Skip to content

Commit

Permalink
8295732: Directly embed TruncatedSeqs in G1Analytics
Browse files Browse the repository at this point in the history 
Reviewed-by: iwalulya, sjohanss
  • Loading branch information
Thomas Schatzl committed Oct 25, 2022
1 parent 3e49f8f commit ef60608
Show file tree
Hide file tree
Showing 2 changed files with 45 additions and 45 deletions.
72 changes: 36 additions & 36 deletions src/hotspot/share/gc/g1/g1Analytics.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -69,14 +69,14 @@ static double non_young_other_cost_per_region_ms_defaults[] = {

G1Analytics::G1Analytics(const G1Predictions* predictor) :
_predictor(predictor),
_recent_gc_times_ms(new TruncatedSeq(NumPrevPausesForHeuristics)),
_concurrent_mark_remark_times_ms(new TruncatedSeq(NumPrevPausesForHeuristics)),
_concurrent_mark_cleanup_times_ms(new TruncatedSeq(NumPrevPausesForHeuristics)),
_alloc_rate_ms_seq(new TruncatedSeq(TruncatedSeqLength)),
_recent_gc_times_ms(NumPrevPausesForHeuristics),
_concurrent_mark_remark_times_ms(NumPrevPausesForHeuristics),
_concurrent_mark_cleanup_times_ms(NumPrevPausesForHeuristics),
_alloc_rate_ms_seq(TruncatedSeqLength),
_prev_collection_pause_end_ms(0.0),
_concurrent_refine_rate_ms_seq(new TruncatedSeq(TruncatedSeqLength)),
_dirtied_cards_rate_ms_seq(new TruncatedSeq(TruncatedSeqLength)),
_dirtied_cards_in_thread_buffers_seq(new TruncatedSeq(TruncatedSeqLength)),
_concurrent_refine_rate_ms_seq(TruncatedSeqLength),
_dirtied_cards_rate_ms_seq(TruncatedSeqLength),
_dirtied_cards_in_thread_buffers_seq(TruncatedSeqLength),
_card_scan_to_merge_ratio_seq(TruncatedSeqLength),
_cost_per_card_scan_ms_seq(TruncatedSeqLength),
_cost_per_card_merge_ms_seq(TruncatedSeqLength),
Expand All @@ -87,21 +87,21 @@ G1Analytics::G1Analytics(const G1Predictions* predictor) :
_constant_other_time_ms_seq(TruncatedSeqLength),
_young_other_cost_per_region_ms_seq(TruncatedSeqLength),
_non_young_other_cost_per_region_ms_seq(TruncatedSeqLength),
_cost_per_byte_ms_during_cm_seq(new TruncatedSeq(TruncatedSeqLength)),
_recent_prev_end_times_for_all_gcs_sec(new TruncatedSeq(NumPrevPausesForHeuristics)),
_cost_per_byte_ms_during_cm_seq(TruncatedSeqLength),
_recent_prev_end_times_for_all_gcs_sec(NumPrevPausesForHeuristics),
_long_term_pause_time_ratio(0.0),
_short_term_pause_time_ratio(0.0) {

// Seed sequences with initial values.
_recent_prev_end_times_for_all_gcs_sec->add(os::elapsedTime());
_recent_prev_end_times_for_all_gcs_sec.add(os::elapsedTime());
_prev_collection_pause_end_ms = os::elapsedTime() * 1000.0;

int index = MIN2(ParallelGCThreads - 1, 7u);

// Start with inverse of maximum STW cost.
_concurrent_refine_rate_ms_seq->add(1/cost_per_logged_card_ms_defaults[0]);
_concurrent_refine_rate_ms_seq.add(1/cost_per_logged_card_ms_defaults[0]);
// Some applications have very low rates for logging cards.
_dirtied_cards_rate_ms_seq->add(0.0);
_dirtied_cards_rate_ms_seq.add(0.0);

_card_scan_to_merge_ratio_seq.set_initial(young_card_scan_to_merge_ratio_defaults[index]);
_cost_per_card_scan_ms_seq.set_initial(young_only_cost_per_card_scan_ms_defaults[index]);
Expand All @@ -114,8 +114,8 @@ G1Analytics::G1Analytics(const G1Predictions* predictor) :
_non_young_other_cost_per_region_ms_seq.add(non_young_other_cost_per_region_ms_defaults[index]);

