| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #include "shakirova_e_elem_matrix_sum/mpi/include/ops_mpi.hpp" | ||
| 2 | |||
| 3 | #include <mpi.h> | ||
| 4 | |||
| 5 | #include <cstddef> | ||
| 6 | #include <cstdint> | ||
| 7 | #include <utility> | ||
| 8 | #include <vector> | ||
| 9 | |||
| 10 | #include "shakirova_e_elem_matrix_sum/common/include/common.hpp" | ||
| 11 | #include "shakirova_e_elem_matrix_sum/common/include/matrix.hpp" | ||
| 12 | |||
| 13 | namespace shakirova_e_elem_matrix_sum { | ||
| 14 | |||
| 15 |
1/2✓ Branch 1 taken 18 times.
✗ Branch 2 not taken.
|
18 | ShakirovaEElemMatrixSumMPI::ShakirovaEElemMatrixSumMPI(const InType &in) { |
| 16 | SetTypeOfTask(GetStaticTypeOfTask()); | ||
| 17 | GetInput() = in; | ||
| 18 | 18 | GetOutput() = 0; | |
| 19 | 18 | } | |
| 20 | |||
| 21 | 18 | bool ShakirovaEElemMatrixSumMPI::ValidationImpl() { | |
| 22 | 18 | int rank = -1; | |
| 23 | |||
| 24 | 18 | MPI_Comm_rank(MPI_COMM_WORLD, &rank); | |
| 25 |
2/2✓ Branch 0 taken 9 times.
✓ Branch 1 taken 9 times.
|
18 | if (rank == 0) { |
| 26 | return GetInput().IsValid(); | ||
| 27 | } | ||
| 28 | |||
| 29 | return true; | ||
| 30 | } | ||
| 31 | |||
| 32 | 18 | bool ShakirovaEElemMatrixSumMPI::PreProcessingImpl() { | |
| 33 | 18 | GetOutput() = 0; | |
| 34 | 18 | return true; | |
| 35 | } | ||
| 36 | |||
| 37 | 18 | bool ShakirovaEElemMatrixSumMPI::RunImpl() { | |
| 38 | 18 | int rank = 0; | |
| 39 | 18 | int p_count = 0; | |
| 40 | |||
| 41 | 18 | MPI_Comm_rank(MPI_COMM_WORLD, &rank); | |
| 42 | 18 | MPI_Comm_size(MPI_COMM_WORLD, &p_count); | |
| 43 | |||
| 44 |
1/2✓ Branch 0 taken 18 times.
✗ Branch 1 not taken.
|
18 | if (p_count <= 0) { |
| 45 | return false; | ||
| 46 | } | ||
| 47 | |||
| 48 | 18 | size_t row_count = GetInput().rows; | |
| 49 | 18 | size_t col_count = GetInput().cols; | |
| 50 | |||
| 51 | 18 | MPI_Bcast(&row_count, 1, MPI_UINT64_T, 0, MPI_COMM_WORLD); | |
| 52 | 18 | MPI_Bcast(&col_count, 1, MPI_UINT64_T, 0, MPI_COMM_WORLD); | |
| 53 | |||
| 54 | 18 | size_t rows_per_process = row_count / p_count; | |
| 55 | 18 | size_t remaining_rows = row_count % p_count; | |
| 56 | |||
| 57 | size_t my_rows = rows_per_process; | ||
| 58 |
1/2✓ Branch 0 taken 18 times.
✗ Branch 1 not taken.
|
18 | if (std::cmp_less(rank, remaining_rows)) { |
| 59 | 3 | my_rows = my_rows + 1; | |
| 60 | } | ||
| 61 | |||
| 62 | 18 | size_t elements_count = my_rows * col_count; | |
| 63 | 18 | std::vector<int64_t> my_data(elements_count); | |
| 64 | |||
| 65 |
2/2✓ Branch 0 taken 9 times.
✓ Branch 1 taken 9 times.
|
18 | if (rank == 0) { |
| 66 |
1/4✓ Branch 1 taken 9 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
9 | std::vector<int> send_counts(p_count); |
| 67 |
1/4✓ Branch 1 taken 9 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
9 | std::vector<int> displacements(p_count, 0); |
| 68 | |||
| 69 |
2/2✓ Branch 0 taken 18 times.
✓ Branch 1 taken 9 times.
|
27 | for (int i = 0; i < p_count; i++) { |
| 70 | size_t proc_rows = rows_per_process; | ||
| 71 | if (std::cmp_less(i, remaining_rows)) { | ||
| 72 | 3 | proc_rows = proc_rows + 1; | |
| 73 | } | ||
| 74 |
2/2✓ Branch 0 taken 9 times.
✓ Branch 1 taken 9 times.
|
18 | send_counts[i] = static_cast<int>(proc_rows * col_count); |
| 75 | |||
| 76 |
2/2✓ Branch 0 taken 9 times.
✓ Branch 1 taken 9 times.
|
18 | if (i > 0) { |
| 77 | 9 | displacements[i] = displacements[i - 1] + send_counts[i - 1]; | |
| 78 | } | ||
| 79 | } | ||
| 80 | |||
| 81 |
1/2✓ Branch 1 taken 9 times.
✗ Branch 2 not taken.
|
9 | MPI_Scatterv(GetInput().data.data(), send_counts.data(), displacements.data(), MPI_INT64_T, my_data.data(), |
| 82 | static_cast<int>(elements_count), MPI_INT64_T, 0, MPI_COMM_WORLD); | ||
| 83 | } else { | ||
| 84 |
1/2✓ Branch 1 taken 9 times.
✗ Branch 2 not taken.
|
9 | MPI_Scatterv(nullptr, nullptr, nullptr, MPI_INT64_T, my_data.data(), static_cast<int>(elements_count), MPI_INT64_T, |
| 85 | 0, MPI_COMM_WORLD); | ||
| 86 | } | ||
| 87 | |||
| 88 | 18 | int64_t local_sum = 0; | |
| 89 |
2/2✓ Branch 0 taken 2491 times.
✓ Branch 1 taken 18 times.
|
2509 | for (const auto &value : my_data) { |
| 90 | 2491 | local_sum = local_sum + value; | |
| 91 | } | ||
| 92 | |||
| 93 | 18 | int64_t total_sum = 0; | |
| 94 |
1/2✓ Branch 1 taken 18 times.
✗ Branch 2 not taken.
|
18 | MPI_Allreduce(&local_sum, &total_sum, 1, MPI_INT64_T, MPI_SUM, MPI_COMM_WORLD); |
| 95 | |||
| 96 |
2/2✓ Branch 0 taken 17 times.
✓ Branch 1 taken 1 times.
|
18 | GetOutput() = total_sum; |
| 97 | |||
| 98 | return true; | ||
| 99 | } | ||
| 100 | |||
| 101 | 18 | bool ShakirovaEElemMatrixSumMPI::PostProcessingImpl() { | |
| 102 | 18 | return true; | |
| 103 | } | ||
| 104 | |||
| 105 | } // namespace shakirova_e_elem_matrix_sum | ||
| 106 |