• file
• revisions
• annotate
• diff
• comparison
• raw

src/share/vm/gc_implementation/shared/gcUtil.cpp@e018e6884bd8 (annotated)

src/share/vm/gc_implementation/shared/gcUtil.cpp

Wed, 23 Dec 2009 09:23:54 -0800

author

ysr

date

Wed, 23 Dec 2009 09:23:54 -0800

changeset 1580

e018e6884bd8

parent 435

a61af66fc99e

child 1907

c18cbe5936b8

permissions

-rw-r--r--

6631166: CMS: better heuristics when combatting fragmentation
Summary: Autonomic per-worker free block cache sizing, tunable coalition policies, fixes to per-size block statistics, retuned gain and bandwidth of some feedback loop filters to allow quicker reactivity to abrupt changes in ambient demand, and other heuristics to reduce fragmentation of the CMS old gen. Also tightened some assertions, including those related to locking.
Reviewed-by: jmasa

duke@435	1	/*
duke@435	2	* Copyright 2002-2005 Sun Microsystems, Inc. All Rights Reserved.
duke@435	3	* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
duke@435	4	*
duke@435	5	* This code is free software; you can redistribute it and/or modify it
duke@435	6	* under the terms of the GNU General Public License version 2 only, as
duke@435	7	* published by the Free Software Foundation.
duke@435	8	*
duke@435	9	* This code is distributed in the hope that it will be useful, but WITHOUT
duke@435	10	* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
duke@435	11	* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
duke@435	12	* version 2 for more details (a copy is included in the LICENSE file that
duke@435	13	* accompanied this code).
duke@435	14	*
duke@435	15	* You should have received a copy of the GNU General Public License version
duke@435	16	* 2 along with this work; if not, write to the Free Software Foundation,
duke@435	17	* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
duke@435	18	*
duke@435	19	* Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
duke@435	20	* CA 95054 USA or visit www.sun.com if you need additional information or
duke@435	21	* have any questions.
duke@435	22	*
duke@435	23	*/
duke@435	24
duke@435	25	# include "incls/_precompiled.incl"
duke@435	26	# include "incls/_gcUtil.cpp.incl"
duke@435	27
duke@435	28	// Catch-all file for utility classes
duke@435	29
duke@435	30	float AdaptiveWeightedAverage::compute_adaptive_average(float new_sample,
duke@435	31	float average) {
duke@435	32	// We smooth the samples by not using weight() directly until we've
duke@435	33	// had enough data to make it meaningful. We'd like the first weight
duke@435	34	// used to be 1, the second to be 1/2, etc until we have 100/weight
duke@435	35	// samples.
duke@435	36	unsigned count_weight = 100/count();
duke@435	37	unsigned adaptive_weight = (MAX2(weight(), count_weight));
duke@435	38
duke@435	39	float new_avg = exp_avg(average, new_sample, adaptive_weight);
duke@435	40
duke@435	41	return new_avg;
duke@435	42	}
duke@435	43
duke@435	44	void AdaptiveWeightedAverage::sample(float new_sample) {
duke@435	45	increment_count();
duke@435	46	assert(count() != 0,
duke@435	47	"Wraparound -- history would be incorrectly discarded");
duke@435	48
duke@435	49	// Compute the new weighted average
duke@435	50	float new_avg = compute_adaptive_average(new_sample, average());
duke@435	51	set_average(new_avg);
duke@435	52	_last_sample = new_sample;
duke@435	53	}
duke@435	54
ysr@1580	55	void AdaptiveWeightedAverage::print() const {
ysr@1580	56	print_on(tty);
ysr@1580	57	}
ysr@1580	58
ysr@1580	59	void AdaptiveWeightedAverage::print_on(outputStream* st) const {
ysr@1580	60	guarantee(false, "NYI");
ysr@1580	61	}
ysr@1580	62
ysr@1580	63	void AdaptivePaddedAverage::print() const {
ysr@1580	64	print_on(tty);
ysr@1580	65	}
ysr@1580	66
ysr@1580	67	void AdaptivePaddedAverage::print_on(outputStream* st) const {
ysr@1580	68	guarantee(false, "NYI");
ysr@1580	69	}
ysr@1580	70
ysr@1580	71	void AdaptivePaddedNoZeroDevAverage::print() const {
ysr@1580	72	print_on(tty);
ysr@1580	73	}
ysr@1580	74
ysr@1580	75	void AdaptivePaddedNoZeroDevAverage::print_on(outputStream* st) const {
