A Comprehensive Study on the Energy Efficiency of Java Thread-Save Collections

Transcript

1. A Comprehensive Study on the Energy Efficiency of Java Thread-Safe Collections Fernando Castor Kenan Liu Gustavo Pinto David Liu 1
2. Disclaimer 2
3. 3 Miguel turns 8mo today!
4. Motivations 4
5. First, energy consumption is not only a concern for unwired devices, but also for data centers 5
6. First, energy consumption is not only a concern for unwired devices, but also for data centers 6 Second, multicore CPUs are ubiquitous
7. First, energy consumption is not only a concern for unwired devices, but also for data centers 7 Second, multicore CPUs are ubiquitous Third, data structures are the fundamentals of computer programming
8. In case you are a Java programmer… 8
9. 9 List<Object> lists = …; • ArrayList • LinkedList
10. 10 List<Object> lists = new ArrayList<>();
11. 11 List<Object> lists = new ArrayList<>(); Thread
12. 12 • ArrayList • LinkedList • Vector • Collections.synchronizedList() • CopyOnWriteArrayList List<Object> lists = …; Non Thread-Safe Thread-Safe
13. 13 List<Object> lists = new Vector<>();
14. 14 List<Object> lists = new Vector<>(); Thread
15. 15 Thread Thread-safe! List<Object> lists = new Vector<>();
16. 16 Thread { } … List<Object> lists = new Vector<>();
17. 17 Thread { } … List<Object> lists = new Vector<>();
18. 18 List<Object> lists = new CopyOnWriteArrayList<>();
19. 19 Thread Thread-safe! List<Object> lists = new CopyOnWriteArrayList<>();
20. List<Object> lists = new CopyOnWriteArrayList<>(); 20 { } Thread Thread-safe!
21. 21 List<Object> lists = …; • ArrayList • LinkedList • Vector • Collections.synchronizedList() • CopyOnWriteArrayList Non Thread-Safe Thread-Safe
22. 22 List<Object> lists = …; Set<Objects> sets = …; Map<Object, Object> maps = …;
23. 23 List<Object> lists = …; Set<Objects> sets = …; Map<Object, Object> maps = …;
24. 24 List<Object> lists = …; Set<Objects> sets = …; Map<Object, Object> maps = …; And how about energy consumption?
25. 16 Collections 25 List ArrayList Vector Collections.syncList() CopyOnWriteArrayList Set LinkedHashSet Collections.syncSet() CopyOnWriteArraySet ConcurrentSkipListSet ConcurrentHashSet ConcurrentHashSetV8 Map LinkedHashMap Hashtable Collections.syncMap() ConcurrentSkipListMap ConcurrentHashMap ConcurrentHashMapV8
26. 16 Collections 26 List ArrayList Vector Collections.syncList() CopyOnWriteArrayList Set LinkedHashSet Collections.syncSet() CopyOnWriteArraySet ConcurrentSkipListSet ConcurrentHashSet ConcurrentHashSetV8 Map LinkedHashMap Hashtable Collections.syncMap() ConcurrentSkipListMap ConcurrentHashMap ConcurrentHashMapV8 Non thread-safe Thread-safe
27. 16 Collections 27 List ArrayList Vector Collections.syncList() CopyOnWriteArrayList Set LinkedHashSet Collections.syncSet() CopyOnWriteArraySet ConcurrentSkipListSet ConcurrentHashSet ConcurrentHashSetV8 Map LinkedHashMap Hashtable Collections.syncMap() ConcurrentSkipListMap ConcurrentHashMap ConcurrentHashMapV8 Java 7 Java 8
28. 16 Collections 28 List ArrayList Vector Collections.syncList() CopyOnWriteArrayList Set LinkedHashSet Collections.syncSet() CopyOnWriteArraySet ConcurrentSkipListSet ConcurrentHashSet ConcurrentHashSetV8 Map LinkedHashMap Hashtable Collections.syncMap() ConcurrentSkipListMap ConcurrentHashMap ConcurrentHashMapV8 x 3 Operations Traversal Insertion Removal
29. 2 Environments AMD CPU: A 2×16-core, running Debian, 2.4 GHz, 64GB of memory, JDK version 1.7 .0 11, build 21. 29 Intel CPU: A 2×8-core (32-cores w/ hyper-threading), running Debian, 2.60GHz, with 64GB of memory, JDK version 1.7 .0 71, build 14.
