shenandoahAdaptiveHeuristics.cpp

/*
 * Copyright (c) 2018, 2019, Red Hat, Inc. All rights reserved.
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 *
 * This code is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 only, as
 * published by the Free Software Foundation.
 *
 * This code is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
 * version 2 for more details (a copy is included in the LICENSE file that
 * accompanied this code).
 *
 * You should have received a copy of the GNU General Public License version
 * 2 along with this work; if not, write to the Free Software Foundation,
 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
 *
 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
 * or visit www.oracle.com if you need additional information or have any
 * questions.
 *
 */

#include "precompiled.hpp"

#include "gc/shenandoah/heuristics/shenandoahAdaptiveHeuristics.hpp"
#include "gc/shenandoah/shenandoahCollectionSet.hpp"
#include "gc/shenandoah/shenandoahFreeSet.hpp"
#include "gc/shenandoah/shenandoahHeap.inline.hpp"
#include "gc/shenandoah/shenandoahHeapRegion.inline.hpp"
#include "logging/log.hpp"
#include "logging/logTag.hpp"
#include "utilities/quickSort.hpp"

// These constants are used to adjust the margin of error for the moving
// average of the allocation rate and cycle time. The units are standard
// deviations.
const double ShenandoahAdaptiveHeuristics::FULL_PENALTY_SD = 0.2;
const double ShenandoahAdaptiveHeuristics::DEGENERATE_PENALTY_SD = 0.1;

// These are used to decide if we want to make any adjustments at all
// at the end of a successful concurrent cycle.
const double ShenandoahAdaptiveHeuristics::LOWEST_EXPECTED_AVAILABLE_AT_END = -0.5;
const double ShenandoahAdaptiveHeuristics::HIGHEST_EXPECTED_AVAILABLE_AT_END = 0.5;

// These values are the confidence interval expressed as standard deviations.
// At the minimum confidence level, there is a 25% chance that the true value of
// the estimate (average cycle time or allocation rate) is not more than
// MINIMUM_CONFIDENCE standard deviations away from our estimate. Similarly, the
// MAXIMUM_CONFIDENCE interval here means there is a one in a thousand chance
// that the true value of our estimate is outside the interval. These are used
// as bounds on the adjustments applied at the outcome of a GC cycle.
const double ShenandoahAdaptiveHeuristics::MINIMUM_CONFIDENCE = 0.319; // 25%
const double ShenandoahAdaptiveHeuristics::MAXIMUM_CONFIDENCE = 3.291; // 99.9%

ShenandoahAdaptiveHeuristics::ShenandoahAdaptiveHeuristics() :
  ShenandoahHeuristics(),
  _margin_of_error_sd(ShenandoahAdaptiveInitialConfidence),
  _spike_threshold_sd(ShenandoahAdaptiveInitialSpikeThreshold),
  _last_trigger(OTHER) { }

ShenandoahAdaptiveHeuristics::~ShenandoahAdaptiveHeuristics() {}

void ShenandoahAdaptiveHeuristics::choose_collection_set_from_regiondata(ShenandoahCollectionSet* cset,
                                                                         RegionData* data, size_t size,
                                                                         size_t actual_free) {
  size_t garbage_threshold = ShenandoahHeapRegion::region_size_bytes() * ShenandoahGarbageThreshold / 100;

  // The logic for cset selection in adaptive is as follows:
  //
  //   1. We cannot get cset larger than available free space. Otherwise we guarantee OOME
  //      during evacuation, and thus guarantee full GC. In practice, we also want to let
  //      application to allocate something. This is why we limit CSet to some fraction of
  //      available space. In non-overloaded heap, max_cset would contain all plausible candidates
  //      over garbage threshold.
  //
  //   2. We should not get cset too low so that free threshold would not be met right
  //      after the cycle. Otherwise we get back-to-back cycles for no reason if heap is
  //      too fragmented. In non-overloaded non-fragmented heap min_garbage would be around zero.
  //
  // Therefore, we start by sorting the regions by garbage. Then we unconditionally add the best candidates
  // before we meet min_garbage. Then we add all candidates that fit with a garbage threshold before
  // we hit max_cset. When max_cset is hit, we terminate the cset selection. Note that in this scheme,
  // ShenandoahGarbageThreshold is the soft threshold which would be ignored until min_garbage is hit.

