GCC Code Coverage Report

Directory: ./
File: tasks/viderman_a_sparse_matrix_mult_crs_complex/all/src/ops_all.cpp
Date: 2026-06-04 20:25:32
Exec Total Coverage
Lines: 188 197 95.4%
Functions: 16 16 100.0%
Branches: 107 160 66.9%
Line Branch Exec Source
1 #include "viderman_a_sparse_matrix_mult_crs_complex/all/include/ops_all.hpp"
2
3 #include <mpi.h>
4 #include <omp.h>
5
6 #include <algorithm>
7 #include <complex>
8 #include <vector>
9
10 #include "viderman_a_sparse_matrix_mult_crs_complex/common/include/common.hpp"
11
12 namespace viderman_a_sparse_matrix_mult_crs_complex {
13
14
1/2
✓ Branch 2 taken 12 times.
✗ Branch 3 not taken.
 12 VidermanASparseMatrixMultCRSComplexALL::VidermanASparseMatrixMultCRSComplexALL(const InType &in) {
15 SetTypeOfTask(GetStaticTypeOfTask());
16 GetInput() = in;
17
1/2
✓ Branch 1 taken 12 times.
✗ Branch 2 not taken.
 12 GetOutput() = CRSMatrix(0, 0);
18 12 }
19
20 12 bool VidermanASparseMatrixMultCRSComplexALL::ValidationImpl() {
21 12 MPI_Initialized(&initialized_pipiline_);
22
23 12 rank_ = 0;
24
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 if (initialized_pipiline_ != 0) {
25 12 MPI_Comm_rank(MPI_COMM_WORLD, &rank_);
26 }
27
28 12 int is_valid = 1;
29
2/2
✓ Branch 0 taken 6 times.
✓ Branch 1 taken 6 times.
 12 if (rank_ == 0) {
30 const auto &input = GetInput();
31 const auto &a = std::get<0>(input);
32 const auto &b = std::get<1>(input);
33
34
4/6
✓ Branch 0 taken 6 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 6 times.
✗ Branch 3 not taken.
✓ Branch 4 taken 1 times.
✓ Branch 5 taken 5 times.
 6 if (!a.IsValid() || !b.IsValid() || a.cols != b.rows) {
35 1 is_valid = 0;
36 }
37 }
38
39
1/2
✓ Branch 0 taken 12 times.
✗ Branch 1 not taken.
 12 if (initialized_pipiline_ != 0) {
40 12 MPI_Bcast(&is_valid, 1, MPI_INT, 0, MPI_COMM_WORLD);
41 }
42
43 12 return is_valid == 1;
44 }
45
46 10 bool VidermanASparseMatrixMultCRSComplexALL::PreProcessingImpl() {
47
1/2
✓ Branch 0 taken 10 times.
✗ Branch 1 not taken.
 10 if (initialized_pipiline_ != 0) {
48 10 MPI_Comm_rank(MPI_COMM_WORLD, &rank_);
49 10 MPI_Comm_size(MPI_COMM_WORLD, &world_size_);
50 } else {
51 ✗ rank_ = 0;
52 ✗ world_size_ = 1;
53 }
54
55
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 5 times.
 10 if (rank_ == 0) {
56 const auto &input = GetInput();
57 const auto &a = std::get<0>(input);
58 5 b_ = std::get<1>(input);
59
60 5 a_rows_ = a.rows;
61 5 a_cols_ = a.cols;
62 }
63
64
1/2
✓ Branch 0 taken 10 times.
✗ Branch 1 not taken.
 10 if (initialized_pipiline_ != 0) {
65 10 MPI_Bcast(&a_rows_, 1, MPI_INT, 0, MPI_COMM_WORLD);
66 10 MPI_Bcast(&a_cols_, 1, MPI_INT, 0, MPI_COMM_WORLD);
67 }
68
69 10 BroadcastData();
70 10 ScatterData();
71
72 10 return true;
73 }
74
75 10 bool VidermanASparseMatrixMultCRSComplexALL::RunImpl() {
76 10 MultiplyLocal();
77
78 10 return true;
79 }
80
81 10 bool VidermanASparseMatrixMultCRSComplexALL::PostProcessingImpl() {
82 10 GatherData();
83
84 10 return true;
85 }
86
87 5 void VidermanASparseMatrixMultCRSComplexALL::MasterScatter() {
88 5 SendDataProcesses();
89
90 5 int count = a_rows_ / world_size_;
91
2/2
✓ Branch 0 taken 2 times.
