1 /*
2  * Copyright (C) 2010 The Guava Authors
3  *
4  * Licensed under the Apache License, Version 2.0 (the "License");
5  * you may not use this file except in compliance with the License.
6  * You may obtain a copy of the License at
7  *
8  * http://www.apache.org/licenses/LICENSE-2.0
9  *
10  * Unless required by applicable law or agreed to in writing, software
11  * distributed under the License is distributed on an "AS IS" BASIS,
12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13  * See the License for the specific language governing permissions and
14  * limitations under the License.
15  */
16 
17 package com.google.common.cache;
18 
19 import com.google.caliper.AfterExperiment;
20 import com.google.caliper.BeforeExperiment;
21 import com.google.caliper.Benchmark;
22 import com.google.caliper.Param;
23 import com.google.common.primitives.Ints;
24 
25 import java.util.Random;
26 import java.util.concurrent.atomic.AtomicLong;
27 
28 /**
29  * Single-threaded benchmark for {@link LoadingCache}.
30  *
31  * @author Charles Fry
32  */
33 public class LoadingCacheSingleThreadBenchmark {
34   @Param({"1000", "2000"}) int maximumSize;
35   @Param("5000") int distinctKeys;
36   @Param("4") int segments;
37 
38   // 1 means uniform likelihood of keys; higher means some keys are more popular
39   // tweak this to control hit rate
40   @Param("2.5") double concentration;
41 
42   Random random = new Random();
43 
44   LoadingCache<Integer, Integer> cache;
45 
46   int max;
47 
48   static AtomicLong requests = new AtomicLong(0);
49   static AtomicLong misses = new AtomicLong(0);
50 
setUp()51   @BeforeExperiment void setUp() {
52     // random integers will be generated in this range, then raised to the
53     // power of (1/concentration) and floor()ed
54     max = Ints.checkedCast((long) Math.pow(distinctKeys, concentration));
55 
56     cache = CacheBuilder.newBuilder()
57         .concurrencyLevel(segments)
58         .maximumSize(maximumSize)
59         .build(
60             new CacheLoader<Integer, Integer>() {
61               @Override public Integer load(Integer from) {
62                 return (int) misses.incrementAndGet();
63               }
64             });
65 
66     // To start, fill up the cache.
67     // Each miss both increments the counter and causes the map to grow by one,
68     // so until evictions begin, the size of the map is the greatest return
69     // value seen so far
70     while (cache.getUnchecked(nextRandomKey()) < maximumSize) {}
71 
72     requests.set(0);
73     misses.set(0);
74   }
75 
time(int reps)76   @Benchmark int time(int reps) {
77     int dummy = 0;
78     for (int i = 0; i < reps; i++) {
79       dummy += cache.getUnchecked(nextRandomKey());
80     }
81     requests.addAndGet(reps);
82     return dummy;
83   }
84 
nextRandomKey()85   private int nextRandomKey() {
86     int a = random.nextInt(max);
87 
88     /*
89      * For example, if concentration=2.0, the following takes the square root of
90      * the uniformly-distributed random integer, then truncates any fractional
91      * part, so higher integers would appear (in this case linearly) more often
92      * than lower ones.
93      */
94     return (int) Math.pow(a, 1.0 / concentration);
95   }
96 
tearDown()97   @AfterExperiment void tearDown() {
98     double req = requests.get();
99     double hit = req - misses.get();
100 
101     // Currently, this is going into /dev/null, but I'll fix that
102     System.out.println("hit rate: " + hit / req);
103   }
104 
105   // for proper distributions later:
106   // import JSci.maths.statistics.ProbabilityDistribution;
107   // int key = (int) dist.inverse(random.nextDouble());
108 }
109