/* * Copyright (c) 2013 The WebRTC project authors. All Rights Reserved. * * Use of this source code is governed by a BSD-style license * that can be found in the LICENSE file in the root of the source * tree. An additional intellectual property rights grant can be found * in the file PATENTS. All contributing project authors may * be found in the AUTHORS file in the root of the source tree. */ #include "video/overuse_frame_detector.h" #include #include #include #include #include #include #include #include "api/video/video_frame.h" #include "common_video/include/frame_callback.h" #include "rtc_base/checks.h" #include "rtc_base/logging.h" #include "rtc_base/timeutils.h" #include "system_wrappers/include/field_trial.h" #if defined(WEBRTC_MAC) && !defined(WEBRTC_IOS) #include #endif // defined(WEBRTC_MAC) && !defined(WEBRTC_IOS) namespace webrtc { namespace { const int64_t kCheckForOveruseIntervalMs = 5000; const int64_t kTimeToFirstCheckForOveruseMs = 100; // Delay between consecutive rampups. (Used for quick recovery.) const int kQuickRampUpDelayMs = 10 * 1000; // Delay between rampup attempts. Initially uses standard, scales up to max. const int kStandardRampUpDelayMs = 40 * 1000; const int kMaxRampUpDelayMs = 240 * 1000; // Expontential back-off factor, to prevent annoying up-down behaviour. const double kRampUpBackoffFactor = 2.0; // Max number of overuses detected before always applying the rampup delay. const int kMaxOverusesBeforeApplyRampupDelay = 4; const auto kScaleReasonCpu = AdaptationObserverInterface::AdaptReason::kCpu; } // namespace CpuOveruseOptions::CpuOveruseOptions() : high_encode_usage_threshold_percent(85), frame_timeout_interval_ms(1500), min_process_count(3), high_threshold_consecutive_count(2), filter_time_ms(5000) { #if defined(WEBRTC_MAC) && !defined(WEBRTC_IOS) // This is proof-of-concept code for letting the physical core count affect // the interval into which we attempt to scale. For now, the code is Mac OS // specific, since that's the platform were we saw most problems. // TODO(torbjorng): Enhance SystemInfo to return this metric. mach_port_t mach_host = mach_host_self(); host_basic_info hbi = {}; mach_msg_type_number_t info_count = HOST_BASIC_INFO_COUNT; kern_return_t kr = host_info(mach_host, HOST_BASIC_INFO, reinterpret_cast(&hbi), &info_count); mach_port_deallocate(mach_task_self(), mach_host); int n_physical_cores; if (kr != KERN_SUCCESS) { // If we couldn't get # of physical CPUs, don't panic. Assume we have 1. n_physical_cores = 1; RTC_LOG(LS_ERROR) << "Failed to determine number of physical cores, assuming 1"; } else { n_physical_cores = hbi.physical_cpu; RTC_LOG(LS_INFO) << "Number of physical cores:" << n_physical_cores; } // Change init list default for few core systems. The assumption here is that // encoding, which we measure here, takes about 1/4 of the processing of a // two-way call. This is roughly true for x86 using both vp8 and vp9 without // hardware encoding. Since we don't affect the incoming stream here, we only // control about 1/2 of the total processing needs, but this is not taken into // account. if (n_physical_cores == 1) high_encode_usage_threshold_percent = 20; // Roughly 1/4 of 100%. else if (n_physical_cores == 2) high_encode_usage_threshold_percent = 40; // Roughly 1/4 of 200%. #endif // defined(WEBRTC_MAC) && !defined(WEBRTC_IOS) // Note that we make the interval 2x+epsilon wide, since libyuv scaling steps // are close to that (when squared). This wide interval makes sure that // scaling up or down does not jump all the way across the interval. low_encode_usage_threshold_percent = (high_encode_usage_threshold_percent - 1) / 2; } // Class for calculating the processing usage on the send-side (the average // processing time of a frame divided by the average time difference between // captured frames). class OveruseFrameDetector::SendProcessingUsage { public: explicit SendProcessingUsage(const CpuOveruseOptions& options) : options_(options) { Reset(); } virtual ~SendProcessingUsage() {} void Reset() { // Start in between the underuse and overuse threshold. load_estimate_ = (options_.low_encode_usage_threshold_percent + options_.high_encode_usage_threshold_percent) / 200.0; } void AddSample(double encode_time, double diff_time) { RTC_CHECK_GE(diff_time, 0.0); // Use the filter update // // load <-- x/d (1-exp (-d/T)) + exp (-d/T) load // // where we must take care for small d, using the proper limit // (1 - exp(-d/tau)) / d = 1/tau - d/2tau^2 + O(d^2) double tau = (1e-3 * options_.filter_time_ms); double e = diff_time / tau; double c; if (e < 0.0001) { c = (1 - e / 2) / tau; } else { c = -expm1(-e) / diff_time; } load_estimate_ = c * encode_time + exp(-e) * load_estimate_; } virtual int Value() { return static_cast(100.0 * load_estimate_ + 0.5); } private: const CpuOveruseOptions options_; double load_estimate_; }; // Class used for manual testing of overuse, enabled via field trial flag. class OveruseFrameDetector::OverdoseInjector : public OveruseFrameDetector::SendProcessingUsage { public: OverdoseInjector(const CpuOveruseOptions& options, int64_t normal_period_ms, int64_t overuse_period_ms, int64_t underuse_period_ms) : OveruseFrameDetector::SendProcessingUsage(options), normal_period_ms_(normal_period_ms), overuse_period_ms_(overuse_period_ms), underuse_period_ms_(underuse_period_ms), state_(State::kNormal), last_toggling_ms_(-1) { RTC_DCHECK_GT(overuse_period_ms, 0); RTC_DCHECK_GT(normal_period_ms, 0); RTC_LOG(LS_INFO) << "Simulating overuse with intervals " << normal_period_ms << "ms normal mode, " << overuse_period_ms << "ms overuse mode."; } ~OverdoseInjector() override {} int Value() override { int64_t now_ms = rtc::TimeMillis(); if (last_toggling_ms_ == -1) { last_toggling_ms_ = now_ms; } else { switch (state_) { case State::kNormal: if (now_ms > last_toggling_ms_ + normal_period_ms_) { state_ = State::kOveruse; last_toggling_ms_ = now_ms; RTC_LOG(LS_INFO) << "Simulating CPU overuse."; } break; case State::kOveruse: if (now_ms > last_toggling_ms_ + overuse_period_ms_) { state_ = State::kUnderuse; last_toggling_ms_ = now_ms; RTC_LOG(LS_INFO) << "Simulating CPU underuse."; } break; case State::kUnderuse: if (now_ms > last_toggling_ms_ + underuse_period_ms_) { state_ = State::kNormal; last_toggling_ms_ = now_ms; RTC_LOG(LS_INFO) << "Actual CPU overuse measurements in effect."; } break; } } rtc::Optional overried_usage_value; switch (state_) { case State::kNormal: break; case State::kOveruse: overried_usage_value.emplace(250); break; case State::kUnderuse: overried_usage_value.emplace(5); break; } return overried_usage_value.value_or(SendProcessingUsage::Value()); } private: const int64_t normal_period_ms_; const int64_t overuse_period_ms_; const int64_t underuse_period_ms_; enum class State { kNormal, kOveruse, kUnderuse } state_; int64_t last_toggling_ms_; }; std::unique_ptr OveruseFrameDetector::CreateSendProcessingUsage( const CpuOveruseOptions& options) { std::unique_ptr instance; std::string toggling_interval = field_trial::FindFullName("WebRTC-ForceSimulatedOveruseIntervalMs"); if (!toggling_interval.empty()) { int normal_period_ms = 0; int overuse_period_ms = 0; int