✓ Branch 1 taken 3 times.
 5 if (a_rows_ % world_size_ > 0) {
92 2 count++;
93 }
94
95 5 local_a_ = CRSMatrix(count, a_cols_);
96 const auto &a = std::get<0>(GetInput());
97
2/4
✓ Branch 0 taken 5 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 5 times.
✗ Branch 3 not taken.
 5 if (count > 0 && !a.row_ptr.empty()) {
98 5 int nnz = a.row_ptr[count] - a.row_ptr[0];
99 5 std::copy(a.row_ptr.begin(), a.row_ptr.begin() + count + 1, local_a_.row_ptr.begin());
100
101 5 local_a_.col_indices.assign(a.col_indices.begin(), a.col_indices.begin() + nnz);
102 5 local_a_.values.assign(a.values.begin(), a.values.begin() + nnz);
103 }
104 5 }
105
106 5 void VidermanASparseMatrixMultCRSComplexALL::WorkerScatter() {
107 5 int count = 0;
108 5 MPI_Recv(&count, 1, MPI_INT, 0, 0, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
109
110 5 local_a_ = CRSMatrix(count, a_cols_);
111
112
2/2
✓ Branch 0 taken 4 times.
✓ Branch 1 taken 1 times.
 5 if (count > 0) {
113 4 int nnz = 0;
114 4 MPI_Recv(&nnz, 1, MPI_INT, 0, 1, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
115
116 4 local_a_.row_ptr.resize(count + 1);
117 4 MPI_Recv(local_a_.row_ptr.data(), count + 1, MPI_INT, 0, 2, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
118
119
2/2
✓ Branch 0 taken 3 times.
✓ Branch 1 taken 1 times.
 4 if (nnz > 0) {
120 3 local_a_.col_indices.resize(nnz);
121 3 local_a_.values.resize(nnz);
122
123 3 MPI_Recv(local_a_.col_indices.data(), nnz, MPI_INT, 0, 3, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
124 3 MPI_Recv(reinterpret_cast<double *>(local_a_.values.data()), nnz * 2, MPI_DOUBLE, 0, 4, MPI_COMM_WORLD,
125 MPI_STATUS_IGNORE);
126 }
127 }
128 5 }
129
130 10 void VidermanASparseMatrixMultCRSComplexALL::ScatterData() {
131
2/4
✓ Branch 0 taken 10 times.
✗ Branch 1 not taken.
✗ Branch 2 not taken.
✓ Branch 3 taken 10 times.
 10 if (initialized_pipiline_ == 0 || world_size_ == 1) {
132 ✗ if (rank_ == 0) {
133 ✗ local_a_ = std::get<0>(GetInput());
134 }
135
136 ✗ return;
137 }
138
139
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 5 times.
 10 if (rank_ == 0) {
140 5 MasterScatter();
141 } else {
142 5 WorkerScatter();
143 }
144 }
145
146 5 void VidermanASparseMatrixMultCRSComplexALL::MasterGather() {
147 CRSMatrix &out = GetOutput();
148 5 out = CRSMatrix(a_rows_, b_.cols);
149
150 int rows_to_copy = std::min(local_c_.rows, a_rows_);
151
2/2
✓ Branch 0 taken 11 times.
✓ Branch 1 taken 5 times.
 16 for (int i = 0; i <= rows_to_copy; ++i) {
152 11 out.row_ptr[i] = local_c_.row_ptr[i];
153 }
154
155 5 out.col_indices = local_c_.col_indices;
156 5 out.values = local_c_.values;
157
158 5 ReceiveDataProcesses(out);
159 5 }
160
161 5 void VidermanASparseMatrixMultCRSComplexALL::WorkerGather() {
162 5 int send_rows = local_c_.rows;
163 5 int send_nnz = static_cast<int>(local_c_.values.size());
164
165 5 MPI_Send(&send_rows, 1, MPI_INT, 0, 0, MPI_COMM_WORLD);
166
167
2/2
✓ Branch 0 taken 4 times.
✓ Branch 1 taken 1 times.
 5 if (send_rows > 0) {
168 4 MPI_Send(&send_nnz, 1, MPI_INT, 0, 1, MPI_COMM_WORLD);
169 4 MPI_Send(local_c_.row_ptr.data(), send_rows + 1, MPI_INT, 0, 2, MPI_COMM_WORLD);
170
171
2/2
✓ Branch 0 taken 2 times.
