1 /* 2 * Copyright (c) 2017-2021 Arm Limited. 3 * 4 * SPDX-License-Identifier: MIT 5 * 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: 12 * 13 * The above copyright notice and this permission notice shall be included in all 14 * copies or substantial portions of the Software. 15 * 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 22 * SOFTWARE. 23 */ 24 #ifndef ARM_COMPUTE_OMPSCHEDULER_H 25 #define ARM_COMPUTE_OMPSCHEDULER_H 26 27 #include "arm_compute/runtime/IScheduler.h" 28 29 namespace arm_compute 30 { 31 /** Pool of threads to automatically split a kernel's execution among several threads. */ 32 class OMPScheduler final : public IScheduler 33 { 34 public: 35 /** Constructor. */ 36 OMPScheduler(); 37 /** Sets the number of threads the scheduler will use to run the kernels. 38 * 39 * @param[in] num_threads If set to 0, then the number returned by omp_get_max_threads() will be used, otherwise the number of threads specified. 40 */ 41 void set_num_threads(unsigned int num_threads) override; 42 /** Returns the number of threads that the OMPScheduler has in its pool. 43 * 44 * @return Number of threads available in OMPScheduler. 45 */ 46 unsigned int num_threads() const override; 47 /** Multithread the execution of the passed kernel if possible. 48 * 49 * The kernel will run on a single thread if any of these conditions is true: 50 * - ICPPKernel::is_parallelisable() returns false 51 * - The scheduler has been initialized with only one thread. 52 * 53 * @param[in] kernel Kernel to execute. 54 * @param[in] hints Hints for the scheduler. 55 */ 56 void schedule(ICPPKernel *kernel, const Hints &hints) override; 57 58 /** Multithread the execution of the passed kernel if possible. 59 * 60 * The kernel will run on a single thread if any of these conditions is true: 61 * - ICPPKernel::is_parallelisable() returns false 62 * - The scheduler has been initialized with only one thread. 63 * 64 * @param[in] kernel Kernel to execute. 65 * @param[in] hints Hints for the scheduler. 66 * @param[in] window Window to use for kernel execution. 67 * @param[in] tensors Vector containing the tensors to operate on. 68 */ 69 void schedule_op(ICPPKernel *kernel, const Hints &hints, const Window &window, ITensorPack &tensors) override; 70 71 protected: 72 /** Execute all the passed workloads 73 * 74 * @note there is no guarantee regarding the order in which the workloads will be executed or whether or not they will be executed in parallel. 75 * 76 * @param[in] workloads Array of workloads to run 77 */ 78 void run_workloads(std::vector<Workload> &workloads) override; 79 80 private: 81 unsigned int _num_threads; 82 }; 83 } // namespace arm_compute 84 #endif /* ARM_COMPUTE_OMPSCHEDULER_H */ 85