underuse_period_ms = 0; if (sscanf(toggling_interval.c_str(), "%d-%d-%d", &normal_period_ms, &overuse_period_ms, &underuse_period_ms) == 3) { if (normal_period_ms > 0 && overuse_period_ms > 0 && underuse_period_ms > 0) { instance.reset(new OverdoseInjector( options, normal_period_ms, overuse_period_ms, underuse_period_ms)); } else { RTC_LOG(LS_WARNING) << "Invalid (non-positive) normal/overuse/underuse periods: " << normal_period_ms << " / " << overuse_period_ms << " / " << underuse_period_ms; } } else { RTC_LOG(LS_WARNING) << "Malformed toggling interval: " << toggling_interval; } } if (!instance) { // No valid overuse simulation parameters set, use normal usage class. instance.reset(new SendProcessingUsage(options)); } return instance; } class OveruseFrameDetector::CheckOveruseTask : public rtc::QueuedTask { public: explicit CheckOveruseTask(OveruseFrameDetector* overuse_detector) : overuse_detector_(overuse_detector) { rtc::TaskQueue::Current()->PostDelayedTask( std::unique_ptr(this), kTimeToFirstCheckForOveruseMs); } void Stop() { RTC_CHECK(task_checker_.CalledSequentially()); overuse_detector_ = nullptr; } private: bool Run() override { RTC_CHECK(task_checker_.CalledSequentially()); if (!overuse_detector_) return true; // This will make the task queue delete this task. overuse_detector_->CheckForOveruse(); rtc::TaskQueue::Current()->PostDelayedTask( std::unique_ptr(this), kCheckForOveruseIntervalMs); // Return false to prevent this task from being deleted. Ownership has been // transferred to the task queue when PostDelayedTask was called. return false; } rtc::SequencedTaskChecker task_checker_; OveruseFrameDetector* overuse_detector_; }; OveruseFrameDetector::OveruseFrameDetector( const CpuOveruseOptions& options, AdaptationObserverInterface* observer, EncodedFrameObserver* encoder_timing, CpuOveruseMetricsObserver* metrics_observer) : check_overuse_task_(nullptr), options_(options), observer_(observer), encoder_timing_(encoder_timing), metrics_observer_(metrics_observer), num_process_times_(0), // TODO(nisse): Use rtc::Optional last_capture_time_us_(-1), last_processed_capture_time_us_(-1), num_pixels_(0), last_overuse_time_ms_(-1), checks_above_threshold_(0), num_overuse_detections_(0), last_rampup_time_ms_(-1), in_quick_rampup_(false), current_rampup_delay_ms_(kStandardRampUpDelayMs), usage_(CreateSendProcessingUsage(options)) { task_checker_.Detach(); } OveruseFrameDetector::~OveruseFrameDetector() { RTC_DCHECK(!check_overuse_task_) << "StopCheckForOverUse must be called."; } void OveruseFrameDetector::StartCheckForOveruse() { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); RTC_DCHECK(!check_overuse_task_); check_overuse_task_ = new CheckOveruseTask(this); } void OveruseFrameDetector::StopCheckForOveruse() { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); check_overuse_task_->Stop(); check_overuse_task_ = nullptr; } void OveruseFrameDetector::EncodedFrameTimeMeasured(int encode_duration_ms) { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); if (!metrics_) metrics_ = rtc::Optional(CpuOveruseMetrics()); metrics_->encode_usage_percent = usage_->Value(); metrics_observer_->OnEncodedFrameTimeMeasured(encode_duration_ms, *metrics_); } bool OveruseFrameDetector::FrameSizeChanged(int num_pixels) const { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); if (num_pixels != num_pixels_) { return true; } return false; } bool OveruseFrameDetector::FrameTimeoutDetected(int64_t now_us) const { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); if (last_capture_time_us_ == -1) return false; return (now_us - last_capture_time_us_) > options_.frame_timeout_interval_ms * rtc::kNumMicrosecsPerMillisec; } void OveruseFrameDetector::ResetAll() { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); usage_->Reset(); last_capture_time_us_ = -1; last_processed_capture_time_us_ = -1; num_process_times_ = 0; metrics_ = rtc::Optional(); } void OveruseFrameDetector::FrameCaptured(int width, int height) { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); if (FrameSizeChanged(width * height)) { ResetAll(); num_pixels_ = width * height; } } void OveruseFrameDetector::FrameEncoded(int64_t capture_time_us, int64_t encode_duration_us) { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); if (FrameTimeoutDetected(capture_time_us)) { ResetAll(); } else if (last_capture_time_us_ != -1) { usage_->AddSample(1e-6 * encode_duration_us, 1e-6 * (capture_time_us - last_capture_time_us_)); } last_capture_time_us_ = capture_time_us; EncodedFrameTimeMeasured(encode_duration_us / rtc::kNumMicrosecsPerMillisec); if (encoder_timing_) { // TODO(nisse): Update encoder_timing_ to also use us units. encoder_timing_->OnEncodeTiming( capture_time_us / rtc::kNumMicrosecsPerMillisec, encode_duration_us / rtc::kNumMicrosecsPerMillisec); } } void OveruseFrameDetector::CheckForOveruse() { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); ++num_process_times_; if (num_process_times_ <= options_.min_process_count || !metrics_) return; int64_t now_ms = rtc::TimeMillis(); if (IsOverusing(*metrics_)) { // If the last thing we did was going up, and now have to back down, we need // to check if this peak was short. If so we should back off to avoid going // back and forth between this load, the system doesn't seem to handle it. bool check_for_backoff = last_rampup_time_ms_ > last_overuse_time_ms_; if (check_for_backoff) { if (now_ms - last_rampup_time_ms_ < kStandardRampUpDelayMs || num_overuse_detections_ > kMaxOverusesBeforeApplyRampupDelay) { // Going up was not ok for very long, back off. current_rampup_delay_ms_ *= kRampUpBackoffFactor; if (current_rampup_delay_ms_ > kMaxRampUpDelayMs) current_rampup_delay_ms_ = kMaxRampUpDelayMs; } else { // Not currently backing off, reset rampup delay. current_rampup_delay_ms_ = kStandardRampUpDelayMs; } } last_overuse_time_ms_ = now_ms; in_quick_rampup_ = false; checks_above_threshold_ = 0; ++num_overuse_detections_; if (observer_) observer_->AdaptDown(kScaleReasonCpu); } else if (IsUnderusing(*metrics_, now_ms)) { last_rampup_time_ms_ = now_ms; in_quick_rampup_ = true; if (observer_) observer_->AdaptUp(kScaleReasonCpu); } int rampup_delay = in_quick_rampup_ ? kQuickRampUpDelayMs : current_rampup_delay_ms_; RTC_LOG(LS_VERBOSE) << " Frame stats: " << " encode usage " << metrics_->encode_usage_percent << " overuse detections " << num_overuse_detections_ << " rampup delay " << rampup_delay; } bool OveruseFrameDetector::IsOverusing(const CpuOveruseMetrics& metrics) { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); if (metrics.encode_usage_percent >= options_.high_encode_usage_threshold_percent) { ++checks_above_threshold_; } else { checks_above_threshold_ = 0; } return checks_above_threshold_ >= options_.high_threshold_consecutive_count; } bool OveruseFrameDetector::IsUnderusing(const CpuOveruseMetrics& metrics, int64_t time_now) { RTC_DCHECK_CALLED_SEQUENTIALLY(&task_checker_); int delay = in_quick_rampup_ ? kQuickRampUpDelayMs : current_rampup_delay_ms_; if (time_now < last_rampup_time_ms_ + delay) return false; return metrics.encode_usage_percent < options_.low_encode_usage_threshold_percent; } } // namespace webrtc