ Branch 1 taken 12 times.
 42 for (int j = 0; j < m.cols_count; ++j) {
45
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 30 times.
 30 if (m.col_ptrs[j] > m.col_ptrs[j + 1]) {
46 return false;
47 }
48 int prev_row = -1;
49
2/2
✓ Branch 0 taken 46 times.
✓ Branch 1 taken 30 times.
 76 for (int idx = m.col_ptrs[j]; idx < m.col_ptrs[j + 1]; ++idx) {
50
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 46 times.
 46 const int row = m.row_indices[idx];
51
2/4
✗ Branch 0 not taken.
✓ Branch 1 taken 46 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 46 times.
 46 if (row < 0 || row >= m.rows_count) {
52 return false;
53 }
54
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 46 times.
 46 if (row <= prev_row) {
55 return false;
56 }
57 prev_row = row;
58 }
59 }
60 return true;
61 }
62
63
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 bool IsValidCCS(const SparseMatrixCCS &m) {
64
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 return HasValidDimensions(m) && HasValidContainers(m) && HasValidColumnOrdering(m);
65 }
66
67
2/2
✓ Branch 0 taken 1 times.
✓ Branch 1 taken 6 times.
 7 void AccumulateColumnProduct(const SparseMatrixCCS &lhs, const SparseMatrixCCS &rhs, int rhs_col,
68 std::vector<double> &acc, std::vector<int> &marker, std::vector<int> &used_rows,
69 std::vector<std::pair<int, double>> &col) {
70 used_rows.clear();
71 col.clear();
72
73
2/2
✓ Branch 0 taken 12 times.
✓ Branch 1 taken 7 times.
 19 for (int idx_b = rhs.col_ptrs[rhs_col]; idx_b < rhs.col_ptrs[rhs_col + 1]; ++idx_b) {
74 12 const int k = rhs.row_indices[idx_b];
75 12 const double rhs_value = rhs.values[idx_b];
76
77
2/2
✓ Branch 0 taken 16 times.
✓ Branch 1 taken 12 times.
 28 for (int idx_a = lhs.col_ptrs[k]; idx_a < lhs.col_ptrs[k + 1]; ++idx_a) {
78
2/2
✓ Branch 0 taken 12 times.
✓ Branch 1 taken 4 times.
 16 const int row = lhs.row_indices[idx_a];
79 16 const double value = lhs.values[idx_a] * rhs_value;
80
81
2/2
✓ Branch 0 taken 12 times.
✓ Branch 1 taken 4 times.
 16 if (marker[row] != rhs_col) {
82
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 marker[row] = rhs_col;
83
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 acc[row] = value;
84 used_rows.push_back(row);
85 } else {
86 4 acc[row] += value;
87 }
88 }
89 }
90
91 7 col.reserve(used_rows.size());
92
2/2
✓ Branch 0 taken 12 times.
✓ Branch 1 taken 7 times.
 19 for (int row : used_rows) {
93
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 const double value = acc[row];
94
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 if (std::abs(value) > kEps) {
95 12 col.emplace_back(row, value);
96 }
97 }
98
99 std::ranges::sort(col, {}, &std::pair<int, double>::first);
100 7 }
101
102 6 void ComputeLocalSlab(const SparseMatrixCCS &lhs, const SparseMatrixCCS &rhs, int j_start, int j_end,
103 std::vector<int> &local_col_ptrs, std::vector<int> &local_row_indices,
104 std::vector<double> &local_values) {
105 6 const int local_cols = j_end - j_start;
106 6 std::vector<std::vector<std::pair<int, double>>> column_results(local_cols);
107
108 6 #pragma omp parallel default(none) shared(lhs, rhs, column_results, j_start, j_end)
109 {
110 std::vector<double> acc(lhs.rows_count, 0.0);
111 std::vector<int> marker(lhs.rows_count, -1);
112 std::vector<int> used_rows;
113 used_rows.reserve(256);
114
115 #pragma omp for schedule(dynamic)
116 for (int j = j_start; j < j_end; ++j) {
117 AccumulateColumnProduct(lhs, rhs, j, acc, marker, used_rows, column_results[j - j_start]);
118 }
119 }
120
121
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 local_col_ptrs.assign(static_cast<size_t>(local_cols) + 1, 0);
122
2/2
✓ Branch 0 taken 7 times.
