diff options
Diffstat (limited to 'src/runtime/IScheduler.cpp')
| -rw-r--r-- | src/runtime/IScheduler.cpp | 99 |
1 files changed, 72 insertions, 27 deletions
diff --git a/src/runtime/IScheduler.cpp b/src/runtime/IScheduler.cpp index df04fed401..ecf84abd2c 100644 --- a/src/runtime/IScheduler.cpp +++ b/src/runtime/IScheduler.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016-2021 Arm Limited. + * Copyright (c) 2016-2023 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -25,14 +25,15 @@ #include "arm_compute/core/CPP/ICPPKernel.h" #include "arm_compute/core/Error.h" +#include "arm_compute/core/Log.h" #include "arm_compute/core/Window.h" + #include "src/common/cpuinfo/CpuInfo.h" #include "src/runtime/SchedulerUtils.h" namespace arm_compute { IScheduler::IScheduler() - : _cpu_info() { // Work out the best possible number of execution threads _num_threads_hint = cpuinfo::num_threads_hint(); @@ -40,7 +41,7 @@ IScheduler::IScheduler() CPUInfo &IScheduler::cpu_info() { - return _cpu_info; + return CPUInfo::get(); } void IScheduler::set_num_threads_with_affinity(unsigned int num_threads, BindFunc func) @@ -59,7 +60,7 @@ void IScheduler::schedule_common(ICPPKernel *kernel, const Hints &hints, const W ARM_COMPUTE_ERROR_ON_MSG(!kernel, "The child class didn't set the kernel"); #ifndef BARE_METAL const Window &max_window = window; - if(hints.split_dimension() == IScheduler::split_dimensions_all) + if (hints.split_dimension() == IScheduler::split_dimensions_all) { /* * if the split dim is size_t max then this signals we should parallelise over @@ -73,27 +74,27 @@ void IScheduler::schedule_common(ICPPKernel *kernel, const Hints &hints, const W std::tie(m_threads, n_threads) = scheduler_utils::split_2d(this->num_threads(), m, n); std::vector<IScheduler::Workload> workloads; - for(unsigned int ni = 0; ni != n_threads; ++ni) + for (unsigned int ni = 0; ni != n_threads; ++ni) { - for(unsigned int mi = 0; mi != m_threads; ++mi) + for (unsigned int mi = 0; mi != m_threads; ++mi) { workloads.push_back( - [ni, mi, m_threads, n_threads, &max_window, &kernel](const ThreadInfo & info) - { - //narrow the window to our mi-ni workload - Window win = max_window.split_window(Window::DimX, mi, m_threads) - .split_window(Window::DimY, ni, n_threads); + [ni, mi, m_threads, n_threads, &max_window, &kernel](const ThreadInfo &info) + { + //narrow the window to our mi-ni workload + Window win = max_window.split_window(Window::DimX, mi, m_threads) + .split_window(Window::DimY, ni, n_threads); - win.validate(); + win.validate(); - Window thread_locator; - thread_locator.set(Window::DimX, Window::Dimension(mi, m_threads)); - thread_locator.set(Window::DimY, Window::Dimension(ni, n_threads)); + Window thread_locator; + thread_locator.set(Window::DimX, Window::Dimension(mi, m_threads)); + thread_locator.set(Window::DimY, Window::Dimension(ni, n_threads)); - thread_locator.validate(); + thread_locator.validate(); - kernel->run_nd(win, info, thread_locator); - }); + kernel->run_nd(win, info, thread_locator); + }); } } run_workloads(workloads); @@ -103,16 +104,16 @@ void IScheduler::schedule_common(ICPPKernel *kernel, const Hints &hints, const W const unsigned int num_iterations = max_window.num_iterations(hints.split_dimension()); const unsigned int num_threads = std::min(num_iterations, this->num_threads()); - if(num_iterations == 0) + if (num_iterations == 0) { return; } - if(!kernel->is_parallelisable() || num_threads == 1) + if (!kernel->is_parallelisable() || num_threads == 