| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #include "batkov_f_vector_sum/mpi/include/ops_mpi.hpp" | ||
| 2 | |||
| 3 | #include <mpi.h> | ||
| 4 | |||
| 5 | #include <cstddef> | ||
| 6 | #include <limits> | ||
| 7 | #include <vector> | ||
| 8 | |||
| 9 | #include "batkov_f_vector_sum/common/include/common.hpp" | ||
| 10 | |||
| 11 | namespace batkov_f_vector_sum { | ||
| 12 | |||
| 13 |
1/2✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
|
8 | BatkovFVectorSumMPI::BatkovFVectorSumMPI(const InType &in) { |
| 14 | SetTypeOfTask(GetStaticTypeOfTask()); | ||
| 15 | |||
| 16 | 8 | int rank = 0; | |
| 17 | 8 | int mpi_size = 0; | |
| 18 |
1/2✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
|
8 | MPI_Comm_rank(MPI_COMM_WORLD, &rank); |
| 19 |
1/2✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
|
8 | MPI_Comm_size(MPI_COMM_WORLD, &mpi_size); |
| 20 | |||
| 21 | 8 | m_rank_ = static_cast<size_t>(rank); | |
| 22 | 8 | m_mpi_size_ = static_cast<size_t>(mpi_size); | |
| 23 | |||
| 24 |
2/2✓ Branch 0 taken 4 times.
✓ Branch 1 taken 4 times.
|
8 | if (m_rank_ == 0) { |
| 25 |
1/2✓ Branch 1 taken 4 times.
✗ Branch 2 not taken.
|
4 | GetInput() = in; |
| 26 | } else { | ||
| 27 | 4 | GetInput() = std::vector<int>(); | |
| 28 | } | ||
| 29 | |||
| 30 | 8 | GetOutput() = 0; | |
| 31 | 8 | } | |
| 32 | |||
| 33 | 8 | bool BatkovFVectorSumMPI::ValidationImpl() { | |
| 34 | 8 | return GetOutput() == 0; | |
| 35 | } | ||
| 36 | |||
| 37 | 8 | bool BatkovFVectorSumMPI::PreProcessingImpl() { | |
| 38 | 8 | return true; | |
| 39 | } | ||
| 40 | |||
| 41 | 8 | bool BatkovFVectorSumMPI::RunImpl() { | |
| 42 | 8 | size_t input_size = 0; | |
| 43 |
2/2✓ Branch 0 taken 4 times.
✓ Branch 1 taken 4 times.
|
8 | if (m_rank_ == 0) { |
| 44 | 4 | input_size = GetInput().size(); | |
| 45 | } | ||
| 46 | |||
| 47 | 8 | MPI_Bcast(&input_size, 1, MPI_UNSIGNED_LONG_LONG, 0, MPI_COMM_WORLD); | |
| 48 | |||
| 49 |
2/2✓ Branch 0 taken 2 times.
✓ Branch 1 taken 6 times.
|
8 | if (input_size == 0) { |
| 50 | 2 | GetOutput() = 0; | |
| 51 | 2 | return true; | |
| 52 | } | ||
| 53 | |||
| 54 | 6 | const size_t base_chunk = input_size / m_mpi_size_; | |
| 55 | 6 | const size_t remaining_elements = input_size % m_mpi_size_; | |
| 56 | |||
| 57 | size_t chunk_size = base_chunk; | ||
| 58 |
2/2✓ Branch 0 taken 1 times.
✓ Branch 1 taken 5 times.
|
6 | if (m_rank_ < remaining_elements) { |
| 59 | 1 | chunk_size = base_chunk + 1; | |
| 60 | } | ||
| 61 | 6 | std::vector<int> local_data(chunk_size); | |
| 62 | |||
| 63 |
1/4✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
6 | std::vector<int> send_counts(m_mpi_size_); |
| 64 |
1/4✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
6 | std::vector<int> displs(m_mpi_size_); |
| 65 | size_t disp = 0; | ||
| 66 | |||
| 67 |
2/2✓ Branch 0 taken 12 times.
✓ Branch 1 taken 6 times.
|
18 | for (size_t i = 0; i < m_mpi_size_; ++i) { |
| 68 | size_t chunk = base_chunk; | ||
| 69 |
2/2✓ Branch 0 taken 2 times.
✓ Branch 1 taken 10 times.
|
12 | if (i < remaining_elements) { |
| 70 | 2 | chunk = base_chunk + 1; | |
| 71 | } | ||
| 72 | |||
| 73 | 12 | int int_chunk = static_cast<int>(chunk); | |
| 74 | 12 | int int_disp = static_cast<int>(disp); | |
| 75 | |||
| 76 | if (int_chunk > std::numeric_limits<int>::max() || int_disp > std::numeric_limits<int>::max()) { | ||
| 77 | return false; | ||
| 78 | } | ||
| 79 | |||
| 80 | 12 | send_counts[i] = int_chunk; | |
| 81 | 12 | displs[i] = int_disp; | |
| 82 | 12 | disp += chunk; | |
| 83 | } | ||
| 84 | |||
| 85 |
1/2✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
|
6 | MPI_Scatterv(GetInput().data(), send_counts.data(), displs.data(), MPI_INT, local_data.data(), |
| 86 | static_cast<int>(chunk_size), MPI_INT, 0, MPI_COMM_WORLD); | ||
| 87 | |||
| 88 | 6 | int local_sum = 0; | |
| 89 |
2/2✓ Branch 0 taken 399 times.
✓ Branch 1 taken 6 times.
|
405 | for (int val : local_data) { |
| 90 | 399 | local_sum += val; | |
| 91 | } | ||
| 92 | |||
| 93 | 6 | int global_sum = 0; | |
| 94 |
1/2✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
|
6 | MPI_Allreduce(&local_sum, &global_sum, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD); |
| 95 | |||
| 96 |
1/2✓ Branch 0 taken 6 times.
✗ Branch 1 not taken.
|
6 | GetOutput() = global_sum; |
| 97 | return true; | ||
| 98 | } | ||
| 99 | |||
| 100 | 8 | bool BatkovFVectorSumMPI::PostProcessingImpl() { | |
| 101 | 8 | return true; | |
| 102 | } | ||
| 103 | |||
| 104 | } // namespace batkov_f_vector_sum | ||
| 105 |