✓ Branch 1 taken 2 times.
 4 if (send_nnz > 0) {
172 2 MPI_Send(local_c_.col_indices.data(), send_nnz, MPI_INT, 0, 3, MPI_COMM_WORLD);
173 2 MPI_Send(reinterpret_cast<const double *>(local_c_.values.data()), send_nnz * 2, MPI_DOUBLE, 0, 4,
174 MPI_COMM_WORLD);
175 }
176 }
177 5 }
178
179 10 void VidermanASparseMatrixMultCRSComplexALL::GatherData() {
180
2/4
✓ Branch 0 taken 10 times.
✗ Branch 1 not taken.
✗ Branch 2 not taken.
✓ Branch 3 taken 10 times.
 10 if (initialized_pipiline_ == 0 || world_size_ == 1) {
181 ✗ if (rank_ == 0) {
182 ✗ GetOutput() = local_c_;
183 }
184
185 ✗ return;
186 }
187
188
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 5 times.
 10 if (rank_ == 0) {
189 5 MasterGather();
190 } else {
191 5 WorkerGather();
192 }
193
194 CRSMatrix &out = GetOutput();
195
196 10 std::vector<int> g_info(3, 0);
197
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 5 times.
 10 if (rank_ == 0) {
198 5 g_info[0] = out.rows;
199 5 g_info[1] = out.cols;
200 5 g_info[2] = static_cast<int>(out.values.size());
201 }
202
203
1/2
✓ Branch 1 taken 10 times.
✗ Branch 2 not taken.
 10 MPI_Bcast(g_info.data(), 3, MPI_INT, 0, MPI_COMM_WORLD);
204
205
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 5 times.
 10 if (rank_ != 0) {
206 5 out.rows = g_info[0];
207 5 out.cols = g_info[1];
208
1/4
✓ Branch 1 taken 5 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
 5 out.row_ptr.assign(out.rows + 1, 0);
209
1/2
✓ Branch 1 taken 5 times.
✗ Branch 2 not taken.
 5 out.col_indices.resize(g_info[2]);
210
1/2
✓ Branch 1 taken 5 times.
✗ Branch 2 not taken.
 5 out.values.resize(g_info[2]);
211 }
212
213
1/2
✓ Branch 0 taken 10 times.
✗ Branch 1 not taken.
 10 if (out.rows >= 0) {
214
1/2
✓ Branch 1 taken 10 times.
✗ Branch 2 not taken.
 10 MPI_Bcast(out.row_ptr.data(), out.rows + 1, MPI_INT, 0, MPI_COMM_WORLD);
215 }
216
217
2/2
✓ Branch 0 taken 6 times.
✓ Branch 1 taken 4 times.
 10 if (g_info[2] > 0) {
218
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 MPI_Bcast(out.col_indices.data(), g_info[2], MPI_INT, 0, MPI_COMM_WORLD);
219
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 MPI_Bcast(reinterpret_cast<double *>(out.values.data()), g_info[2] * 2, MPI_DOUBLE, 0, MPI_COMM_WORLD);
220 }
221 }
222
223 5 void VidermanASparseMatrixMultCRSComplexALL::SendDataProcesses() {
224 const auto &a = std::get<0>(GetInput());
225 5 int rows_per_proc = a_rows_ / world_size_;
226 5 int remainder = a_rows_ % world_size_;
227
228
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 5 times.
 10 for (int process = 1; process < world_size_; ++process) {
229
1/2
✓ Branch 0 taken 5 times.
✗ Branch 1 not taken.
 5 int start = (process * rows_per_proc) + std::min(process, remainder);
230
1/2
✓ Branch 0 taken 5 times.
✗ Branch 1 not taken.
 5 int count = rows_per_proc + (process < remainder ? 1 : 0);
231 5 MPI_Send(&count, 1, MPI_INT, process, 0, MPI_COMM_WORLD);
232
233
2/2
✓ Branch 0 taken 4 times.
✓ Branch 1 taken 1 times.
 5 if (count > 0) {
234 4 int nnz = a.row_ptr[start + count] - a.row_ptr[start];
235 4 MPI_Send(&nnz, 1, MPI_INT, process, 1, MPI_COMM_WORLD);
236
237 4 std::vector<int> p_row_ptr(count + 1);
238
239 4 int offset = a.row_ptr[start];
240
2/2
✓ Branch 0 taken 8 times.