ysr@1580	76	guarantee(false, "NYI");
ysr@1580	77	}
ysr@1580	78
duke@435	79	void AdaptivePaddedAverage::sample(float new_sample) {
ysr@1580	80	// Compute new adaptive weighted average based on new sample.
duke@435	81	AdaptiveWeightedAverage::sample(new_sample);
duke@435	82
ysr@1580	83	// Now update the deviation and the padded average.
duke@435	84	float new_avg = average();
duke@435	85	float new_dev = compute_adaptive_average(fabsd(new_sample - new_avg),
duke@435	86	deviation());
duke@435	87	set_deviation(new_dev);
duke@435	88	set_padded_average(new_avg + padding() * new_dev);
duke@435	89	_last_sample = new_sample;
duke@435	90	}
duke@435	91
duke@435	92	void AdaptivePaddedNoZeroDevAverage::sample(float new_sample) {
duke@435	93	// Compute our parent classes sample information
duke@435	94	AdaptiveWeightedAverage::sample(new_sample);
duke@435	95
duke@435	96	float new_avg = average();
duke@435	97	if (new_sample != 0) {
duke@435	98	// We only create a new deviation if the sample is non-zero
duke@435	99	float new_dev = compute_adaptive_average(fabsd(new_sample - new_avg),
duke@435	100	deviation());
duke@435	101
duke@435	102	set_deviation(new_dev);
duke@435	103	}
duke@435	104	set_padded_average(new_avg + padding() * deviation());
duke@435	105	_last_sample = new_sample;
duke@435	106	}
duke@435	107
duke@435	108	LinearLeastSquareFit::LinearLeastSquareFit(unsigned weight) :
duke@435	109	_sum_x(0), _sum_y(0), _sum_xy(0),
duke@435	110	_mean_x(weight), _mean_y(weight) {}
duke@435	111
duke@435	112	void LinearLeastSquareFit::update(double x, double y) {
duke@435	113	_sum_x = _sum_x + x;
duke@435	114	_sum_x_squared = _sum_x_squared + x * x;
duke@435	115	_sum_y = _sum_y + y;
duke@435	116	_sum_xy = _sum_xy + x * y;
duke@435	117	_mean_x.sample(x);
duke@435	118	_mean_y.sample(y);
duke@435	119	assert(_mean_x.count() == _mean_y.count(), "Incorrect count");
duke@435	120	if ( _mean_x.count() > 1 ) {
duke@435	121	double slope_denominator;
duke@435	122	slope_denominator = (_mean_x.count() * _sum_x_squared - _sum_x * _sum_x);
duke@435	123	// Some tolerance should be injected here. A denominator that is
duke@435	124	// nearly 0 should be avoided.
duke@435	125
duke@435	126	if (slope_denominator != 0.0) {
duke@435	127	double slope_numerator;
duke@435	128	slope_numerator = (_mean_x.count() * _sum_xy - _sum_x * _sum_y);
duke@435	129	_slope = slope_numerator / slope_denominator;
duke@435	130
duke@435	131	// The _mean_y and _mean_x are decaying averages and can
duke@435	132	// be used to discount earlier data. If they are used,
duke@435	133	// first consider whether all the quantities should be
duke@435	134	// kept as decaying averages.
duke@435	135	// _intercept = _mean_y.average() - _slope * _mean_x.average();
duke@435	136	_intercept = (_sum_y - _slope * _sum_x) / ((double) _mean_x.count());
duke@435	137	}
duke@435	138	}
duke@435	139	}
duke@435	140
duke@435	141	double LinearLeastSquareFit::y(double x) {
duke@435	142	double new_y;
duke@435	143
duke@435	144	if ( _mean_x.count() > 1 ) {
duke@435	145	new_y = (_intercept + _slope * x);
duke@435	146	return new_y;
duke@435	147	} else {
duke@435	148	return _mean_y.average();
duke@435	149	}
duke@435	150	}
duke@435	151
duke@435	152	// Both decrement_will_decrease() and increment_will_decrease() return
duke@435	153	// true for a slope of 0. That is because a change is necessary before
duke@435	154	// a slope can be calculated and a 0 slope will, in general, indicate
duke@435	155	// that no calculation of the slope has yet been done. Returning true
duke@435	156	// for a slope equal to 0 reflects the intuitive expectation of the
duke@435	157	// dependence on the slope. Don't use the complement of these functions
duke@435	158	// since that untuitive expectation is not built into the complement.
duke@435	159	bool LinearLeastSquareFit::decrement_will_decrease() {
duke@435	160	return (_slope >= 0.00);
duke@435	161	}
duke@435	162
duke@435	163	bool LinearLeastSquareFit::increment_will_decrease() {
duke@435	164	return (_slope <= 0.00);
duke@435	165	}