30. 2 Environments AMD CPU: A 2×16-core, running Debian, 2.4 GHz, 64GB of memory, JDK version 1.7 .0 11, build 21. 30 Intel CPU: A 2×8-core (32-cores w/ hyper-threading), running Debian, 2.60GHz, with 64GB of memory, JDK version 1.7 .0 71, build 14. Hardware-based energy measurement Software-based energy measurement
31. 31 More details about how the experiments were conducted? http:/ /icsme2016.github.io/program/accepted.html
32. Results 32
33. 33 Traversal Insertion Removal Intel CPU AMD CPU Traversal Insertion Removal Lists
34. 34 ArrayList Vector Lists Collections.synchronizedList()
35. 35 Lists ArrayList Vector Collections.synchronizedList() Energy (Joules)
36. 36 Lists ArrayList Vector Energy (Joules) Power (Watts) Collections.synchronizedList()
37. 37 Lists ArrayList Vector Energy (Joules) Power (Watts) Collections.synchronizedList()
38. Traversal Insertion Removal Intel CPU AMD CPU 38 38 Traversal Insertion Removal Lists
39. 39 Traversal Insertion Removal Intel CPU AMD CPU 39 Traversal Insertion Removal Lists
40. 40 Maps Intel CPU Traversal Insertion Removal AMD CPU Traversal Insertion Removal
41. 41 Maps Intel CPU Traversal Insertion Removal AMD CPU Traversal Insertion Removal
42. 42 Maps Intel CPU Traversal Insertion Removal AMD CPU Traversal Insertion Removal
43. 43 Maps Intel CPU Traversal Insertion Removal AMD CPU Traversal Insertion Removal
44. 44 Maps Intel CPU Traversal Insertion Removal AMD CPU Traversal Insertion Removal
45. 45 Maps Intel CPU Traversal Insertion Removal AMD CPU Traversal Insertion Removal
46. 46 Maps Intel CPU Traversal Insertion Removal AMD CPU Traversal Insertion Removal Less energy than the non thread-safe implementation!
47. Case Study 47
48. 48 Tomcat > A web server > More than 170K lines of Java code > More than 300 Hashtables
49. 49 Tomcat Xalan > Parses XML in HTML documents > More than 188K lines of Java code > More than 140 Hashtables > A web server > More than 170K lines of Java code > More than 300 Hashtables
50. 50 For each Hashtable instance, change it for a ConcurrentHashMap one. Do it again for ConcurrentHashMapV8 Task:
51. 51 Tomcat Hashtable to CHM: -12.21% Hashtable to CHM8: -17 .82%
52. 52 Tomcat Xalan Hashtable to CHM: -12.21% Hashtable to CHM8: -17 .82% Hashtable to CHM: -5.82% Hashtable to CHM8: -9.32%
53. 53 So, lets change all Hashtable instances to ConcurrentHashMap! Not so fast, young padawan…
54. 54 Hashtable ConcurrentHashMap Map 
 implements
55. 55 Hashtable ConcurrentHashMap Cloneable 
 Map 
 implements
56. 56 Hashtable ConcurrentHashMap List<Object> obj = new Hashtable<>(); obj.clone(); Cloneable 
 Map 
 / / works fine implements
57. 57 Hashtable ConcurrentHashMap List<Object> obj = new Hashtable<>(); obj.clone(); Cloneable 
 Map 
 / / works fine / / compiler error implements List<Object> obj = new ConcurrentHashMap<>(); obj.clone();
58. 58 Hashtable ConcurrentHashMap List<Object> obj = new Hashtable<>(); obj.clone(); Cloneable 
 Map 
 / / works fine / / compiler error implements List<Object> obj = new ConcurrentHashMap<>(); obj.clone(); Danny Dig Opportunity for improving refactoring tools!
59. Take Away Messages 59
60. 60 Small changes can have GREAT impacts, when it comes to energy consumption
61. 61 Doug Lea Stay up-to-date with the new releases of the java.util.concurrent library
62. In Summary 62
64. 64 THANKS
65. A Comprehensive Study on the Energy Efficiency of Java Thread-Safe Collections Fernando Castor Kenan Liu Gustavo Pinto David Liu 65
66. Experimental Environments 66 AMD CPU: A 2×16-core, running Debian, 2.4 GHz, 64GB of memory, JDK version 1.7 .0 11, build 21. Intel CPU: A 2×8-core (32-cores w/ hyper-threading), running Debian, 2.60GHz, with 64GB of memory, JDK version 1.7 .0 71, build 14.
67. 67 Maps (scaling number of threads) Intel CPU Traversal Insertion