✓ Branch 1 taken 4 times.
 12 for (int i = 0; i <= count; ++i) {
241 8 p_row_ptr[i] = a.row_ptr[start + i] - offset;
242 }
243
244
1/2
✓ Branch 1 taken 4 times.
✗ Branch 2 not taken.
 4 MPI_Send(p_row_ptr.data(), count + 1, MPI_INT, process, 2, MPI_COMM_WORLD);
245
246
2/2
✓ Branch 0 taken 3 times.
✓ Branch 1 taken 1 times.
 4 if (nnz > 0) {
247
1/2
✓ Branch 1 taken 3 times.
✗ Branch 2 not taken.
 3 MPI_Send(&a.col_indices[offset], nnz, MPI_INT, process, 3, MPI_COMM_WORLD);
248
1/2
✓ Branch 1 taken 3 times.
✗ Branch 2 not taken.
 3 MPI_Send(reinterpret_cast<const double *>(&a.values[offset]), nnz * 2, MPI_DOUBLE, process, 4, MPI_COMM_WORLD);
249 }
250 }
251 }
252 5 }
253
254 5 void VidermanASparseMatrixMultCRSComplexALL::ReceiveDataProcesses(CRSMatrix &out) const {
255 5 int current_global_row = local_c_.rows;
256
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 5 times.
 10 for (int process = 1; process < world_size_; ++process) {
257 5 int p_rows = 0;
258 5 MPI_Recv(&p_rows, 1, MPI_INT, process, 0, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
259
260
2/2
✓ Branch 0 taken 4 times.
✓ Branch 1 taken 1 times.
 5 if (p_rows > 0) {
261 4 int p_nnz = 0;
262 4 MPI_Recv(&p_nnz, 1, MPI_INT, process, 1, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
263
264 4 std::vector<int> p_ptr(p_rows + 1);
265
1/2
✓ Branch 1 taken 4 times.
✗ Branch 2 not taken.
 4 MPI_Recv(p_ptr.data(), p_rows + 1, MPI_INT, process, 2, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
266
267 4 int offset = static_cast<int>(out.values.size());
268
269
2/2
✓ Branch 0 taken 2 times.
✓ Branch 1 taken 2 times.
 4 if (p_nnz > 0) {
270
1/4
✓ Branch 1 taken 2 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
 2 std::vector<int> p_cols(p_nnz);
271
1/4
✓ Branch 1 taken 2 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
 2 std::vector<Complex> p_vals(p_nnz);
272
273
1/2
✓ Branch 1 taken 2 times.
✗ Branch 2 not taken.
 2 MPI_Recv(p_cols.data(), p_nnz, MPI_INT, process, 3, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
274
1/2
✓ Branch 1 taken 2 times.
✗ Branch 2 not taken.
 2 MPI_Recv(reinterpret_cast<double *>(p_vals.data()), p_nnz * 2, MPI_DOUBLE, process, 4, MPI_COMM_WORLD,
275 MPI_STATUS_IGNORE);
276
277
1/2
✓ Branch 1 taken 2 times.
✗ Branch 2 not taken.
 2 out.col_indices.insert(out.col_indices.end(), p_cols.begin(), p_cols.end());
278
2/4
✓ Branch 1 taken 2 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 2 times.
✗ Branch 4 not taken.
 2 out.values.insert(out.values.end(), p_vals.begin(), p_vals.end());
279 }
280
2/2
✓ Branch 0 taken 4 times.
✓ Branch 1 taken 4 times.
 8 for (int i = 1; i <= p_rows; ++i) {
281
1/2
✓ Branch 0 taken 4 times.
✗ Branch 1 not taken.
 4 if (current_global_row + i <= a_rows_) {
282 4 out.row_ptr[current_global_row + i] = offset + p_ptr[i];
283 }
284 }
285
286
1/2
✓ Branch 0 taken 4 times.
✗ Branch 1 not taken.
 4 current_global_row += p_rows;
287 }
288 }
289 5 }
290
291 10 void VidermanASparseMatrixMultCRSComplexALL::BroadcastData() {
292
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 10 times.
 10 if (initialized_pipiline_ == 0) {
293 ✗ return;
294 }
295
296 10 MPI_Bcast(&b_.rows, 1, MPI_INT, 0, MPI_COMM_WORLD);
297 10 MPI_Bcast(&b_.cols, 1, MPI_INT, 0, MPI_COMM_WORLD);
298
299
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 5 times.