1) { ThreadInfo info; - info.cpu_info = &_cpu_info; - if(tensors.empty()) + info.cpu_info = &cpu_info(); + if (tensors.empty()) { kernel->run(max_window, info); } @@ -124,14 +125,15 @@ void IScheduler::schedule_common(ICPPKernel *kernel, const Hints &hints, const W else { unsigned int num_windows = 0; - switch(hints.strategy()) + switch (hints.strategy()) { case StrategyHint::STATIC: num_windows = num_threads; break; case StrategyHint::DYNAMIC: { - const unsigned int granule_threshold = (hints.threshold() <= 0) ? num_threads : static_cast<unsigned int>(hints.threshold()); + const unsigned int granule_threshold = + (hints.threshold() <= 0) ? num_threads : static_cast<unsigned int>(hints.threshold()); // Make sure we don't use some windows which are too small as this might create some contention on the ThreadFeeder num_windows = num_iterations > granule_threshold ? granule_threshold : num_iterations; break; @@ -139,16 +141,19 @@ void IScheduler::schedule_common(ICPPKernel *kernel, const Hints &hints, const W default: ARM_COMPUTE_ERROR("Unknown strategy"); } + // Make sure the smallest window is larger than minimum workload size + num_windows = adjust_num_of_windows(max_window, hints.split_dimension(), num_windows, *kernel, cpu_info()); + std::vector<IScheduler::Workload> workloads(num_windows); - for(unsigned int t = 0; t < num_windows; ++t) + for (unsigned int t = 0; t < num_windows; ++t) { //Capture 't' by copy, all the other variables by reference: - workloads[t] = [t, &hints, &max_window, &num_windows, &kernel, &tensors](const ThreadInfo & info) + workloads[t] = [t, &hints, &max_window, &num_windows, &kernel, &tensors](const ThreadInfo &info) { Window win = max_window.split_window(hints.split_dimension(), t, num_windows); win.validate(); - if(tensors.empty()) + if (tensors.empty()) { kernel->run(win, info); } @@ -172,4 +177,44 @@ void IScheduler::run_tagged_workloads(std::vector<Workload> &workloads, const ch run_workloads(workloads); } +std::size_t IScheduler::adjust_num_of_windows(const Window &window, + std::size_t split_dimension, + std::size_t init_num_windows, + const ICPPKernel &kernel, + const CPUInfo &cpu_info) +{ + // Mitigation of the narrow split issue, which occurs when the split dimension is too small to split (hence "narrow"). + if (window.num_iterations(split_dimension) < init_num_windows) + { + auto recommended_split_dim = Window::DimX; + for (std::size_t dims = Window::DimY; dims <= Window::DimW; ++dims) + { + if (window.num_iterations(recommended_split_dim) < window.num_iterations(dims)) + { + recommended_split_dim = dims; + } + } + ARM_COMPUTE_LOG_INFO_MSG_WITH_FORMAT_CORE( + "%zu dimension is not a suitable dimension to split the workload. Recommended: %zu recommended_split_dim", + split_dimension, recommended_split_dim); + } + + for (auto t = init_num_windows; t > 0; --t) // Trying the highest number of windows ,init_num_windows, first + { + // Try splitting the workload into t, subject to each subworkload size <= mws. + if ((window.num_iterations(split_dimension) / kernel.get_mws(cpu_info, t)) >= t) + { + if (t != init_num_windows) + { + ARM_COMPUTE_LOG_INFO_MSG_CORE( + "The scheduler is using a different thread count than the one assigned by the user."); + } + return t; + } + } + ARM_COMPUTE_LOG_INFO_MSG_CORE( + "The scheduler is using single thread instead of the thread count assigned by the user."); + return 1; // If the workload is so small that it can't be split, we should run a single thread +} + } // namespace arm_compute |