| 1 | // clang-format off |
| 2 | // RUN: %libomptarget-compile-generic |
| 3 | // RUN: env LIBOMPTARGET_INFO=16 \ |
| 4 | // RUN: %libomptarget-run-generic 2>&1 | %fcheck-generic --check-prefix=DEFAULT |
| 5 | // RUN: %libomptarget-compile-generic -g |
| 6 | // RUN: env LIBOMPTARGET_INFO=16 \ |
| 7 | // RUN: %libomptarget-run-generic 2>&1 | %fcheck-generic --check-prefix=DEFAULT |
| 8 | |
| 9 | // REQUIRES: amdgpu |
| 10 | |
| 11 | __attribute__((optnone)) int optnone() { return 1; } |
| 12 | |
| 13 | int main() { |
| 14 | int N = optnone() * 4098 * 32; |
| 15 | |
| 16 | // DEFAULT: [[NT:(128|256)]] (MaxFlatWorkGroupSize: [[NT]] |
| 17 | #pragma omp target teams distribute parallel for simd |
| 18 | for (int i = 0; i < N; ++i) { |
| 19 | optnone(); |
| 20 | } |
| 21 | // DEFAULT: [[NT:(128|256)]] (MaxFlatWorkGroupSize: [[NT]] |
| 22 | #pragma omp target teams distribute parallel for simd |
| 23 | for (int i = 0; i < N; ++i) { |
| 24 | optnone(); |
| 25 | } |
| 26 | // DEFAULT: [[NT:(128|256)]] (MaxFlatWorkGroupSize: [[NT]] |
| 27 | #pragma omp target teams distribute parallel for simd |
| 28 | for (int i = 0; i < N; ++i) { |
| 29 | optnone(); |
| 30 | } |
| 31 | // DEFAULT: [[NT:(128|256)]] (MaxFlatWorkGroupSize: [[NT]] |
| 32 | #pragma omp target |
| 33 | #pragma omp teams distribute parallel for |
| 34 | for (int i = 0; i < N; ++i) { |
| 35 | optnone(); |
| 36 | } |
| 37 | // DEFAULT: 42 (MaxFlatWorkGroupSize: 1024 |
| 38 | #pragma omp target thread_limit(optnone() * 42) |
| 39 | #pragma omp teams distribute parallel for |
| 40 | for (int i = 0; i < N; ++i) { |
| 41 | optnone(); |
| 42 | } |
| 43 | // DEFAULT: 42 (MaxFlatWorkGroupSize: 42 |
| 44 | #pragma omp target thread_limit(optnone() * 42) ompx_attribute(__attribute__((amdgpu_flat_work_group_size(42, 42)))) |
| 45 | #pragma omp teams distribute parallel for |
| 46 | for (int i = 0; i < N; ++i) { |
| 47 | optnone(); |
| 48 | } |
| 49 | // DEFAULT: 42 (MaxFlatWorkGroupSize: 42 |
| 50 | #pragma omp target ompx_attribute(__attribute__((amdgpu_flat_work_group_size(42, 42)))) |
| 51 | #pragma omp teams distribute parallel for |
| 52 | for (int i = 0; i < N; ++i) { |
| 53 | optnone(); |
| 54 | } |
| 55 | // DEFAULT: MaxFlatWorkGroupSize: 1024 |
| 56 | #pragma omp target |
| 57 | #pragma omp teams distribute parallel for num_threads(optnone() * 42) |
| 58 | for (int i = 0; i < N; ++i) { |
| 59 | optnone(); |
| 60 | } |
| 61 | // DEFAULT: MaxFlatWorkGroupSize: 1024 |
| 62 | #pragma omp target teams distribute parallel for thread_limit(optnone() * 42) |
| 63 | for (int i = 0; i < N; ++i) { |
| 64 | optnone(); |
| 65 | } |
| 66 | // DEFAULT: MaxFlatWorkGroupSize: 1024 |
| 67 | #pragma omp target teams distribute parallel for num_threads(optnone() * 42) |
| 68 | for (int i = 0; i < N; ++i) { |
| 69 | optnone(); |
| 70 | } |
| 71 | // DEFAULT: 9 (MaxFlatWorkGroupSize: 9 |
| 72 | #pragma omp target |
| 73 | #pragma omp teams distribute parallel for num_threads(9) |
| 74 | for (int i = 0; i < N; ++i) { |
| 75 | optnone(); |
| 76 | } |
| 77 | // DEFAULT: 4 (MaxFlatWorkGroupSize: 4 |
| 78 | #pragma omp target thread_limit(4) |
| 79 | #pragma omp teams distribute parallel for |
| 80 | for (int i = 0; i < N; ++i) { |
| 81 | optnone(); |
| 82 | } |
| 83 | // DEFAULT: 4 (MaxFlatWorkGroupSize: 4 |
| 84 | #pragma omp target |
| 85 | #pragma omp teams distribute parallel for thread_limit(4) |
| 86 | for (int i = 0; i < N; ++i) { |
| 87 | optnone(); |
| 88 | } |
| 89 | // DEFAULT: 9 (MaxFlatWorkGroupSize: 9 |
| 90 | #pragma omp target teams distribute parallel for num_threads(9) |
| 91 | for (int i = 0; i < N; ++i) { |
| 92 | optnone(); |
| 93 | } |
| 94 | // DEFAULT: 4 (MaxFlatWorkGroupSize: 4 |
| 95 | #pragma omp target teams distribute parallel for simd thread_limit(4) |
| 96 | for (int i = 0; i < N; ++i) { |
| 97 | optnone(); |
| 98 | } |
| 99 | } |
| 100 | |
| 101 | |