| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #include "zhurin_i_matrix_sums/mpi/include/ops_mpi.hpp" | ||
| 2 | |||
| 3 | #include <mpi.h> | ||
| 4 | |||
| 5 | #include <cstdint> | ||
| 6 | #include <vector> | ||
| 7 | |||
| 8 | #include "zhurin_i_matrix_sums/common/include/common.hpp" | ||
| 9 | |||
| 10 | namespace zhurin_i_matrix_sums { | ||
| 11 | |||
| 12 |
1/2✓ Branch 1 taken 20 times.
✗ Branch 2 not taken.
|
20 | ZhurinIMatrixSumsMPI::ZhurinIMatrixSumsMPI(const InType &in) { |
| 13 | SetTypeOfTask(GetStaticTypeOfTask()); | ||
| 14 | GetInput() = in; | ||
| 15 | 20 | GetOutput() = 0.0; | |
| 16 | 20 | } | |
| 17 | |||
| 18 | 20 | bool ZhurinIMatrixSumsMPI::ValidationImpl() { | |
| 19 |
1/2✓ Branch 0 taken 20 times.
✗ Branch 1 not taken.
|
20 | return (static_cast<uint64_t>(std::get<0>(GetInput())) * std::get<1>(GetInput()) == std::get<2>(GetInput()).size()) && |
| 20 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 20 times.
|
20 | (GetOutput() == 0.0); |
| 21 | } | ||
| 22 | |||
| 23 | 20 | bool ZhurinIMatrixSumsMPI::PreProcessingImpl() { | |
| 24 | 20 | GetOutput() = 0.0; | |
| 25 | 20 | return true; | |
| 26 | } | ||
| 27 | |||
| 28 | 20 | bool ZhurinIMatrixSumsMPI::RunImpl() { | |
| 29 | 20 | int rank = 0; | |
| 30 | 20 | int wsize = 0; | |
| 31 | 20 | MPI_Comm_size(MPI_COMM_WORLD, &wsize); | |
| 32 | 20 | MPI_Comm_rank(MPI_COMM_WORLD, &rank); | |
| 33 | |||
| 34 | 20 | int rows = static_cast<int>(std::get<0>(GetInput())); | |
| 35 | 20 | int columns = static_cast<int>(std::get<1>(GetInput())); | |
| 36 | |||
| 37 | 20 | MPI_Bcast(&rows, 1, MPI_INT, 0, MPI_COMM_WORLD); | |
| 38 | 20 | MPI_Bcast(&columns, 1, MPI_INT, 0, MPI_COMM_WORLD); | |
| 39 | 20 | int size = rows * columns; | |
| 40 | |||
| 41 | 20 | int elems_per_proc = size / wsize; | |
| 42 | 20 | int remainder = size % wsize; | |
| 43 | |||
| 44 | 20 | std::vector<int> counts(wsize, 0); | |
| 45 |
1/4✓ Branch 1 taken 20 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
20 | std::vector<int> displs(wsize, 0); |
| 46 | |||
| 47 | int displ = 0; | ||
| 48 |
2/2✓ Branch 0 taken 40 times.
✓ Branch 1 taken 20 times.
|
60 | for (int proc = 0; proc < wsize; proc++) { |
| 49 |
2/2✓ Branch 0 taken 34 times.
✓ Branch 1 taken 6 times.
|
74 | counts[proc] = elems_per_proc + (proc < remainder ? 1 : 0); |
| 50 | 40 | displs[proc] = displ; | |
| 51 | 40 | displ += counts[proc]; | |
| 52 | } | ||
| 53 | |||
| 54 | 20 | std::vector<double> matrix; | |
| 55 |
2/2✓ Branch 0 taken 10 times.
✓ Branch 1 taken 10 times.
|
20 | if (rank == 0) { |
| 56 |
1/2✓ Branch 1 taken 10 times.
✗ Branch 2 not taken.
|
10 | matrix = std::get<2>(GetInput()); |
| 57 | } | ||
| 58 | |||
| 59 |
1/4✓ Branch 1 taken 20 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
20 | std::vector<double> local_buff(counts[rank], 0); |
| 60 | |||
| 61 |
3/4✓ Branch 0 taken 10 times.
✓ Branch 1 taken 10 times.
✓ Branch 3 taken 20 times.
✗ Branch 4 not taken.
|
30 | MPI_Scatterv(rank == 0 ? matrix.data() : nullptr, counts.data(), displs.data(), MPI_DOUBLE, local_buff.data(), |
| 62 |
2/2✓ Branch 0 taken 10 times.
✓ Branch 1 taken 10 times.
|
20 | counts[rank], MPI_DOUBLE, 0, MPI_COMM_WORLD); |
| 63 | |||
| 64 | 20 | double local_sum = 0.0; | |
| 65 |
2/2✓ Branch 0 taken 79 times.
✓ Branch 1 taken 20 times.
|
99 | for (int i = 0; i < counts[rank]; i++) { |
| 66 | 79 | local_sum += local_buff[i]; | |
| 67 | } | ||
| 68 | |||
| 69 | 20 | double global_sum = 0.0; | |
| 70 |
1/2✓ Branch 1 taken 20 times.
✗ Branch 2 not taken.
|
20 | MPI_Allreduce(&local_sum, &global_sum, 1, MPI_DOUBLE, MPI_SUM, MPI_COMM_WORLD); |
| 71 | |||
| 72 |
2/2✓ Branch 0 taken 19 times.
✓ Branch 1 taken 1 times.
|
20 | GetOutput() = global_sum; |
| 73 | 20 | return true; | |
| 74 | } | ||
| 75 | |||
| 76 | 20 | bool ZhurinIMatrixSumsMPI::PostProcessingImpl() { | |
| 77 | 20 | return true; | |
| 78 | } | ||
| 79 | |||
| 80 | } // namespace zhurin_i_matrix_sums | ||
| 81 |