 10 int nnz = (rank_ == 0) ? static_cast<int>(b_.values.size()) : 0;
300 10 MPI_Bcast(&nnz, 1, MPI_INT, 0, MPI_COMM_WORLD);
301
302
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 5 times.
 10 if (rank_ != 0) {
303 5 b_.row_ptr.assign(b_.rows + 1, 0);
304 5 b_.col_indices.resize(nnz);
305 5 b_.values.resize(nnz);
306 }
307
308
1/2
✓ Branch 0 taken 10 times.
✗ Branch 1 not taken.
 10 if (!b_.row_ptr.empty()) {
309 10 MPI_Bcast(b_.row_ptr.data(), b_.rows + 1, MPI_INT, 0, MPI_COMM_WORLD);
310 }
311
312
2/2
✓ Branch 0 taken 6 times.
✓ Branch 1 taken 4 times.
 10 if (nnz > 0) {
313 6 MPI_Bcast(b_.col_indices.data(), nnz, MPI_INT, 0, MPI_COMM_WORLD);
314 6 MPI_Bcast(reinterpret_cast<double *>(b_.values.data()), nnz * 2, MPI_DOUBLE, 0, MPI_COMM_WORLD);
315 }
316 }
317
318 10 void VidermanASparseMatrixMultCRSComplexALL::ComputeMultiply(int i, std::vector<int> &cols,
319 std::vector<Complex> &vals) {
320 10 std::vector<Complex> accumulator(b_.cols, Complex(0, 0));
321 bool has_data = false;
322
323
2/2
✓ Branch 0 taken 7 times.
✓ Branch 1 taken 10 times.
 17 for (int j = local_a_.row_ptr[i]; j < local_a_.row_ptr[i + 1]; ++j) {
324 7 int a_col = local_a_.col_indices[j];
325 7 Complex a_val = local_a_.values[j];
326
327
2/2
✓ Branch 0 taken 7 times.
✓ Branch 1 taken 7 times.
 14 for (int k = b_.row_ptr[a_col]; k < b_.row_ptr[a_col + 1]; ++k) {
328 7 accumulator[b_.col_indices[k]] += a_val * b_.values[k];
329
330 has_data = true;
331 }
332 }
333
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 5 times.
 10 if (has_data) {
334
2/2
✓ Branch 0 taken 9 times.
✓ Branch 1 taken 5 times.
 14 for (int j = 0; j < b_.cols; ++j) {
335
2/2
✓ Branch 0 taken 7 times.
✓ Branch 1 taken 2 times.
 9 if (std::norm(accumulator[j]) > 1e-18) {
336 cols.push_back(j);
337 vals.push_back(accumulator[j]);
338 }
339 }
340 }
341 10 }
342
343 10 void VidermanASparseMatrixMultCRSComplexALL::MultiplyLocal() {
344 10 local_c_ = CRSMatrix(local_a_.rows, b_.cols);
345
3/4
✓ Branch 0 taken 9 times.
✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 9 times.
 10 if (local_a_.rows <= 0 || b_.cols <= 0) {
346 1 return;
347 }
348
349 9 std::vector<std::vector<int>> row_cols(local_a_.rows);
350
1/2
✓ Branch 1 taken 9 times.
✗ Branch 2 not taken.
 9 std::vector<std::vector<Complex>> row_vals(local_a_.rows);
351
352 9 #pragma omp parallel for default(none) shared(row_cols, row_vals)
353 for (int i = 0; i < local_a_.rows; ++i) {
354 ComputeMultiply(i, row_cols[i], row_vals[i]);
355 }
356
357
2/2
✓ Branch 0 taken 10 times.
✓ Branch 1 taken 9 times.
 19 for (int i = 0; i < local_a_.rows; ++i) {
358
1/2
✓ Branch 1 taken 10 times.
✗ Branch 2 not taken.
 10 local_c_.col_indices.insert(local_c_.col_indices.end(), row_cols[i].begin(), row_cols[i].end());
359
1/2
✓ Branch 1 taken 10 times.
✗ Branch 2 not taken.
 10 local_c_.values.insert(local_c_.values.end(), row_vals[i].begin(), row_vals[i].end());
360 10 local_c_.row_ptr[i + 1] = static_cast<int>(local_c_.values.size());
361 }
362 9 }
363
364 } // namespace viderman_a_sparse_matrix_mult_crs_complex
365