✓ Branch 1 taken 6 times.
 13 for (int col_idx = 0; col_idx < local_cols; ++col_idx) {
123 7 local_col_ptrs[col_idx + 1] = local_col_ptrs[col_idx] + static_cast<int>(column_results[col_idx].size());
124 }
125 6 const int local_nnz = local_col_ptrs[local_cols];
126
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 local_row_indices.resize(local_nnz);
127
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 local_values.resize(local_nnz);
128
129
2/2
✓ Branch 0 taken 7 times.
✓ Branch 1 taken 6 times.
 13 for (int col_idx = 0; col_idx < local_cols; ++col_idx) {
130 7 int write = local_col_ptrs[col_idx];
131
2/2
✓ Branch 0 taken 12 times.
✓ Branch 1 taken 7 times.
 19 for (const auto &[row, value] : column_results[col_idx]) {
132 12 local_row_indices[write] = row;
133 12 local_values[write] = value;
134 12 ++write;
135 }
136 }
137 6 }
138
139 } // namespace
140
141
1/2
✓ Branch 2 taken 6 times.
✗ Branch 3 not taken.
 6 DilshodovASpmmDoubleCssAll::DilshodovASpmmDoubleCssAll(const InType &in) {
142 SetTypeOfTask(GetStaticTypeOfTask());
143 GetInput() = in;
144 6 }
145
146
1/2
✓ Branch 0 taken 6 times.
✗ Branch 1 not taken.
 6 bool DilshodovASpmmDoubleCssAll::ValidationImpl() {
147 const auto &lhs = std::get<0>(GetInput());
148 const auto &rhs = std::get<1>(GetInput());
149
3/6
✓ Branch 0 taken 6 times.
✗ Branch 1 not taken.
✗ Branch 2 not taken.
✓ Branch 3 taken 6 times.
✗ Branch 4 not taken.
✓ Branch 5 taken 6 times.
 6 return IsValidCCS(lhs) && IsValidCCS(rhs) && lhs.cols_count == rhs.rows_count;
150 }
151
152 6 bool DilshodovASpmmDoubleCssAll::PreProcessingImpl() {
153 6 GetOutput() = SparseMatrixCCS{};
154 6 return true;
155 }
156
157 6 bool DilshodovASpmmDoubleCssAll::RunImpl() {
158 const auto &lhs = std::get<0>(GetInput());
159 const auto &rhs = std::get<1>(GetInput());
160 auto &out = GetOutput();
161
162 6 int rank = 0;
163 6 int size = 1;
164 6 MPI_Comm_rank(MPI_COMM_WORLD, &rank);
165 6 MPI_Comm_size(MPI_COMM_WORLD, &size);
166
167 6 out.rows_count = lhs.rows_count;
168 6 out.cols_count = rhs.cols_count;
169
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 6 times.
 6 out.col_ptrs.assign(static_cast<size_t>(out.cols_count) + 1, 0);
170 out.row_indices.clear();
171 out.values.clear();
172
173 6 std::vector<int> col_starts(static_cast<size_t>(size) + 1);
174
2/2
✓ Branch 0 taken 18 times.
✓ Branch 1 taken 6 times.
 24 for (int proc = 0; proc <= size; ++proc) {
175 18 col_starts[proc] = static_cast<int>((static_cast<int64_t>(proc) * rhs.cols_count) / size);
176 }
177
178
1/2
✓ Branch 0 taken 6 times.
✗ Branch 1 not taken.