  size_t capacity    = ShenandoahHeap::heap()->soft_max_capacity();
  size_t max_cset    = (size_t)((1.0 * capacity / 100 * ShenandoahEvacReserve) / ShenandoahEvacWaste);
  size_t free_target = (capacity / 100 * ShenandoahMinFreeThreshold) + max_cset;
  size_t min_garbage = (free_target > actual_free ? (free_target - actual_free) : 0);

  log_info(gc, ergo)("Adaptive CSet Selection. Target Free: " SIZE_FORMAT "%s, Actual Free: "
                     SIZE_FORMAT "%s, Max CSet: " SIZE_FORMAT "%s, Min Garbage: " SIZE_FORMAT "%s",
                     byte_size_in_proper_unit(free_target), proper_unit_for_byte_size(free_target),
                     byte_size_in_proper_unit(actual_free), proper_unit_for_byte_size(actual_free),
                     byte_size_in_proper_unit(max_cset),    proper_unit_for_byte_size(max_cset),
                     byte_size_in_proper_unit(min_garbage), proper_unit_for_byte_size(min_garbage));

  // Better select garbage-first regions
  QuickSort::sort<RegionData>(data, (int)size, compare_by_garbage, false);

  size_t cur_cset = 0;
  size_t cur_garbage = 0;

  for (size_t idx = 0; idx < size; idx++) {
    ShenandoahHeapRegion* r = data[idx]._region;

    size_t new_cset    = cur_cset + r->get_live_data_bytes();
    size_t new_garbage = cur_garbage + r->garbage();

    if (new_cset > max_cset) {
      break;
    }

    if ((new_garbage < min_garbage) || (r->garbage() > garbage_threshold)) {
      cset->add_region(r);
      cur_cset = new_cset;
      cur_garbage = new_garbage;
    }
  }
}

void ShenandoahAdaptiveHeuristics::record_cycle_start() {
  ShenandoahHeuristics::record_cycle_start();
  _allocation_rate.allocation_counter_reset();
}

void ShenandoahAdaptiveHeuristics::record_success_concurrent() {
  ShenandoahHeuristics::record_success_concurrent();

  size_t available = ShenandoahHeap::heap()->free_set()->available();

  _available.add(available);
  double z_score = 0.0;
  if (_available.sd() > 0) {
    z_score = (available - _available.avg()) / _available.sd();
  }

  log_debug(gc, ergo)("Available: " SIZE_FORMAT " %sB, z-score=%.3f. Average available: %.1f %sB +/- %.1f %sB.",
                      byte_size_in_proper_unit(available), proper_unit_for_byte_size(available),
                      z_score,
                      byte_size_in_proper_unit(_available.avg()), proper_unit_for_byte_size(_available.avg()),
                      byte_size_in_proper_unit(_available.sd()), proper_unit_for_byte_size(_available.sd()));

  // In the case when a concurrent GC cycle completes successfully but with an
  // unusually small amount of available memory we will adjust our trigger
  // parameters so that they are more likely to initiate a new cycle.
  // Conversely, when a GC cycle results in an above average amount of available
  // memory, we will adjust the trigger parameters to be less likely to initiate
  // a GC cycle.
  //
  // The z-score we've computed is in no way statistically related to the
  // trigger parameters, but it has the nice property that worse z-scores for
  // available memory indicate making larger adjustments to the trigger
  // parameters. It also results in fewer adjustments as the application
  // stabilizes.
  //
  // In order to avoid making endless and likely unnecessary adjustments to the
  // trigger parameters, the change in available memory (with respect to the
  // average) at the end of a cycle must be beyond these threshold values.
  if (z_score < LOWEST_EXPECTED_AVAILABLE_AT_END ||
      z_score > HIGHEST_EXPECTED_AVAILABLE_AT_END) {
    // The sign is flipped because a negative z-score indicates that the
    // available memory at the end of the cycle is below average. Positive
    // adjustments make the triggers more sensitive (i.e., more likely to fire).
    // The z-score also gives us a measure of just how far below normal. This
    // property allows us to adjust the trigger parameters proportionally.
    //
    // The `100` here is used to attenuate the size of our adjustments. This
    // number was chosen empirically. It also means the adjustments at the end of
    // a concurrent cycle are an order of magnitude smaller than the adjustments
    // made for a degenerated or full GC cycle (which themselves were also
    // chosen empirically).
    adjust_last_trigger_parameters(z_score / -100);
  }
}

void ShenandoahAdaptiveHeuristics::record_success_degenerated() {
  ShenandoahHeuristics::record_success_degenerated();
  // Adjust both trigger's parameters in the case of a degenerated GC because
  // either of them should have triggered earlier to avoid this case.
  adjust_margin_of_error(DEGENERATE_PENALTY_SD);
  adjust_spike_threshold(DEGENERATE_PENALTY_SD);
}