// start conservatively (around 50ms is about right)
_concurrent_mark_remark_times_ms->add(0.05);
_concurrent_mark_cleanup_times_ms->add(0.20);
_concurrent_mark_remark_times_ms.add(0.05);
_concurrent_mark_cleanup_times_ms.add(0.20);
}

bool G1Analytics::enough_samples_available(TruncatedSeq const* seq) {
Expand Down Expand Up @@ -147,20 +147,20 @@ double G1Analytics::predict_zero_bounded(G1PhaseDependentSeq const* seq, bool fo
}

int G1Analytics::num_alloc_rate_ms() const {
return _alloc_rate_ms_seq->num();
return _alloc_rate_ms_seq.num();
}

void G1Analytics::report_concurrent_mark_remark_times_ms(double ms) {
_concurrent_mark_remark_times_ms->add(ms);
_concurrent_mark_remark_times_ms.add(ms);
}

void G1Analytics::report_alloc_rate_ms(double alloc_rate) {
_alloc_rate_ms_seq->add(alloc_rate);
_alloc_rate_ms_seq.add(alloc_rate);
}

void G1Analytics::compute_pause_time_ratios(double end_time_sec, double pause_time_ms) {
double long_interval_ms = (end_time_sec - oldest_known_gc_end_time_sec()) * 1000.0;
double gc_pause_time_ms = _recent_gc_times_ms->sum() - _recent_gc_times_ms->oldest() + pause_time_ms;
double gc_pause_time_ms = _recent_gc_times_ms.sum() - _recent_gc_times_ms.oldest() + pause_time_ms;
_long_term_pause_time_ratio = gc_pause_time_ms / long_interval_ms;
_long_term_pause_time_ratio = clamp(_long_term_pause_time_ratio, 0.0, 1.0);

Expand All @@ -170,15 +170,15 @@ void G1Analytics::compute_pause_time_ratios(double end_time_sec, double pause_ti
}

void G1Analytics::report_concurrent_refine_rate_ms(double cards_per_ms) {
_concurrent_refine_rate_ms_seq->add(cards_per_ms);
_concurrent_refine_rate_ms_seq.add(cards_per_ms);
}

void G1Analytics::report_dirtied_cards_rate_ms(double cards_per_ms) {
_dirtied_cards_rate_ms_seq->add(cards_per_ms);
_dirtied_cards_rate_ms_seq.add(cards_per_ms);
}

void G1Analytics::report_dirtied_cards_in_thread_buffers(size_t cards) {
_dirtied_cards_in_thread_buffers_seq->add(double(cards));
_dirtied_cards_in_thread_buffers_seq.add(double(cards));
}

void G1Analytics::report_cost_per_card_scan_ms(double cost_per_card_ms, bool for_young_only_phase) {
Expand All @@ -199,7 +199,7 @@ void G1Analytics::report_rs_length_diff(double rs_length_diff, bool for_young_on

void G1Analytics::report_cost_per_byte_ms(double cost_per_byte_ms, bool mark_or_rebuild_in_progress) {
if (mark_or_rebuild_in_progress) {
_cost_per_byte_ms_during_cm_seq->add(cost_per_byte_ms);
_cost_per_byte_ms_during_cm_seq.add(cost_per_byte_ms);
} else {
_copy_cost_per_byte_ms_seq.add(cost_per_byte_ms);
}
Expand All @@ -226,23 +226,23 @@ void G1Analytics::report_rs_length(double rs_length, bool for_young_only_phase)
}

double G1Analytics::predict_alloc_rate_ms() const {
if (enough_samples_available(_alloc_rate_ms_seq)) {
return predict_zero_bounded(_alloc_rate_ms_seq);
if (enough_samples_available(&_alloc_rate_ms_seq)) {
return predict_zero_bounded(&_alloc_rate_ms_seq);
} else {
return 0.0;
}
}

double G1Analytics::predict_concurrent_refine_rate_ms() const {
return predict_zero_bounded(_concurrent_refine_rate_ms_seq);
return predict_zero_bounded(&_concurrent_refine_rate_ms_seq);
}

double G1Analytics::predict_dirtied_cards_rate_ms() const {
return predict_zero_bounded(_dirtied_cards_rate_ms_seq);
return predict_zero_bounded(&_dirtied_cards_rate_ms_seq);
}

size_t G1Analytics::predict_dirtied_cards_in_thread_buffers() const {
return predict_size(_dirtied_cards_in_thread_buffers_seq);
return predict_size(&_dirtied_cards_in_thread_buffers_seq);
}

size_t G1Analytics::predict_scan_card_num(size_t rs_length, bool for_young_only_phase) const {
Expand All @@ -258,10 +258,10 @@ double G1Analytics::predict_card_scan_time_ms(size_t card_num, bool for_young_on
}

double G1Analytics::predict_object_copy_time_ms_during_cm(size_t bytes_to_copy) const {
if (!enough_samples_available(_cost_per_byte_ms_during_cm_seq)) {
if (!enough_samples_available(&_cost_per_byte_ms_during_cm_seq)) {
return (1.1 * bytes_to_copy) * predict_zero_bounded(&_copy_cost_per_byte_ms_seq);
} else {
return bytes_to_copy * predict_zero_bounded(_cost_per_byte_ms_during_cm_seq);
return bytes_to_copy * predict_zero_bounded(&_cost_per_byte_ms_during_cm_seq);
}
}

