2 * Copyright (c) 2017-2018 ARM Limited.
4 * SPDX-License-Identifier: MIT
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to
8 * deal in the Software without restriction, including without limitation the
9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10 * sell copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
13 * The above copyright notice and this permission notice shall be included in all
14 * copies or substantial portions of the Software.
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
24 #include "arm_compute/runtime/OMP/OMPScheduler.h"
26 #include "arm_compute/core/CPP/ICPPKernel.h"
27 #include "arm_compute/core/Error.h"
28 #include "arm_compute/core/Helpers.h"
29 #include "arm_compute/core/Utils.h"
30 #include "arm_compute/runtime/CPUUtils.h"
34 using namespace arm_compute;
36 OMPScheduler &OMPScheduler::get()
38 static OMPScheduler scheduler;
42 OMPScheduler::OMPScheduler() // NOLINT
43 : _num_threads(omp_get_max_threads())
45 get_cpu_configuration(_cpu_info);
48 unsigned int OMPScheduler::num_threads() const
53 void OMPScheduler::set_num_threads(unsigned int num_threads)
55 const unsigned int num_cores = omp_get_max_threads();
56 _num_threads = (num_threads == 0) ? num_cores : num_threads;
59 void OMPScheduler::schedule(ICPPKernel *kernel, unsigned int split_dimension)
61 ARM_COMPUTE_ERROR_ON_MSG(!kernel, "The child class didn't set the kernel");
64 info.cpu_info = &_cpu_info;
66 const Window &max_window = kernel->window();
67 const unsigned int num_iterations = max_window.num_iterations(split_dimension);
68 info.num_threads = std::min(num_iterations, _num_threads);
70 if(!kernel->is_parallelisable() || info.num_threads == 1)
72 kernel->run(max_window, info);
76 #pragma omp parallel firstprivate(info) num_threads(info.num_threads)
78 const int tid = omp_get_thread_num();
79 Window win = max_window.split_window(split_dimension, tid, info.num_threads);
81 kernel->run(win, info);