void ShenandoahAdaptiveHeuristics::record_success_full() {
  ShenandoahHeuristics::record_success_full();
  // Adjust both trigger's parameters in the case of a full GC because
  // either of them should have triggered earlier to avoid this case.
  adjust_margin_of_error(FULL_PENALTY_SD);
  adjust_spike_threshold(FULL_PENALTY_SD);
}

static double saturate(double value, double min, double max) {
  return MAX2(MIN2(value, max), min);
}

bool ShenandoahAdaptiveHeuristics::should_start_gc() {
  ShenandoahHeap* heap = ShenandoahHeap::heap();
  size_t max_capacity = heap->max_capacity();
  size_t capacity = heap->soft_max_capacity();
  size_t available = heap->free_set()->available();
  size_t allocated = heap->bytes_allocated_since_gc_start();

  // Make sure the code below treats available without the soft tail.
  size_t soft_tail = max_capacity - capacity;
  available = (available > soft_tail) ? (available - soft_tail) : 0;

  // Track allocation rate even if we decide to start a cycle for other reasons.
  double rate = _allocation_rate.sample(allocated);
  _last_trigger = OTHER;

  size_t min_threshold = capacity / 100 * ShenandoahMinFreeThreshold;
  if (available < min_threshold) {
    log_info(gc)("Trigger: Free (" SIZE_FORMAT "%s) is below minimum threshold (" SIZE_FORMAT "%s)",
                 byte_size_in_proper_unit(available),     proper_unit_for_byte_size(available),
                 byte_size_in_proper_unit(min_threshold), proper_unit_for_byte_size(min_threshold));
    return true;
  }

  const size_t max_learn = ShenandoahLearningSteps;
  if (_gc_times_learned < max_learn) {
    size_t init_threshold = capacity / 100 * ShenandoahInitFreeThreshold;
    if (available < init_threshold) {
      log_info(gc)("Trigger: Learning " SIZE_FORMAT " of " SIZE_FORMAT ". Free (" SIZE_FORMAT "%s) is below initial threshold (" SIZE_FORMAT "%s)",
                   _gc_times_learned + 1, max_learn,
                   byte_size_in_proper_unit(available),      proper_unit_for_byte_size(available),
                   byte_size_in_proper_unit(init_threshold), proper_unit_for_byte_size(init_threshold));
      return true;
    }
  }

  // Check if allocation headroom is still okay. This also factors in:
  //   1. Some space to absorb allocation spikes
  //   2. Accumulated penalties from Degenerated and Full GC
  size_t allocation_headroom = available;

  size_t spike_headroom = capacity / 100 * ShenandoahAllocSpikeFactor;
  size_t penalties      = capacity / 100 * _gc_time_penalties;

  allocation_headroom -= MIN2(allocation_headroom, spike_headroom);
  allocation_headroom -= MIN2(allocation_headroom, penalties);

  double avg_cycle_time = _gc_time_history->davg() + (_margin_of_error_sd * _gc_time_history->dsd());
  double avg_alloc_rate = _allocation_rate.upper_bound(_margin_of_error_sd);
  if (avg_cycle_time > allocation_headroom / avg_alloc_rate) {
    log_info(gc)("Trigger: Average GC time (%.2f ms) is above the time for average allocation rate (%.0f %sB/s) to deplete free headroom (" SIZE_FORMAT "%s) (margin of error = %.2f)",
                 avg_cycle_time * 1000,
                 byte_size_in_proper_unit(avg_alloc_rate), proper_unit_for_byte_size(avg_alloc_rate),
                 byte_size_in_proper_unit(allocation_headroom), proper_unit_for_byte_size(allocation_headroom),
                 _margin_of_error_sd);

    log_info(gc, ergo)("Free headroom: " SIZE_FORMAT "%s (free) - " SIZE_FORMAT "%s (spike) - " SIZE_FORMAT "%s (penalties) = " SIZE_FORMAT "%s",
                       byte_size_in_proper_unit(available),           proper_unit_for_byte_size(available),
                       byte_size_in_proper_unit(spike_headroom),      proper_unit_for_byte_size(spike_headroom),
                       byte_size_in_proper_unit(penalties),           proper_unit_for_byte_size(penalties),
                       byte_size_in_proper_unit(allocation_headroom), proper_unit_for_byte_size(allocation_headroom));