 6 const int j_start = col_starts[rank];
179 6 const int j_end = col_starts[rank + 1];
180 6 const int local_cols = j_end - j_start;
181
182 6 std::vector<int> local_col_ptrs;
183 6 std::vector<int> local_row_indices;
184 6 std::vector<double> local_values;
185
186
1/2
✓ Branch 0 taken 6 times.
✗ Branch 1 not taken.
 6 if (local_cols > 0) {
187
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 ComputeLocalSlab(lhs, rhs, j_start, j_end, local_col_ptrs, local_row_indices, local_values);
188 } else {
189 ✗ local_col_ptrs.assign(1, 0);
190 }
191
192
1/2
✓ Branch 0 taken 6 times.
✗ Branch 1 not taken.
 6 const int local_nnz = local_col_ptrs.empty() ? 0 : local_col_ptrs.back();
193
194
2/6
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 6 times.
✗ Branch 5 not taken.
✗ Branch 6 not taken.
✗ Branch 7 not taken.
 6 std::vector<int> nnz_per_proc(size, 0);
195
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 MPI_Allgather(&local_nnz, 1, MPI_INT, nnz_per_proc.data(), 1, MPI_INT, MPI_COMM_WORLD);
196
197
1/4
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
 6 std::vector<int> nnz_displs(size, 0);
198 int total_nnz = 0;
199
2/2
✓ Branch 0 taken 12 times.
✓ Branch 1 taken 6 times.
 18 for (int proc = 0; proc < size; ++proc) {
200 12 nnz_displs[proc] = total_nnz;
201 12 total_nnz += nnz_per_proc[proc];
202 }
203
204
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 out.row_indices.resize(total_nnz);
205
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 out.values.resize(total_nnz);
206
207
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 MPI_Allgatherv(local_row_indices.data(), local_nnz, MPI_INT, out.row_indices.data(), nnz_per_proc.data(),
208 nnz_displs.data(), MPI_INT, MPI_COMM_WORLD);
209
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 MPI_Allgatherv(local_values.data(), local_nnz, MPI_DOUBLE, out.values.data(), nnz_per_proc.data(), nnz_displs.data(),
210 MPI_DOUBLE, MPI_COMM_WORLD);
211
212
1/4
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
 6 std::vector<int> cols_per_proc(size, 0);
213
1/4
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
 6 std::vector<int> col_displs(size, 0);
214
2/2
✓ Branch 0 taken 12 times.
✓ Branch 1 taken 6 times.
 18 for (int proc = 0; proc < size; ++proc) {
215 12 cols_per_proc[proc] = col_starts[proc + 1] - col_starts[proc];
216 12 col_displs[proc] = col_starts[proc];
217 }
218
219
1/4
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
 6 std::vector<int> local_inner(static_cast<size_t>(local_cols));
220
2/2
✓ Branch 0 taken 7 times.
✓ Branch 1 taken 6 times.
 13 for (int col_idx = 0; col_idx < local_cols; ++col_idx) {
221 7 local_inner[col_idx] = local_col_ptrs[col_idx] + nnz_displs[rank];
222 }
223
224
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 MPI_Allgatherv(local_inner.data(), local_cols, MPI_INT, out.col_ptrs.data(), cols_per_proc.data(), col_displs.data(),
225 MPI_INT, MPI_COMM_WORLD);
226
1/2
✓ Branch 0 taken 6 times.
✗ Branch 1 not taken.
 6 out.col_ptrs[out.cols_count] = total_nnz;
227
228
1/2
✓ Branch 0 taken 6 times.
✗ Branch 1 not taken.
 6 out.non_zeros = total_nnz;
229 6 return true;
230 }
231
232 6 bool DilshodovASpmmDoubleCssAll::PostProcessingImpl() {
233 6 GetOutput().non_zeros = static_cast<int>(GetOutput().values.size());
234 6 return true;
235 }
236
237 } // namespace dilshodov_a_spmm_double_css
238