Expand All @@ -286,11 +286,11 @@ double G1Analytics::predict_non_young_other_time_ms(size_t non_young_num) const
}

double G1Analytics::predict_remark_time_ms() const {
return predict_zero_bounded(_concurrent_mark_remark_times_ms);
return predict_zero_bounded(&_concurrent_mark_remark_times_ms);
}

double G1Analytics::predict_cleanup_time_ms() const {
return predict_zero_bounded(_concurrent_mark_cleanup_times_ms);
return predict_zero_bounded(&_concurrent_mark_cleanup_times_ms);
}

size_t G1Analytics::predict_rs_length(bool for_young_only_phase) const {
Expand All @@ -303,19 +303,19 @@ size_t G1Analytics::predict_pending_cards(bool for_young_only_phase) const {
}

double G1Analytics::oldest_known_gc_end_time_sec() const {
return _recent_prev_end_times_for_all_gcs_sec->oldest();
return _recent_prev_end_times_for_all_gcs_sec.oldest();
}

double G1Analytics::most_recent_gc_end_time_sec() const {
return _recent_prev_end_times_for_all_gcs_sec->last();
return _recent_prev_end_times_for_all_gcs_sec.last();
}

void G1Analytics::update_recent_gc_times(double end_time_sec,
double pause_time_ms) {
_recent_gc_times_ms->add(pause_time_ms);
_recent_prev_end_times_for_all_gcs_sec->add(end_time_sec);
_recent_gc_times_ms.add(pause_time_ms);
_recent_prev_end_times_for_all_gcs_sec.add(end_time_sec);
}

void G1Analytics::report_concurrent_mark_cleanup_times_ms(double ms) {
_concurrent_mark_cleanup_times_ms->add(ms);
_concurrent_mark_cleanup_times_ms.add(ms);
}
18 changes: 9 additions & 9 deletions src/hotspot/share/gc/g1/g1Analytics.hpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -38,17 +38,17 @@ class G1Analytics: public CHeapObj<mtGC> {
const G1Predictions* _predictor;

// These exclude marking times.
TruncatedSeq* _recent_gc_times_ms;
TruncatedSeq _recent_gc_times_ms;

TruncatedSeq* _concurrent_mark_remark_times_ms;
TruncatedSeq* _concurrent_mark_cleanup_times_ms;
TruncatedSeq _concurrent_mark_remark_times_ms;
TruncatedSeq _concurrent_mark_cleanup_times_ms;

TruncatedSeq* _alloc_rate_ms_seq;
TruncatedSeq _alloc_rate_ms_seq;
double _prev_collection_pause_end_ms;

TruncatedSeq* _concurrent_refine_rate_ms_seq;
TruncatedSeq* _dirtied_cards_rate_ms_seq;
TruncatedSeq* _dirtied_cards_in_thread_buffers_seq;
TruncatedSeq _concurrent_refine_rate_ms_seq;
TruncatedSeq _dirtied_cards_rate_ms_seq;
TruncatedSeq _dirtied_cards_in_thread_buffers_seq;
// The ratio between the number of scanned cards and actually merged cards, for
// young-only and mixed gcs.
G1PhaseDependentSeq _card_scan_to_merge_ratio_seq;
Expand All @@ -69,10 +69,10 @@ class G1Analytics: public CHeapObj<mtGC> {
TruncatedSeq _young_other_cost_per_region_ms_seq;
TruncatedSeq _non_young_other_cost_per_region_ms_seq;

TruncatedSeq* _cost_per_byte_ms_during_cm_seq;
TruncatedSeq _cost_per_byte_ms_during_cm_seq;

// Statistics kept per GC stoppage, pause or full.
TruncatedSeq* _recent_prev_end_times_for_all_gcs_sec;
TruncatedSeq _recent_prev_end_times_for_all_gcs_sec;

// Cached values for long and short term pause time ratios. See
// compute_pause_time_ratios() for how they are computed.
Expand Down

1 comment on commit ef60608

@openjdk-notifier
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Review

Issues

Please sign in to comment.