    _last_trigger = RATE;
    return true;
  }

  bool is_spiking = _allocation_rate.is_spiking(rate, _spike_threshold_sd);
  if (is_spiking && avg_cycle_time > allocation_headroom / rate) {
    log_info(gc)("Trigger: Average GC time (%.2f ms) is above the time for instantaneous allocation rate (%.0f %sB/s) to deplete free headroom (" SIZE_FORMAT "%s) (spike threshold = %.2f)",
                 avg_cycle_time * 1000,
                 byte_size_in_proper_unit(rate), proper_unit_for_byte_size(rate),
                 byte_size_in_proper_unit(allocation_headroom), proper_unit_for_byte_size(allocation_headroom),
                 _spike_threshold_sd);
    _last_trigger = SPIKE;
    return true;
  }

  return ShenandoahHeuristics::should_start_gc();
}

void ShenandoahAdaptiveHeuristics::adjust_last_trigger_parameters(double amount) {
  switch (_last_trigger) {
    case RATE:
      adjust_margin_of_error(amount);
      break;
    case SPIKE:
      adjust_spike_threshold(amount);
      break;
    case OTHER:
      // nothing to adjust here.
      break;
    default:
      ShouldNotReachHere();
  }
}

void ShenandoahAdaptiveHeuristics::adjust_margin_of_error(double amount) {
  _margin_of_error_sd = saturate(_margin_of_error_sd + amount, MINIMUM_CONFIDENCE, MAXIMUM_CONFIDENCE);
  log_debug(gc, ergo)("Margin of error now %.2f", _margin_of_error_sd);
}

void ShenandoahAdaptiveHeuristics::adjust_spike_threshold(double amount) {
  _spike_threshold_sd = saturate(_spike_threshold_sd - amount, MINIMUM_CONFIDENCE, MAXIMUM_CONFIDENCE);
  log_debug(gc, ergo)("Spike threshold now: %.2f", _spike_threshold_sd);
}

ShenandoahAllocationRate::ShenandoahAllocationRate() :
  _last_sample_time(os::elapsedTime()),
  _last_sample_value(0),
  _interval_sec(1.0 / ShenandoahAdaptiveSampleFrequencyHz),
  _rate(int(ShenandoahAdaptiveSampleSizeSeconds * ShenandoahAdaptiveSampleFrequencyHz), ShenandoahAdaptiveDecayFactor),
  _rate_avg(int(ShenandoahAdaptiveSampleSizeSeconds * ShenandoahAdaptiveSampleFrequencyHz), ShenandoahAdaptiveDecayFactor) {
}

double ShenandoahAllocationRate::sample(size_t allocated) {
  double now = os::elapsedTime();
  double rate = 0.0;
  if (now - _last_sample_time > _interval_sec) {
    if (allocated >= _last_sample_value) {
      rate = instantaneous_rate(now, allocated);
      _rate.add(rate);
      _rate_avg.add(_rate.avg());
    }

    _last_sample_time = now;
    _last_sample_value = allocated;
  }
  return rate;
}

double ShenandoahAllocationRate::upper_bound(double sds) const {
  // Here we are using the standard deviation of the computed running
  // average, rather than the standard deviation of the samples that went
  // into the moving average. This is a much more stable value and is tied
  // to the actual statistic in use (moving average over samples of averages).
  return _rate.davg() + (sds * _rate_avg.dsd());
}

void ShenandoahAllocationRate::allocation_counter_reset() {
  _last_sample_time = os::elapsedTime();
  _last_sample_value = 0;
}

bool ShenandoahAllocationRate::is_spiking(double rate, double threshold) const {
  if (rate <= 0.0) {
    return false;
  }

  double sd = _rate.sd();
  if (sd > 0) {
    // There is a small chance that that rate has already been sampled, but it
    // seems not to matter in practice.
    double z_score = (rate - _rate.avg()) / sd;
    if (z_score > threshold) {
      return true;
    }
  }
  return false;
}

double ShenandoahAllocationRate::instantaneous_rate(size_t allocated) const {
  return instantaneous_rate(os::elapsedTime(), allocated);
}

double ShenandoahAllocationRate::instantaneous_rate(double time, size_t allocated) const {
  size_t last_value = _last_sample_value;
  double last_time = _last_sample_time;
  size_t allocation_delta = (allocated > last_value) ? (allocated - last_value) : 0;
  double time_delta_sec = time - last_time;
  return (time_delta_sec > 0)  ? (allocation_delta / time_delta_sec) : 0;
}