eaca9f8846
Reviewed-by: sjohanss, sangheki
149 lines
4.7 KiB
C++
149 lines
4.7 KiB
C++
/*
|
|
* Copyright (c) 2016, 2019, Oracle and/or its affiliates. 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/g1/g1Predictions.hpp"
|
|
#include "unittest.hpp"
|
|
|
|
#include "utilities/ostream.hpp"
|
|
|
|
static const double epsilon = 1e-6;
|
|
|
|
// Some basic formula tests with confidence = 0.0
|
|
TEST_VM(G1Predictions, basic_predictions) {
|
|
G1Predictions predictor(0.0);
|
|
TruncatedSeq s;
|
|
|
|
double p0 = predictor.predict(&s);
|
|
ASSERT_LT(p0, epsilon) << "Initial prediction of empty sequence must be 0.0";
|
|
|
|
s.add(5.0);
|
|
double p1 = predictor.predict(&s);
|
|
ASSERT_NEAR(p1, 5.0, epsilon);
|
|
|
|
for (int i = 0; i < 40; i++) {
|
|
s.add(5.0);
|
|
}
|
|
double p2 = predictor.predict(&s);
|
|
ASSERT_NEAR(p2, 5.0, epsilon);
|
|
}
|
|
|
|
// The following tests checks that the initial predictions are based on
|
|
// the average of the sequence and not on the stddev (which is 0).
|
|
TEST_VM(G1Predictions, average_not_stdev_predictions) {
|
|
G1Predictions predictor(0.5);
|
|
TruncatedSeq s;
|
|
|
|
s.add(1.0);
|
|
double p1 = predictor.predict(&s);
|
|
ASSERT_GT(p1, s.davg()) << "First prediction must be greater than average";
|
|
|
|
s.add(1.0);
|
|
double p2 = predictor.predict(&s);
|
|
ASSERT_GT(p1, p2) << "First prediction must be greater than second";
|
|
|
|
s.add(1.0);
|
|
double p3 = predictor.predict(&s);
|
|
ASSERT_GT(p2, p3) << "Second prediction must be greater than third";
|
|
|
|
s.add(1.0);
|
|
s.add(1.0); // Five elements are now in the sequence.
|
|
double p4 = predictor.predict(&s);
|
|
ASSERT_LT(p4, p3) << "Fourth prediction must be smaller than third";
|
|
ASSERT_NEAR(p4, 1.0, epsilon);
|
|
}
|
|
|
|
// The following tests checks that initially prediction based on
|
|
// the average is used, that gets overridden by the stddev prediction at
|
|
// the end.
|
|
TEST_VM(G1Predictions, average_stdev_predictions) {
|
|
G1Predictions predictor(0.5);
|
|
TruncatedSeq s;
|
|
|
|
s.add(0.5);
|
|
double p1 = predictor.predict(&s);
|
|
ASSERT_GT(p1, s.davg()) << "First prediction must be greater than average";
|
|
|
|
s.add(0.2);
|
|
double p2 = predictor.predict(&s);
|
|
ASSERT_GT(p1, p2) << "First prediction must be greater than second";
|
|
|
|
s.add(0.5);
|
|
double p3 = predictor.predict(&s);
|
|
ASSERT_GT(p2, p3) << "Second prediction must be greater than third";
|
|
|
|
s.add(0.2);
|
|
s.add(2.0);
|
|
double p4 = predictor.predict(&s);
|
|
ASSERT_GT(p4, p3) << "Fourth prediction must be greater than third";
|
|
}
|
|
|
|
// Some tests to verify bounding between [0 .. 1]
|
|
TEST_VM(G1Predictions, unit_predictions) {
|
|
G1Predictions predictor(0.5);
|
|
TruncatedSeq s;
|
|
|
|
double p0 = predictor.predict_in_unit_interval(&s);
|
|
ASSERT_LT(p0, epsilon) << "Initial prediction of empty sequence must be 0.0";
|
|
|
|
s.add(100.0);
|
|
double p1 = predictor.predict_in_unit_interval(&s);
|
|
ASSERT_NEAR(p1, 1.0, epsilon);
|
|
|
|
// Feed the sequence additional positive values to test the high bound.
|
|
for (int i = 0; i < 3; i++) {
|
|
s.add(2.0);
|
|
}
|
|
ASSERT_NEAR(predictor.predict_in_unit_interval(&s), 1.0, epsilon);
|
|
|
|
// Feed the sequence additional large negative value to test the low bound.
|
|
for (int i = 0; i < 4; i++) {
|
|
s.add(-200.0);
|
|
}
|
|
ASSERT_NEAR(predictor.predict_in_unit_interval(&s), 0.0, epsilon);
|
|
}
|
|
|
|
// Some tests to verify bounding between [0 .. +inf]
|
|
TEST_VM(G1Predictions, lower_bound_zero_predictions) {
|
|
G1Predictions predictor(0.5);
|
|
TruncatedSeq s;
|
|
|
|
double p0 = predictor.predict_zero_bounded(&s);
|
|
ASSERT_LT(p0, epsilon) << "Initial prediction of empty sequence must be 0.0";
|
|
|
|
s.add(100.0);
|
|
// Feed the sequence additional positive values to see that the high bound is not
|
|
// bounded by e.g. 1.0
|
|
for (int i = 0; i < 3; i++) {
|
|
s.add(2.0);
|
|
}
|
|
ASSERT_GT(predictor.predict_zero_bounded(&s), 1.0);
|
|
|
|
// Feed the sequence additional large negative value to test the low bound.
|
|
for (int i = 0; i < 4; i++) {
|
|
s.add(-200.0);
|
|
}
|
|
ASSERT_NEAR(predictor.predict_zero_bounded(&s), 0.0, epsilon);
|
|
}
|